Improvement strategies for yield potential, disease resistance and drought

tolerance of Zimbabwean maize inbred lines

By

Thokozile Ndhlela

Submitted in accordance with the academic requirements for the degree

Philosophiae Doctor

Department of Plant Sciences (Plant Breeding)

Faculty of Natural and Agricultural Sciences

University of the Free State, South Africa

Promoter: Prof. M.T. Labuschagne

Co-promoters: Prof. L. Herselman
Dr. C. Magorokosho

June 2012
 Declaration
I declare that the thesis hereby submitted by me for the degree Philosophiae Doctor in
Agriculture at the University of the Free State is my own independent work and has not
previously been submitted by me at another university/faculty. I further cede copyright of the
thesis in favour of the University of the Free State.

…………………... …………………………
Thokozile Ndhlela Date

i
 Dedication
To my late husband (Solomon), my sons (Eugene, Ginola, Winstone and Munashe), my
father (Moses) and mother (Lillian).

ii
 Acknowledgements
I would like to start by recognising the great role played by CIMMYT in making this dream
come true through the financial support rendered and the Department of Research and
Specialist Services, my employer, for affording me the opportunity to further my studies.
Secondly my special thanks go to Dr Marianne Banziger for inspiring me to do the study.
Thirdly my immense thanks go to my supevisors Prof M. Labuschagne, Prof L. Herselman
and Dr C. Magorokosho for without their guidance execution of the whole study was not
going to be feasible.

At CIMMYT I would like to thank Sebastian Mawere, Martin Shoko, Stanely Gokoma,
Joseph Makamba, Emma Maramba, Simbarashe Chisoro and Semai for the various
contributions they made during the study. I would also like to thank my colleagues at the
Crop Breeding Institute namely Charles Mutimaamba, Ronica Mukaro, Purity Mazibuko,
Rebecca Pfachi and team leaders at various Research Stations for their assistance during
evaluation of the trials. Special thanks also go to my fellow students Xavier Mhike,
Dimakatso Masindeni, Martin Chamunorwa, Nyika Rwatirera, Tariro Kusada, Linda Phiri,
Terrance Tapera, Elliot Tembo and Fortunus Kapinga for their input and support. Of special
mention is Dr P. Setimela for the assistance rendered in the GGE biplot analysis and Dr S.
Kassa for the assistance rendered in the analysis of molecular data. At the University of Free
State special mention goes to the administrator Sadie Geldenhuys who went out of her way to
make sure I were comfortable and for being motherly to me.

Last but not least of special mention are my beloved sons Eugene, Ginola, Winstone and
Munashe, my mother Lillian, my father Moses and my siblings Themba, Sithabiso,
Thandinkosi and Dingindhlela for their continued support and enduring my absence during
the execution of the study.

iii
 Contents
Page
Declaration i
Dedication ii
Acknowledgements iii
Contents iv
List of tables xii
List of figures xix
Abbreviations and symbols xxiii
CHAPTER 1 1
General introduction 1
1.1 Importance of maize and production constraints in Africa 1
1.2 Maize production in Zimbabwe 1
1.3 Maize production constraints in Zimbabwe 6
1.4 Zimbabwe national maize breeding programme 8
1.5 Overall objective 9
1.6 Specific objectives 9
References 10
CHAPTER 2 14
Literature review 14
2.1 Introduction 14
2.2 Major abiotic stress factors affecting maize production 14
2.2.1 Effects of drought on maize 15
2.2.2 Breeding for drought tolerance in maize 16
2.2.3 Suitable secondary traits used in selection for drought tolerance 18
2.2.4 Managed drought 19
2.2.5 Effects of low nitrogen on maize performance 20
2.2.6 Breeding for low nitrogen tolerance in maize 20
2.3 Combining ability and gene action 21
2.4 Heterosis and genetic diversity 23
2.4.1 Heterosis 23

iv
2.4.2 Heterotic groups 24
2.4.3 Genetic diversity and characterisation 26
2.4.4 Molecular markers 27
2.4.5 Choosing a marker 29
2.4.6 Single nucleotide polymorphism (SNP) 29
2.4.7 Correlation between genetic distance and heterosis 31
2.5 Genotype by environment interaction and assessment of stability 32
2.5.1 Additive main effects and multiplicative interaction 33
2.5.2 Genotype and genotype by environment interaction biplot analysis 34
2.6 Conclusions 34
2.7 References 35
CHAPTER 3 51
Combining ability between Zimbabwean and CIMMYT maize inbred lines
under stress and non-stress conditions 51
Abstract 51
3.1 Introduction 52
3.2 Materials and methods 54
3.2.1 Germplasm 54
3.2.2 Testing environments 54
3.2.3 Management 55
3.2.4 Experimental design and data collection 56
3.2.5 Statistical analysis 56
3.3 Results 58
3.3.1 Analysis of variance and hybrid mean performance across all environments 58
3.3.2 Performance per se of inbred lines 66
3.3.3 Combining ability and heritability 66
3.3.4 Correlation between grain yield and secondary traits 67
3.3.5 Relative contribution of general combining ability and specific combining 70
ability sums of squares to variation
3.3.6 Importance of maternal and paternal effects 71
3.3.7 General combining ability effects across environments 72

v
3.3.8 General combining ability effects under optimum conditions 77
3.3.9 General combining ability effects under managed drought conditions 79
3.3.10 General combining ability effects under low nitrogen conditions 79
3.3.11 Specific combining ability effects across all environments 81
3.3.12 Specific combining ability effects under optimum conditions 86
3.3.13 Specific combining ability effects under managed drought conditions 86
3.3.14 Specific combining ability effects under low nitrogen conditions 87
3.4 Discussion 88
3.5 Conclusions 95
3.6 References 96
CHAPTER 4 101
Genotype by environment interaction and stability analysis for grain yield of
single cross hybrids 101
Abstract 101
4.1 Introduction 102
4.2 Materials and methods 103
4.2.1 Germplasm 103
4.2.2 Sites 104
4.2.3 Experimental design and data collected 104
4.2.4 Statistical analysis 104
4.3 Results 106
4.3.1 Analysis of variance within years and across years 106
4.3.2 Additive main effect and multiplicative interaction analysis 107
4.3.3 Genotype and genotype by environment interaction biplot analysis for all 80
genotypes 108
4.3.4 Genotype and genotype by environment interaction biplot analysis for 20
best performing hybrids 115
4.4 Discussion 121
4.5 Conclusions 125
4.6 References 126
CHAPTER 5 130

vi
Genetic variation among CIMMYT and Zimbabwean maize inbred lines 130
Abstract 130
5.1 Introduction 133
5.2 Materials and methods 133
5.2.1 Germplasm selection 133
5.2.2 Site selection 133
5.2.3 Experimental design and morphological traits 133
5.2.4 Deoxyribonucleic acid extraction 133
5.2.5 Single nucleotide polymorphism genotyping 135
5.2.6 Statistical analysis 136
5.3 Results 138
5.3.1 Performance of inbred lines as measured using morphological traits 138
5.3.2 Correlation coefficients among morphological traits 144
5.3.3 Genetic distances and heterotic grouping among lines based on
morphological data 147
5.3.4 Single nucleotide polymorphism performance and quality 151
5.3.5 Genetic distances and heterotic grouping of lines based on single nucleotide
polymorphism markers 153
5.3.6 Comparison of dendrograms based on morphological and single nucleotide
polymorphism data 160
5.4 Discussion 161
5.5 Conclusions 167
5.6 References 168
CHAPTER 6 174
Relationships between heterosis, genetic distances and combining ability data
in maize hybrids 174
Abstract 174
6.1 Introduction 175
6.2 Materials and methods 177
6.2.1 Germplasm 177
6.2.2 Sites 177

vii
6.2.3 DNA extraction and SNP genotyping 177
6.2.4 Statistical analysis 177
6.3 Results 178
6.3.1 Grain yield, specific combining ability, mid- and high-parent heterosis across
all environments 179
6.3.2 Mean grain yield, specific combining ability, mid- and high-parent heterosis
under optimum conditions 182
6.3.3 Mean grain yield, specific combining ability, mid- and high-parent heterosis
under low nitrogen conditions 184
6.3.4 Mean grain yield, mid- and high-parent heterosis and specific combining
ability under drought conditions 184
6.3.5 Heterotic grouping in relation to field heterosis 185
6.3.6 Correlation between genetic distance, specific combining ability, high- and
mid-parent heterosis and F1 grain yield 188
6.4 Discussion 193
6.5 Conclusions 197
6.6 References 198
CHAPTER 7 202
Performance of F3 testcrosses developed from CIMMYT drought tolerant
donors and Zimbabwean elite inbred lines 202
Abstract 202
7.1 Introduction 203
7.2 Materials and methods 204
7.2.1 Germplasm 204
7.2.2 Evaluation sites 205
7.2.3 Management 206
7.2.4 Data collection and analysis 206
7.3 Results 208
7.3.1 Performance of early maturing testcrosses under managed drought conditions 208
7.3.2 Performance of early maturing testcrosses under optimum conditions 212
7.3.3 Performance of early maturing testcrosses across environments 212

viii
7.3.3.1 Variance components for early maturing testcrosses 215
7.3.3.2 Correlation between grain yield and secondary traits for early maturing
testcrosses under managed drought conditions 215
7.3.4 Performance of late maturing testcrosses under drought conditions 219
7.3.5 Performance of late maturing testcrosses under optimum conditions 222
7.3.6 Performance of late maturing testcrosses under combined environments 223
7.3.7 Variance components for late maturing testcrosses 226
7.3.8 Correlation between grain yield and secondary traits under managed drought
for late maturing testcrosses 226
7.4 Discussion 230
7.5 Conclusions 233
7.6 References 234
CHAPTER 8 238
Performance and yield prediction of three-way hybrids from drought tolerant
single cross hybrids 238
Abstract 238
8.1 Introduction 239
8.2 Materials and methods 240
8.2.1 Germplasm 240
8.2.2 Evaluation sites 241
8.2.3 Trial management 241
8.2.4 Management of drought site 242
8.2.5 Data collection 242
8.2.6 Statistical analysis 242
8.3 Results 242
8.3.1 Performance of hybrids under managed drought conditions 243
8.3.2 Performance of three way hybrids under optimum conditions 246
8.3.3 Combined analysis 248
8.3.4 Correlation between the predicted and observed mean yield 252
8.4 Discussion 253
8.5 Conclusions 256

ix
8.6 References 256
CHAPTER 9 259
General conclusions and recommendations 259
SUMMARY 264
OPSOMMING 266
Appendices 268
Appendix 1 Single cross hybrids 268
Appendix 2 Performance of genotypes for grain yield and other agronomic traits
across 14 environments in the 2009/10 and 2010/11 seasons 270
Appendix 3 Performance of genotypes for grain yield and other agronomic traits
across optimum sites in the 2009/10 and 2010/11 seasons 272
Appendix 4 Performance of genotypes for grain yield and other agronomic traits
across managed drought sites in the 2009/10 and 2010/11 seasons 274
Appendix 5 Performance of genotypes for grain yield and other agronomic traits
across low nitrogen sites 276
Appendix 6 Line general combining ability effects for grain yield across different
environments 278
Appendix 7 Tester general combining ability effects for grain yield across different
environments 278
Appendix 8 Mean grain yield (t ha-1) for 80 genotypes across seven environments 279
Appendix 9 Minor allele frequency and corresponding number of single nucleotide
polymorphism markers 281
Appendix 10 Polymorphic information content values and corresponding number
of single nucleotide polymorphism markers 282
Appendix 11 F1 mean grain yield (t ha-1), specific combining ability, mid- and
high-parent heterosis and genetic distance under optimum conditions 283
-1
Appendix 12 F1 mean grain yield (t ha ), specific combining ability, mid- and
high-parent heterosis and genetic distance under low nitrogen conditions 285
Appendix 13 Mean performance of three-way hybrids for grain yield and other
agronomic traits under managed drought in the 2011 winter season 287
Appendix 14 Mean performance of three-way hybrids for grain yield and other

x
agronomic traits under optimum conditions in the 2011 winter season 290
Appendix 15 Mean performance of three-way hybrids for grain yield and other
agronomic traits in combined analysis in the 2011 winter season 293

xi
 List of tables

Table Page
1.1 Maize area, yield and production for the 2010/11 season as
compared with the 2009/10 season 5
3.1 Germplasm used to produce the single cross hybrids 54
3.2 Amount of rainfall received and irrigation applied in the 2009/10 56
and 2010/11 seasons
3.3 Agronomic traits that were measured and derived 59
3.4 Combined analysis of variance of 14 sites in the 2009/10 and
2010/11 seasons for grain yield and other agronomic traits 61
3.5 Combined analysis of variance across 14 sites for senescence and
diseases in the 2009/10 and 2010/11 seasons 62
3.6 Performance of hybrids for grain yield and other agronomic traits
across 14 sites in the 2009/10 and 2010/11 seasons 63
3.7 Performance of hybrids for grain yield and other agronomic traits
across six optimum sites in the 2009/10 and 2010/11 seasons 64
3.8 Performance of hybrids for grain yield and other agronomic traits
across two managed drought sites in the 2009/10 and 2010/11 65
seasons
3.9 Performance of hybrids across two low nitrogen sites in the 2009/10
and 2010/11 seasons 67
-1
3.10 Performance of inbred parents for grain yield (t ha ) across
different environments in the 2009/10 and 2010/11 seasons 68
3.11 General and specific combining ability variances and heritability
estimates for the measured traits 69
3.12 Correlation coefficients between grain yield and other secondary
traits under managed drought conditions for hybrid trials 69
3.13 Correlation coefficients between grain yield and other secondary
traits under low nitrogen conditions for hybrid trials 70
3.14 Correlation coefficients between grain yield and other secondary

xii
 traits under optimum conditions for hybrid trials 71
3.15 Percentage of sum of squares attributable to general combining
ability and specific combining ability effects for yield and other
traits across sites as well as optimum, managed drought and low 72
nitrogen conditions
3.16 General combining ability due to female and male mean squares
under different environments 73
3.17 Line general combining ability values for other agronomic traits for
all environments 74
3.18 Tester general combining ability effects for other agronomic traits
under all environments 76
3.19 Line general combining ability effects for anthesis days and other
agronomic traits under optimum conditions 78
3.20 Tester general combining ability effects for grain yield and other
agronomic traits under optimum conditions 80
3.21 Line general combining ability effects for anthesis days and other
secondary traits under managed drought conditions 82
3.22 Line general combining ability effects of other agronomic traits
under low nitrogen conditions 83
3.23 Tester general combining ability effects of other agronomic traits
under low nitrogen conditions 84
3.24 Specific combining ability effects for grain yield across all
environments 85
3.25 Specific combining ability effects for anthesis days across all
environments 85
3.26 Specific combining ability for anthesis silking interval across all
environments 86
3.27 Specific combining ability for grain yield under optimum conditions 87
3.28 Specific combining ability effects under managed drought 87
conditions
3.29 Specific combining ability effects under low nitrogen conditions 88

xiii
4.1 Site annual average rainfall and soil type 104
4.2 Analysis of variance for grain yield across environments in the
2009/10 and 2010/11seasons 106
4.3 Combined analysis of variance for grain yield of 80 genotypes
across seven environments 107
4.4 Analysis of variance for additive main effects and multiplicative
interaction model for grain yield across seven environments for the
2009/10 and 2010/11 seasons 108
4.5 Additive main effects and multiplicative interaction analysis of
yield data of 80 maize genotypes tested across seven environments
in the 2009/10 and 2010/11 seasons 109
4.6 Correlation coefficients among test environments 114
4.7 Mean grain yield (t ha-1) for 20 genotypes across seven
environments in two seasons 117
5.1 Mean squares for grain yield and other agronomic traits across five
sites in the 2009/10 season 139
5.2 Mean squares for grain yield and other agronomic traits across five
sites in the 2010/11 season 139
5.3 Mean squares for grain yield and other traits in the 2009/10 and
2010/11 seasons 140
5.4 Mean performance of maize inbred lines for 14 traits evaluated in
the 2009/10 and 2010/11 seasons 141
5.5 Genetic and phenotypic variances and heritability estimates 144
5.6 Estimates of genotypic and phenotypic coefficients of variation and
genetic advance of the maize inbred lines across all environments in
the 2009/10 and 2010/11 seasons 144
5.7 Eigenvectors, eigenvalues, individual and cumulative percentage of
variation explained by first nine principal components for 14
morphological traits of maize inbred lines 145
5.8 Pearson coefficient correlations for grain yield and other
morphological traits measured from the inbred lines in the 2009/10

xiv
 and 2010/11 seasons 146
5.9 Estimates of genetic distances based on Euclidean distances and
morphological data for all pair-wise comparisons of 23 inbred lines 148
5.10 Distribution of single nucleotide polymorphism markers over the 10
maize chromosomes 153
5.11 Number of heterozygous loci and percentage homozygosity of
maize inbred lines 154
5.12 Estimates of genetic distances based on single nucleotide
polymorphism and Rogers’ distances for all pairwise comparisons 155
6.1 Hybrid mean grain yield, specific combining ability, mid- and high-
parent heterosis and genetic distance across all environments 180
6.2 F1, parental, mid- and high-parent heterosis means for anthesis days
and other agronomic traits across all environments 181
6.3 F1, parental, mid- and high-parent heterosis means for anthesis days
and other agronomic traits under optimum conditions 183
6.4 F1, parental, mid- and high-parent heterosis means for anthesis days
and other agronomic traits under low nitrogen conditions 183
6.5 F1, parental, mid- and high-parent heterosis means for anthesis days
and other agronomic traits under managed drought conditions 184
6.6 F1 mean grain yield (t ha-1), specific combining ability, mid- and
high-parent heterosis and genetic distance under optimum 185
conditions
6.7 F1 mean grain yield (t ha-1), specific combining ability, mid- and
high-parent heterosis and genetic distance under low nitrogen 186
conditions
6.8 Hybrid F1 grain yield (t ha-1), mid- and high-parent heterosis,
specific combining ability and genetic distance under drought 187
conditions
6.9 Mid- and high-parent heterosis of hybrids as well as known
heterotic groupings and grouping according to single nucleotide
polymorphism markers 188

xv
6.10 Average mid- and high-parent heterosis, and correlation among F1
grain yield, mid- and high-parent heterosis and specific combining
ability for all hybrids across all environments, optimum, drought
and low nitrogen environments 190
7.1 Pedigree, source and heterotic grouping of the inbred lines used to
develop the F3 population 205
7.2 Analysis of variance for grain yield under managed drought
conditions at Chisumbanje and Save Valley for early maturing
testcrosses in the 2011 winter season 209
7.3 Analysis of variance for anthesis days and other agronomic traits
under managed drought conditions across three sites for early
maturing testcrosses in the 2011 winter season 209
7.4 Analysis of variance for ears per plant, ear aspect, texture and ear
rot under managed drought conditions across three sites for early
maturing test crosses in the 2011 winter season 210
7.5 Performance of early maturing testcrosses for grain yield and other
agronomic traits under managed drought 211
7.6 Analysis of variance for grain yield for early maturing testcrosses
under optimum conditions in the 2011 winter season 213
7.7 Analysis of variance for anthesis days and other agronomic traits
under optimum conditions across three sites for early maturing
testcrosses in the 2011 winter season 213
7.8 Performance of early maturing testcrosses for grain yield and other
agronomic traits under optimum conditions 214
7.9 Analysis of variance for grain yield across environments for early
maturing testcrosses 215
7.10 Analysis of variance for anthesis days and other agronomic traits
under combined environments for early maturing testcrosses 216
7.11 Performance of early maturing testcrosses for grain yield and other
agronomic traits under drought and optimum conditions in the
2010/11 season 217

xvi
7.12 Genetic and phenotypic variance, repeatability and genetic gain for
early maturing testcrosses for the measured traits 219
7.13 Correlation coefficients between grain yield and secondary traits
under managed drought conditions 219
7.14 Analysis of variance for grain yield and other agronomic traits for
late maturing testcrosses under drought conditions in the 2011
winter season 220
7.15 Performance of late maturing testcrosses for grain yield and other
agronomic traits under drought conditions 221
7.16 Analysis of variance for grain yield for late maturing testcrosses
under optimum conditions in the 2011 winter season 223
7.17 Analysis of variance for anthesis silking interval and other
agronomic traits for late maturing testcrosses under optimum
conditions in the 2011 winter season 224
7.18 Performance of late maturing testcrosses for grain yield and other
agronomic traits under optimum conditions 225
7.19 Analysis of variance for grain yield for late testcrosses under both
drought and optimum conditions in the 2011 winter season 226
7.20 Analysis of variance for anthesis days and other agronomic traits for
late maturing testcrosses under drought and optimum conditions in
the 2011 winter season 227
7.21 Performance of late maturing testcrosses for grain yield and other
agronomic traits under drought and optimum conditions in the 2011
winter season 228
7.22 Genetic and phenotypic variances, repeatability, heritability and
genetic gain for grain yield and other agronomic traits measured in
late maturing testcrosses 230
7.23 Correlation of grain yield and secondary traits for late maturing
testcrosses under managed drought 230
8.1 Germplasm used in constituting the three-way hybrids 241
8.2 Analysis of variance for grain yield and other agronomic traits

xvii
 under managed drought conditions in the 2011 winter season 243
8.3 Mean performance for grain yield and other agronomic traits under
managed drought in the 2011 winter season 244
8.4 Pearson’s coefficient of correlation between grain yield and other
agronomic traits under managed drought conditions 245
8.5 Genotypic and phenotypic variances and broad sense heritability
estimates for the measured traits under managed drought conditions 246
8.6 ANOVA for grain yield and other agronomic traits under optimum
conditions in the 2011 winter season 247
8.7 Mean performance of three-way hybrids for grain yield and other
agronomic traits under optimum conditions in the 2011 winter 247
season
8.8 Pearson’s coefficient of correlation of grain yield with other
agronomic traits under optimum conditions 248
8.9 Genotypic and phenotypic variance estimates and broad sense
heritability of the agronomic traits under optimum conditions 249
8.10 Combined analysis of variance for agronomic traits in the 2011 249
winter season
8.11 Genotypic and phenotypic variances and broad sense heritability for
critical agronomic traits in combined analysis 250
8.12 Pearson’s correlation coefficients among agronomic variables in
combined analysis 250
8.13 Mean performance of the hybrids for grain yield and other
agronomic traits in combined analysis in the 2011 winter season 251

xviii
 List of figures

Figure Page
1.1a Sector contribution to national maize production in Zimbabwe in
the 2009/10 season 2
1.1b Sector contribution to national maize production in Zimbabwe in
the 2010/11 season 3
1.2 Maize production trends in Zimbabwe from 2000-2011 3
1.3 National yield comparison per sector in the 2009/10 and 2010/11
seasons in Zimbabwe 5
3.1 Line general combining ability (GCA) for grain yield for all
environments 74
3.2 Line general combining ability (GCA) values for anthesis days
for all environments 75
3.3 Tester general combining ability (GCA) effects for grain yield for
all environments 75
3.4 Tester general combining ability (GCA) effects for anthesis days
for all environments 76
3.5 Line general combining ability (GCA) effects for grain yield
under optimum conditions 77
3.6 Line general combining ability (GCA) effects for grain yield
under managed drought conditions 81
3.7 Line general combining ability (GCA) effects for grain yield
under low nitrogen conditions 82
3.8 Tester general combining ability (GCA) effects for grain yield
under low nitrogen conditions 83
4.1 Additive main effect and multiplicative interaction biplot for
genotype grain yield in seven environments for two seasons
combined 110
4.2 Additive main effect and multiplicative interaction biplot for
genotype grain yield across environments across two seasons 110

xix
4.3 Additive main effects and multiplicative interaction biplot for
environment means across two seasons 111
4.4 Genotype and genotype by environment interaction biplot
analysis of yield across seven environments and two seasons 111
4.5 Yield stability and performance of genotypes for seven
environments and two seasons 112
4.6 Polygon view of the genotype and genotype by environment
interaction biplot based on symmetrical scaling for the “which-
won-where” pattern for genotypes and environments 113
4.7 Genotype and genotype by environment interaction biplot based
on environment-focused scaling for environments 114
4.8 Hierarchical cluster analysis of the seven environments 116
4.9 Genotype and genotype by environment interaction biplot based
on genotype-focused scaling for the top 20 yielding genotypes 118
4.10 Grain yield stability and performance of the 20 top yielding
genotypes in seven environments across two seasons 119
4.11 Relationship amongst testing environments and genotype by
testing environments for the 20 top yielding genotypes
120
4.12 Genotype and genotype by environment interaction biplot based
on genotype and environment focused scaling for comparison of
genotypes and environments for the top 20 yielding genotypes 120
4.13 Polygon views of the genotype and genotype by environment
interaction biplot based on symmetrical scaling for the “which-
won-where” pattern for genotypes and environments for the 20
top yielding genotypes 121
5.1 Grain yield performance and ears per plant for the lines 142
5.2 Response of lines to ear rot and foliar diseases 142
5.3 Unweighted pair-group method with arithmetic average algorithm
cluster analysis of 23 maize inbred lines based on morphological
data combined over two seasons and seven locations 149

xx
5.4 Example of information extracted from each single nucleotide
polymorphism marker using the single nucleotide polymorphism
viewer. Data presented is for single nucleotide polymorphism
marker PHM12749_13, which detects a C/G single nucleotide
polymorphism in the maize genome 151
5.5 Frequency distribution of minor alleles among 23 inbred lines
based on 1 129 single nucleotide polymorphism (SNP) markers 152
5.6 Polymorphic information content (PIC) among 23 inbred lines
based on 1 129 single nucleotide polymorphism (SNP) markers 153
5.7 Neighbour-joining cluster analysis for the 23 maize inbred lines
based on Rogers’ dissimilarity coefficient using single nucleotide
polymorphism data 156
5.8 Neighbour-joining cluster analysis for the 19 maize inbred lines
based on Rogers’ dissimilarity coefficient using single nucleotide
polymorphism data (Lines with a high percentage missing data
excluded from the analysis) 158
5.9 Principal component analysis for 23 maize inbred lines based on
single nucleotide polymorphism data 159
6.1 The high- and mid-parent heterosis for 10 selected hybrids across
all environments 181
6.2 Relation of per se performance of hybrids with high- and mid-
parent heterosis under drought conditions 191
6.3 Relation of specific combining ability with high- and mid-parent
heterosis across all environments 191
6.4 Relation of specific combining ability with per se performance of
hybrids 192
6.5 Relation of genetic distance with high- and mid-parent heterosis
across all environments 192
6.6 Relation of genetic distance with specific combining ability
across all environments 193
7.1 Mean grain yield for early maturing testcrosses under optimum,

xxi
 drought and combined environments 218
7.2 Mean grain yield for early maturing testcrosses under drought
conditions across two environments in the 2011 winter season 218
7.3 Mean grain yield for late maturing testcrosses under optimum,
drought and combined environments in the 2011 winter season 229
7.4 Mean grain yield of late maturing testcrosses under drought
conditions across three environments in 2011 winter season 229
8.1 Predicted and observed mean yield for the best 10 and poorest 10
hybrids for environments contained environments 252

xxii
 Abbreviations and symbols
AEC An environment coordination
AFLP Amplified fragment length polymorphism
AMMI Additive main effects and multiplicative interaction
ANOVA Analysis of variance
ART Agricultural Research Trust
bp Base pairs
CA Communal area
CIMMYT International Maize and Wheat Improvement Center
cm Centimetre (s)
CML CIMMYT maize line
COI Crossover interaction
CRS Chiredzi Research Station
CTAB Cetyltrimethylammonium bromide
DNA Deoxyribonucleic acid
DR&SS Department of Research and Specialist Services
E Environment
EDTA Ethylenediaminetetra acetate
ET Exhollium turcicum
ExY Environment by year interaction
F1 First filial generation
F2 Second filial generation
F3 Third filial generation
FAM 6-carboxyfluorescein
FAO Food and Agriculture Organisation
FRET Fluorescence resonance energy transfer
g Gram (s)
G Genotype
GA Genetic advance
GCA General combining ability
GCAf General combining ability due to females

xxiii
GCAm General combining ability due to males
GCV Genotypic coefficient of variation
GD Genetic distance
GxE Genotype by environment interaction
GxLxY Genotype by location by year interaction
GGE Genotype and genotype by environment interaction
GLS Grey leaf spot
GxY Genotype by year interaction
ha Hectare (s)
h2B Broad sense heritability
HP High-parent
HPH High-parent heterosis
H2O Water
HRS Harare Research Station
IITA International Institute of Tropical Agriculture
IPCA Interaction principal component analysis
KASPar KBioscience competitive allele-specific polymerase chain reaction
kg ha-1 Kilogram per hectare
kg ha-1 yr-1 Kilogram per hectare per year
KRI Kadoma Research Institute
LSCFA Large scale commercial farmers
m Metre (s)
MARS Marker-assisted recurrent selection
MAS Marker-assisted selection
masl Metre (s) above sea level
Max Maximum
MET Multi-environment trial data
MEYT Multi-environment yield trial
MgCl2 Magnesium chloride
-1 -1
Mg ha cycle Megagram (s) per hectare per cycle
Min Minimum

xxiv
min Minute
ml Millitre
mm Millimetre (s)
mM Millimolar (s)
MP Mid-parent
MPH Mid-parent heterosis
MSV Maize streak virus
MT Metric ton
N Nitrogen
NaCl Sodium chloride
NARS National Agriculture Research Systems
NCDII North Carolina Design II
ng Nanogram (s)
NPPES Natal Potchefstroom Pearl Elite Selection
OPV Open pollinated variety
PC Principal component
PCA Principal component analysis
PCR Polymerase chain reaction
PCV Phenotypic coefficient of variation
pH Soil acidity or alkalinity
PIC Polymorphic information content
ppm Parts per million
QTL Quantitative trait loci
r Pearson correlation coefficient
R2 Coefficient of determination
RAPD Random amplified polymorphic DNA
RARS Rattray Arnold Research Station
RFLP Restriction fragment length polymorphism
ROX 6-Carboxyl-X-Rhodamine, succinimdyl ester
rpm Revolutions per minute
SAHN Sequential agglomerative hierarchical nested cluster analysis

xxv
SC Southern Cross
SCA Specific combining ability
sec Second (s)
SNP Single nucleotide polymorphism
SREG Site regression
SSR Simple sequence repeat
SVD Singular value decomposition
SV Singular value
t ha-1 Ton per hectare
Taq Thermus aquaticus
TE Tris/EDTA
Tris 2-amino-2-hydroxymethylpropane-1,3-diol
UK United Kingdom
UPGMA Unweighted pair-group method with arithmetic averages
USA United States of America
v/v Percent volume by volume
VIC 2΄chloro-7΄-phenyl-1,4-dichloro-6-carboxyfluorescein
w/v Percent weight by volume
Y Year
σ2g Genotypic variance
σ2p Phenotypic variance
σ2e Error variance
∑ Summation
% Percent
µl Microlitre
°C Degrees Celcius

xxvi
 CHAPTER 1
General introduction

1.1 Importance of maize and production constraints in Africa

In eastern, central and southern Africa, maize (Zea mays L.) is the major staple food crop
cultivated and consumed by most households. Approximately a quarter of a billion Africans
depend on maize as their staple food and they eat on average a quarter of a kilo or more
maize and maize products every day (African Press Agency, 2007). The successful and
continuous production of maize is key to global food security (Edmeades et al., 2000) and
any change leading to reduced production and subsequently reduced deliveries to the
markets, result in hunger especially in the disadvantaged communities (African Press
Agency, 2007). The two main abiotic stress factors that have hindered agricultural
production in the past include drought stress and poor soil fertility (Beck et al., 1996) and
will continue to have large negative effects on agricultural production in the coming years,
mostly in Asia and Africa (Rijsberman, 2006). The Food and Agricultural Organisation
(FAO) estimated that sub-Saharan Africa is the most severely affected region where almost
half of the land surface is exposed to a high risk of meteorological drought (Ribaut et al.,
2004). This effect is intensely influenced by continuing changes in the global climate (Hillel
and Rosenzweig, 2002). As water continues to be a limiting factor in crop cultivation,
breeding for drought tolerant genotypes becomes more and more imperative. Public and
private plant breeders endeavour to incorporate breeding for abiotic stress tolerance into their
breeding objectives in order to produce stress tolerant hybrids and open pollinated varieties.

1.2 Maize production in Zimbabwe

Maize is the principal food crop and is the main source of carbohydrates for the majority of
the Zimbabwe populace. The country requires 1.8 million ton of maize for human and animal
consumption and 300 000 ton as national strategic reserves per annum. It is produced by
large and small scale commercial farmers for both food (grain and fresh green maize) and
livestock feed (grain and silage). The requirement is divided into the following proportions;
64% for human consumption, 22% for livestock and poultry feed and 14% for other
industrial uses (Mashingaidze, 2006). Zimbabwe’s maize production trends are characterised

1
by extreme variability associated with the incidence of mid-season dry spells and high small
scale contribution to national production (Figure 1.1a and 1.1b). It also varies annually
according to input support programmes. As shown in Figure 1.2, national production has
been oscillating up and down due to various constraints that farmers faced over the years.
After 2001 the area under production continued to increase with the land reform programme
but on the other hand production remained low (Figure 1.2). The communal sector continues
to be the main producer of maize in the country (Figure 1.1a and 1.1b). These farmers are
faced with many challenges such as biotic and abiotic stresses, unavailability of inputs and
poorly adapted varieties. In the 2009/10 season the sector contributed 40% of the national
production, whilst in 2010/11 it contributed 43%. Of the total land area in Zimbabwe
approximately 50% is communal farming area, where about 70% of the population lives with
an average of 2 ha per household set aside for crop cultivation.

Figure 1.1a Sector contribution to national maize production in Zimbabwe in

2009/10 season.
OR=old resettlement; SSCFA=small scale commercial farmers; LSCFA=large scale commercial farmers;
CA= communal area; A1 and A2=newly resettled under land reform programme.
Source: AGRITEX Crop and Livestock Assessment Report, 2011.

2
While commercial maize production increased by two thirds between 1979 and 1985, small
scale production more than tripled (Rohrbach, 1989). The increase in small scale area under

Figure 1.1b Sector contribution to national maize production in Zimbabwe in

2010/11 season.
OR=old resettlement; SSCFA=small scale commercial farmers; LSCFA=large scale commercial farmers;
CA= communal area; A1 and A2=newly resettled under land reform programme.
Source: AGRITEX Crop and Livestock Assessment Report, 2011.
Production (ton)

Year
Figure 1.2 Maize production trends in Zimbabwe from 2000-2011.
Source: AGRITEX Crop and Livestock Assessment Report, 2011.

3
maize production was due to rapid expansion of government and private sector support for
small scale farmers after independence, major investments in market infrastructure,
expansion of a new smallholder credit programme, improved extension assistance and higher
maize prices (Rohrbach, 1989). Announcement by the Zimbabwe government in late 1986 of
a pre-planting producer price cut of 35% for deliveries of more than 91 metric ton (MT) saw
a 50% reduction in maize area planted by the large scale commercial sector, whilst the small
scale maize area remained roughly constant (Rohrbach, 1989). Small scale farmers had
effectively been granted primary responsibility for the production and supply of the nation’s
main staple.

In effect, the post-1979 surge in small scale production transformed the communal sector
from a relatively minor participant in the national maize economy to the principal source of
national production growth (Rohrbach, 1989). Hence the variability in national maize
production levels has increased with the growth of small scale maize production. Zimbabwe
is currently facing some of the largest fluctuations in cereal grain production of any country
in Africa. The 2010/11 maize production was estimated at 1 451 629 MT, from an area of 2
096 035 ha and an average yield of 0.69 t ha-1 (AGRITEX Crop and Livestock Assessment
Report, 2011). The production estimate was about 9% more than the 2009/10 production
estimate of about 1 327 572 MT. The maize area, yield estimate and production per province
are presented in Table 1.1. Mashonaland West had the highest and Matebeleland South the
lowest production. Generally high potential maize producing areas did not experience severe
dry spells, whereas the southern parts of the country namely Masvingo, Matebeleland North,
Matebeleland South and some parts of Manicaland, Midlands and Mashonaland East and
Central were affected by severe mid-season dry spells, which adversely affected production
in these areas (AGRITEX Crop and Livestock Assessment Report, 2011). There was an
increase in yield estimates from the 2009/10 to 2010/11 for large scale commercial farmers
(LSCFA), whilst for the rest of the sectors there was either a decrease or no change at all
(Figure 1.3). The yield for communal area (CA) remained below 0.5 t ha-1 for both seasons
and yet this is the sector contributing a large percentage to the total national production.

4
 Table 1.1 Maize area, yield and production for the 2010/11 season as compared with
the 2009/10 season

Area (ha) Production (MT) Yield t ha-1

% % %
Province 2010/11 2009/10 change 2010/11 2009/10 change 2010/11 2009/10 change

Manicaland 262 106 237 052 11 159 885 118 658 35 0.61 0.50 22
Mash Central 231 814 179 839 29 296 722 223 516 33 1.28 1.20 7
Mash East 247 511 243 995 1 148 507 181 994 -18 0.60 0.70 -14
Mash West 379 066 263 621 44 451 089 336 855 34 1.19 1.30 -8
Masvingo 276 105 229 887 20 80 070 56 201 43 0.29 0.20 45
Mat North 166 265 100 936 65 79 807 73 311 9 0.48 0.70 -31
Mat South 148 922 139 643 7 35 741 58 290 -39 0.24 0.40 -40
Midlands 384 246 408 569 -6 199 808 278 747 -28 0.52 0.70 -25
Total 2 096 035 1 803 542 16 1 451 629 1 327 572 9 0.69 0.70 -1.4
Mash=Mashonaland; Mat=Matebeleland.
Source: AGRITEX Crop and Livestock Assessment Report, 2011.
Yield t ha-1

Figure 1.3 National yield comparison per sector in the 2009/10 and 2010/11 seasons
in Zimbabwe.
OR=old resettlement; SSCFA=small scale commercial farmers; LSCFA=large scale commercial farmers;
CA=communal area; A1 and A2=newly resettled under land reform programme.
Source AGRITEX Crop and Livestock Assessment Report, 2011.

5
1.3 Maize production constraints in Zimbabwe
The communal farmers are faced with a lot of challenges, amongst them the occurrence of
dry spells, as they are mostly located in the drier parts of the country. The main maize
production constraints in Zimbabwe include drought stress, low soil fertility and
susceptibility to current major diseases. Downing (1992) found that with a temperature
increase of 2ºC the wet regions of Zimbabwe (with a water surplus) declined by a third from
9% to about 2.5% and that the drier regions will be double in area in future. A further
increase in temperature to 4ºC reduced the summer water surplus zones to less than 2% and a
similar scenario was observed in the 1991/92 season drought (Downing, 1992). About 160
million ha of maize is grown under rain-fed conditions globally and annual yield losses
attributed to drought are estimated at around 25% (Edmeades, 2008). The losses are expected
to be greater in sub-tropical countries that rely on unpredictable and erratic rainfall (Mhike et
al., 2011). When sub-Saharan Africa’s recurrent droughts ruin harvests, lives and livelihoods
are threatened and even destroyed and Zimbabwe is not exempted from these droughts.
Maize is affected by drought mainly through reduction of the growing season and through
erratic stress that occurs at any time during the growth of the crop. Annual maize production
in Zimbabwe ranges between 1.8 to 2.1 million ton with an average yield of 1.2 t ha-1 in the
small scale sector and 4.5 t ha-1 in the large scale commercial sector (Mhike et al, 2011).

International Maize and Wheat Improvement Center (CIMMYT) and International Institute
for Tropical Agriculture (IITA), working closely with various partners in sub-Saharan Africa,
have developed drought tolerant varieties that have benefited farmers, especially those
located in the drier regions. Farmers realise higher economic returns to labour, other inputs
and land with the use of drought tolerant varieties. Therefore development of new drought
tolerant genotypes can contribute to food security worldwide. New drought tolerant maize
varieties play a major role in alleviating the effects of drought that are expected to increase
due to global warming. In developing countries maize is also grown under low nitrogen (N)
conditions (McCown et al., 1992; Oikeh and Horst, 2001) and this is due to restricted N use
and low N uptake in drought susceptible areas, high price ratios between fertiliser and grain,
scarcity of fertiliser or lack of credit for farmers (Banziger et al., 1997). N availability is
estimated to be the principal limiting factor in more than 20% of arable land (Lafitte and

6
Edmeades, 1988). N deprivation hastens senescence of lower leaves (Wolfe et al., 1988;
Moll et al., 1994), reduces radiation use efficacy (Uhart and Andrade, 1995) and prolongs
anthesis silking interval (Jacobs and Pearson, 1991; Edmeades et al., 2000). These factors
result in maize being barren and eventually reduced yields.

Maize diseases of economic importance in Zimbabwe include maize streak virus (MSV),
grey leaf spot (GLS) and Exhollium turcicum (Leornard and Snuggs) (ET). MSV is
predominantly a disease of maize in Africa and the most devastating, although it has also
been reported in South and South East Asia (Shepherd et al., 2007). According to researchers
at IITA the disease was discovered in South Africa in 1901 (Shepherd et al., 2007). In
Zimbabwe the disease is most prevalent in irrigation schemes although it can also be found in
rain-fed crops and more specifically if the crop is planted late. MSV is transmitted by a
species of leaf hoppers belonging to the genus Cicadulina. The leaf hoppers are minute and
whitish in colour with a shape of an adult cockroach when observed under a magnifying lens
(CIMMYT, 2004). Maize yield losses attributed to MSV vary from season to season and
losses usually depend on the number of plants that are infected with the disease and the
crop’s growth stage when the infection took place. Approximately 50% of calories in local
diets emanate from maize, therefore yield losses attributed to MSV result in starvation and
overall food insecurity (Shepherd et al., 2007). Yield losses due to MSV range from close to
zero to nearly 100% (Stevens, 2008).

Another major disease causing yield losses in maize worldwide is grey leaf spot (GLS). The
disease is caused by the fungus Cercospora zea-maydis (Tehon and E.Y. Daniels). The
occurrence of the pathogen in KwaZulu-Natal, South Africa, in the late 1980s was its first
official report from the African continent and has since become pandemic, causing yield
losses of up to 60% (Ringer and Grybauskas, 1995). The ideal conditions that are conducive
for early season lesions and more severe disease attack include high early season rains and
prolonged periods of high humidity between November and December (Ringer and
Grybauskas, 1995). However, late season infections have been found to be more serious
because they affect the upper canopy which contributes 75-90% of the photosynthate for
grain filling (Allison and Watson, 1966).

7
1.4 Zimbabwe national maize breeding programme
Since its inception in 1909, the Zimbabwe National Maize Breeding Programme has
managed to develop high performance germplasm adapted to tropical and mid-altitude
growing regions roughly extending from 1 000-1 800 m above sea level (masl) and less than
23º from the equator (Doswell et al., 1996). Hybrid breeding in Zimbabwe started in 1932
and it was based on the populations Southern Cross, Salisbury White and to a lesser extent
Hickory King (Olver, 1988). The Southern Rhodesian Department of Agriculture imported
Hickory King, which was among a group of high yielding United States of America (USA)
open pollinated varieties and distributed it to farmers between 1900 and 1905 (Weinmann,
1972). A commercial single cross hybrid SR52 based on inbred lines SC5522 (SC from
Southern Cross) and N3-2-3-3 (N3 from Salisbury White) was released in 1960 (Doswell et
al., 1996).

Lines based on combining ability groups developed from material related to SC, N3 and
K64r/M162W are now the main components of hybrid breeding efforts by a majority of
national breeding programmes in eastern and the majority of southern African countries.
K64r originated from Kansas and is a direct import from the USA, whilst M162W is an
improved version of K64r ( Mickelson et al., 2001). Gene introgression has been done in the
national programme using germplasm mainly from CIMMYT, IITA and other National
Agriculture Research Systems (NARS). Some authors have emphasised the use of exotic
germplasm to widen the genetic base of germplasm used by maize breeders (Beck et al.,
1991; Vasal et al., 1992; Ron Para and Hallauer, 1997). Introducing exotic germplasm is
often suggested as a method to increase genetic diversity between populations in opposite
heterotic groups, thereby increasing the magnitude of heterosis.

In Zimbabwe, the predominant maize inbred lines used in the most successful and current
commercial hybrids and their derivatives were developed in the last century. Breeding gains
have not been significant in the National Breeding Programme in the last few years mainly
because of inadequate funding. As such, yield potential, disease resistance and drought stress
tolerance are lacking in most of the current maize germplasm in the National Breeding
Programme. There is thus an urgent need to improve the current elite lines used by the

8
National Breeding Programme to boost their yield potential and at the same time introduce
resistance to current major diseases and general tolerance to drought stress.

Gene introgression of drought tolerance and disease resistance genes from CIMMYT
germplasm into the National Breeding Programme elite inbred lines has been initiated at the
Department of Research and Specialists Services (DR&SS) in Zimbabwe. In order to
determine the best parents for this project, it is important to understand the heterotic
relationships between the CIMMYT and National Breeding Programme lines with a view to
selecting good parents to initiate crosses for pedigree, backcross and potential marker-
assisted recurrent selection (MARS) populations for line extraction. The resultant new lines
will then be used in developing new improved drought tolerant and disease resistant hybrids
and open pollinated varieties (OPVs) for release. In addition, classification of inbred lines
into heterotic groups will facilitate exploitation of heterosis which can contribute to hybrid
performance.

1.5 Overall objective

The overall objective of the study was to identify improvement strategies for yield potential
and tolerance to biotic and abiotic stress factors of Zimbabwean maize inbred lines

1.6 Specific objectives

(i) To estimate combining ability and heterosis for grain yield and other agronomic traits
between DR&SS and CIMMYT white maize inbred lines under stress and non-stress
environments
(ii) To analyse genotype by environment (G x E) interaction and stability of single cross
hybrids for grain yield
(iii) To examine genetic diversity among DR&SS and CIMMYT white maize inbred lines
using single nucleotide polymorphism (SNP) analysis and morphological traits
(iv) To assess the relationship between genetic diversity of DR&SS and CIMMYT
parental inbred lines and F1 performance, heterosis and specific combining ability
(SCA) effects of hybrids under abiotic stress and non-stress environments

9
(v) To estimate test-cross performance of F3 segregating populations developed from
CIMMYT drought tolerant donors and DR&SS elite inbred lines under drought and
non-drought conditions
(vi) To estimate performance and yield prediction of three-way hybrids from drought
tolerant single cross hybrids.

1.7 References
African Press Agency (APA), International Maize and Wheat Improvement Center. 2007.
Enhanced conditions - Drought tolerant maize to give African farmers options even
with global warming. Science/Daily. http://www.sciencedaily.com. Accessed online
25 March 2009.
AGRITEX Crop and Livestock Assessment Report. 2011. Second Round Crop and Livestock
Assessment Report. Ministry of Agriculture, Mechanisation and Irrigation
Development, Harare, Zimbabwe. pp 4-17.
Allison, J.C. and D.J. Watson. 1966. The production and distribution of dry matter in maize
after flowering. Annals of Botany 30: 365-381.
Banziger, M., F.J. Betran and H.R. Lafitte. 1997. Efficiency of high-nitrogen selection
environments for improving maize for low-nitrogen target environments. Crop
Science 37: 1103-1109.
Beck, D., F.J. Betran, M. Banziger, J.M. Ribaut, M. Willcox, S.K. Vasal and A. Ortega.
1996. Progress in developing drought and low soil nitrogen tolerance in maize. In:
Wilkinson, D.B. (ed) Proceedings of the 51st Annual Corn and Sorghum Research
Conference. Washington ASTA. pp 85-111.
Beck, D.L., S.K. Vasal and J. Crossa. 1991. Heterosis and combining ability among
subtropical and temperate intermediate maturity maize germplasm. Crop Science 31:
68-73.
CIMMYT. 2004. Maize Diseases: A guide for field identification. 4th Edition, Mexico, D.F.:
CIMMYT.
Doswell, C.R., R.L. Paliwal and R.P. Cantrell. 1996. Maize in the Third World. Westview
Press. U.S.A.

10
Downing, T.E. 1992. Climate change and vulnerable places: Global Food Security and
Country Studies in Zimbabwe, Kenya, Senegal and Chile. Research Report No. 1.
Environmental Change Unit, University of Oxford. pp 54.
Edmeades, G.O. 2008. Drought tolerance in maize: An emerging reality. In: James, C. (ed)
Global Status of Commercialised Biotechnology/GM Crops. ISAAA Brief 39: 196-
315.
Edmeades G.O., J. Bolanos, A. Elings, J.M. Ribaut, M. Banziger and M.E. Westgate. 2000.
The role and regulation of the anthesis-silking interval in maize. In: Westagate, M.E.
and K.J. Boote (eds) Physiology and modeling kernel set in maize. CSSA Special
Publication no. 29 CSSA. Madison WI. pp 43-73.
Hillel D. and C. Rosenzweig. 2002. Desertification in relation to climate variability and
change. Advances in Agronomy 77: 1-38.
Jacobs, B.C. and C.J. Pearson. 1991. Potential yield of maize, determined by rates of growth
and development of ears. Field Crops Research 27: 281-298.
Lafitte, H.R. and G.O. Edmeades. 1988. An update on selection under stress: Selection
Criteria. In: Gelaw, B. (ed) Second Eastern, Central and Southern African Regional
Maize Workshop. The College Press, Harare, Zimbabwe. pp 309-331.
Mashingaidze, K. 2006. Maize research and development. In: Rukuni, M., P. Tawonezvi and
C. Eicher (eds) Zimbabwe’s Agricultural Revolution Revisited. University of
Zimbabwe Publication. pp 363-377.
McCown, R.L., B.A. Keating, M.E. Probert and R.K. Jones. 1992. Strategies for sustainable
crop production in semi-arid Africa. Outlook Agriculture 21: 21-31.
Mhike, X., P. Okori, C. Magorokosho and T. Ndhlela. 2011. Validation of the use of
secondary traits and selection indices for drought tolerance in tropical maize (Zea
mays L.) African Journal of Plant Science 5: 96-102.
Mickelson, H.R., H. Cordova, K.V. Pixley and M.S. Bjarnason. 2001. Heterotic relationships
among nine temperate and subtropical maize populations. Crop Science 41: 1012-
1020.
Moll, R.H., W.A. Jackson and R.L. Mikkelsen. 1994. Recurrent selection for maize grain
yield: dry matter and nitrogen accumulation and partitioning changes. Crop Science
34: 874-881.

11
Oikeh, S.O. and W.J. Horst. 2001. Agro-physiological responses of tropical maize cultivars
to nitrogen fertilization in the moist savannah of West Africa. In: Horst, W.J. (ed)
Plant Nutrition-Food Security and Sustainability of Agro-ecosystems. Kluwer
Academic Publishers, Dordrecht, Netherlands. pp 804-805.
Olver, R.C. 1988. Zimbabwe maize breeding program. Towards self sufficiency: Proceedings
of the Second Eastern, Central and Southern Africa Regional Maize Workshop.
March 15-21. CIMMYT, Harare pp 34-43.
Ribaut, J.M., D. Hoisington, M. Banziger, T.L. Setter and G.O. Edmeades. 2004. Genetic
dissection of drought tolerance in maize: A case study. In: Nguyen, H.T. and A. Blum
(eds) Physiology and Biotechnology Intergration for Plant Breeding. Marcel Dekker
Inc, New York. pp 571-609.
Rijsberman, F.R. 2006. Water scarcity: Fact or fiction? Agricultural Water Management 80:
5- 22.
Ringer, C.E. and A.P. Grybauskas. 1995. Infection cycle components and disease progress of
grey leaf spot on field cover. Plant Disease 79: 24-28.
Rohrbach, D.D. 1989. The economics of smallholder maize production in Zimbabwe:
Implicaions for food security. Paper No. 11, MSU International Development Papers.
Department of Agricultural Economics, Michigan State University, East Lansing,
Michigan.
Ron Parra, J. and A.R. Hallauer. 1997. Utilisation of exotic maize germplasm. Plant
Breeding Revolution 14:165-187.
Shepherd, D.N., T. Mangwende, D.P. Martin, M. Bezvidenhout, J.A. Thompson and E.P.
Rybicki. 2007. Inhibition of maize streak virus (MSV) replication by transient and
transgenic expression of MSV replication-associated protein mutants. Journal of
General Virology 88: 325-336.
Stevens, R. 2008. Review: Prospects for using marker assisted breeding to improve maize
production in Africa. Journal of the Science of Food and Agriculture 88: 745-755.
Uhart, S.A. and F.H. Andrade. 1995. Nitrogen deficiency in maize: I. Effects on crop growth,
development, dry matter partitioning and kernel set. Crop Science 35: 1367-1383.

12
Vasal, S.K., G. Srinivasan, D.L. Beck, J. Crossa, S. Pandey and C. De Leon. 1992. Heterosis
and combining ability of CIMMYT’s tropical late white maize germplasm. Maydica
37: 217-223.
Weinmann, H. 1972. Agriculural research and development in Southern Rhodesia, 1890-
1923. Department of Agriculture Occasional Paper 4. Harare: University of
Zimbabwe. pp 69-70.
Wolfe, D.W., D.W. Henderson, T.C. Hsiao and A. Alvino. 1988. Interactive water and
nitrogen effects on senescence of maize. I. Leaf area duration, nitrogen distribution
and yield. Agronomy Journal 80: 859-864.

13
 CHAPTER 2
Literature review

2.1 Introduction
Literature on principles of main concepts that are relevant in this research is reviewed in this
chapter. These include i) understanding the effects of drought on maize, progress in breeding
for drought, secondary traits used in selection for drought as well as managed drought
screening, effects and breeding for low N conditions in maize, ii) combining ability, heterosis
and heterotic groups, iii) genetic characterisation, molecular markers, SNPs and correlation
between heterosis and genetic distances and iv) G x E interaction.

2.2 Major abiotic stress factors affecting maize production

Drought and low N are the two major abiotic stress factors affecting maize production in sub-
Saharan Africa. In Zimbabwe the major maize producers are small scale farmers who are
mainly located in dry regions of the country with low soil inherent fertility. Initial efforts in
the National Breeding Programme were towards breeding maize varieties for high rainfall
regions with optimum fertilisation. It is therefore important that efforts are made towards
improving new maize varieties for tolerance to these stresses. Currently few drought and low
N tolerant maize varieties have been released by the National Breeding Programme. Maize
yields are mostly affected by drought through reduction of the growing season and erratic
mid-season dry spells that take place at any time during the growth of the crop (Edmeades et
al., 1994). Maize is mainly susceptible to drought stress that takes place just before and
during flowering when its yield potential is determined (Malosetti et al., 2007).

Drought is a water deficit in the plant’s environment that has the potential to reduce crop
yield (Cooper et al., 2006). It has devastating economical and sociological effects. Drought
incidents are predicted to increase due to long term effects of global warming (Cook et al.,
2007). It is difficult to forecast manifestation of natural drought making it challenging or
almost impossible to differentiate between stress and non-stress agricultural systems (Cooper
et al., 2006). In the semi-arid tropics the effect of drought is intensified by extremely erratic

14
rainfall, high temperatures, high levels of solar radiation and poor soil productiveness (Cook
et al., 2007).

2.2.1 Effects of drought on maize

Maize inflorescence consists of separate male and female flowers making it more vulnerable
to drought stress during flowering time (Prine, 1971; Grant et al., 1989). Tassel development
and pollen shed in maize are less sensitive to fluctuations in moisture availability compared
to silk growth. The allocation of nutrients to ears, ovules and silks is reduced under drought
as a result of the dominance effects of the apical tassel. Silk emergence in relation to male
flowering is delayed when drought takes place just before flowering and this result in an
increased anthesis silking interval (Bolanos and Edmeades, 1993a). When the anthesis
silking interval is lengthened the pollen might arrive when silks have dried up (Bassetti and
Westgate, 1993) or after ovaries have used up their starch reserves (Saini and Westgate,
2000; Zinselmeier et al., 2000). This scenario results in retarded ear and silk growth and
accelerated kernel and ear abortion (Westgate and Boyer, 1986; Edmeades et al., 1993).

The maize crop has been found to be more susceptible to moisture stress one week before to
two weeks after flowering (Grant et al., 1989). Grain abortion normally takes place during
the first 2-3 weeks after the emergence of silks (Westgate and Boyer, 1986; Schussler and
Westgate, 1991). It is intensified by any stress that decreases canopy photosynthesis and
movement of assimilates to the developing ear. This scenario results in the growing ear being
deprived of the necessary nutrients (Stevens, 2008). Therefore the amount of assimilates
reduces to below threshold levels required to sustain grain development and growth
(Edmeades and Daynard, 1979; Tollenaar et al., 1992). The decrease in photosynthesis can
be due to a decrease in radiation interception associated with increased leaf rolling (Bolanos
et al., 1993). Reduction in photosynthetic rate decreases the volume of nutrients available for
distribution to the sink organs (Kim et al., 2000). The amount of stress that drought imposes
on the maize crop results in modifications of photosynthetic pigments and constituents
(Loggini et al., 1999). It also causes damage to photosynthetic organs (Fu and Huang, 2001)
and the calvin cycle enzyme activity is reduced (Monakhova and Cheryadév, 2004).
Carbohydrate metabolism activity in the plant’s reproductive organs is also negatively

15
affected (Liu et al., 2004). Maize is more vulnerable to drought compared to sorghum as a
result of its shallow root system, enlarged leaf surface area, increased transpiration rate,
slower grain development rate and extended grain filling period (Sinclair and Muchow,
2001).

2.2.2 Breeding for drought tolerance in maize

Maize is considered the most susceptible cereal to drought stress, with the exception of rice
(Banziger and Araus, 2007). Maize yields remain below 2 t ha-1 in most countries in sub-
Saharan Africa and yields vary from year to year (FAO, 2011). Maize is the staple food crop
of importance to over 300 million people in eastern and southern African countries (Heisey
and Edmeades, 1999). Rainfall distribution and amount have been found to have a direct
effect on maize productivity in these two regions. In southern Africa the 2002/03 drought left
about 14 million people exposed to starvation and the food deficit was 3.3 MT (World Food
Programme, 2003). The World Food Programme was expected to provide food aid to 7.8
million people in five East African countries (Somalia, Ethiopia, Djibout, Kenya and
Uganda) as a result of consecutive seasons of drought (World Food Progamme, 2009).
Again East Africa experienced a severe drought in 2011 that left more than 10 million people
relying on food aid (World Food Programme, 2011). Therefore, in order for farmers to
realise increased and stable yields and for seed merchants to be in a position to market a
variety widely, it is critical that drought tolerance is incorporated in maize breeding strategies
(Campos et al., 2004). Hence, improvement or development of maize genotypes with high
and constant yields under drought stress conditions is essential.

Among abiotic stresses, breeding for drought tolerance is one of the most challenging
endevours, because selected germplasm ought to perform exceptionally well not only under
drought stress but also under optimum conditions. Since water is a scarce resource,
improving varieties for drought tolerance is an important approach in reducing this problem.
It is important in breeding for drought tolerance to consider breeding for other stress factors
as well (Beebe et al., 2008). Progress in breeding for drought tolerance has been slow as a
result of the complex nature of the trait and an improved understanding of the fundamental
mechanisms of drought would hasten progress in breeding for the trait (Ribaut et al., 2002).

16
In an effort to improve maize productivity, maize breeders have exerted enormous efforts to
breed hybrids with drought tolerance (Bruce et al., 2002). The efficiency in selection of
germplasm for drought tolerance can be improved through use of managed drought
environments. This can be done during the off-season (winter) with the use of controlled
irrigation whereby the occurrence, extent and amount of drought stress on the crop are
controlled (Banziger et al., 2000). As a result of significant G x E interaction, it is important
that genotypes screened for drought tolerance are evaluated in the target locations before they
are incorporated as parents in the breeding programmes.

Although progress in drought tolerance can be achieved through conventional selection

methods, trials must usually be replicated across a large number of locations and across
several years before the expression of the trait can be conclusively identified. Stevens (2008)
reported that CIMMYT has been involved in breeding for drought tolerance in maize over
the past three decades. A unique selection programme aimed at improving tolerance to
drought using the lowland tropical maize populations Tuxpeño Sequía, LaPosta Sequia and
Pool 26 Sequia was started at CIMMYT in the 1970s. Average maize yields under drought
conditions were 126 kg ha-1 per cycle (Stevens, 2008). Progress in selection for drought
tolerance of Tuxpeño Sequía resulted in average breeding gain of 108 kg ha-1 yr-1 with yield
levels ranging from 1-8 t ha-1 (Bolanos and Edmeades, 1993a). A significant gain of 9.4% for
ears per plant and reduction of anthesis silking interval from 34.2 days in the first cycle to 9.8
days in cycle eight were further reported by Bolanos and Edmeades (1993a). Bolanos and
Edmeades (1993b) reported an increase in anthesis silking interval to 18.8 days under severe
stress. The three populations outperformed their original cycles of selection and check
varieties under drought conditions and yields ranged from 1.0-4.5 Mg ha-1 and 5.8-10.4 Mg
ha-1 under drought and optimum conditions respectively (Edmeades et al., 1999). The authors
further reported yield gains ranging from 0.08 - 0.29 Mg ha-1 cycle-1 under drought
conditions and 0.04 - 0.18 Mg ha-1 cycle-1 under optimum conditions. Campos et al. (2004)
reported 146 kg ha-1 gains annually when drought stress was introduced at flowering and 76
kg ha-1 when stress was introduced at mid-grain filling stage. Drought tolerant germplasm
was introgressed into local African germplasm, producing varieties with stable and superior
performance in southern and eastern Africa. Superior maize OPVs with yield performance

17
comparable to commercial hybrids under moderate to severe moisture stress were also
produced.

Grain yield is considered the primary trait for selection under drought stress conditions.
Nonetheless, reduced heritability and variance of yield components make selection based
only on grain yield inefficient (Stevens, 2008). The major strategy in breeding for drought
tolerance in maize has been direct selection for high yield. Grain yield under severe mid and
late season moisture stress has been improved by 30-50% in three late maturing maize
populations through recurrent selection at rates of up to 12% per selection cycle (Edmeades
et al., 1999). It has been reported that use of both secondary traits and grain yield in
improving germplasm for drought stress tolerance has resulted in significant selection
progress (Mhike et al., 2011).

2.2.3 Suitable secondary traits used in selection for drought tolerance

Appropriate secondary traits selected for under drought stress should be genetically related to
grain yield, have high heritability estimates, be consistent and easy to select for and not too
expensive. These traits must furthermore be measurable at or before flowering so that
undesirable parents are selected against and these traits should not be accompanied by yield
loss under optimum environments (Edmeades et al., 1998). Critical secondary traits under
drought identified at CIMMYT and Pioneer Hi-Bred include reduced prolificacy, anthesis
silking interval, stay green and to a lesser extent leaf rolling (Banziger et al., 2000). Anthesis
silking interval is measured as the number of days between silk emergence and pollen
shedding and increases under drought stress as a result of retarded ear and silk growth
(Bolanos and Edmeades, 1993b). Work done at CIMMYT has revealed that the heritability
estimate for anthesis silking interval is related to or greater than the heritability for grain
yield. A high negative correlation of anthesis silking interval with grain yield and other
related traits such as kernel number and number of ears per plant has been reported. Anthesis
silking interval and ears per plant have been widely used in breeding for stress tolerance in
maize (Bolanos and Edmeades, 1993b; Banziger et al., 2000). These two traits have shown
good genetic variability under drought stress conditions and high heritability.

18
The variation in number of kernels has a major effect on maize grain yield under drought
(Bolanos and Edmeades, 1996). Bolanos and Edmeades (1993a) observed a 90% drop in
yield as anthesis silking interval increased from -0.4-10 days, whilst Du Plessis and Dijkhuis
(1967) reported 82% drop in grain yield as anthesis silking interval increased from 0-28 days.
In genotypes selected for short anthesis silking intervals and increased grain yield under
drought the bulk of the carbohydrates are channeled towards development of the ear and less
towards the growth of tassels and vegetative organs (Edmeades et al., 1993). In tropical
maize gains in selection have been linked with improved synchronisation in silking and
pollen shedding, reduced barreness, reduced tassel size, increased harvest index, delayed leaf
senescence and reduced root length density in the upper soil profile with no alterations in
water uptake or biomass (Bolanos and Edmeades, 1993a; b; Bolanos et al., 1993; Chapman
and Edmeades, 1999). Genotypes are selected under managed drought stress based on grain
yield performance and appropriate secondary traits.

2.2.4 Managed drought

Managed drought stress screening is usually done off-season (winter) with the use of
irrigation. Drought stress on genotypes is induced either at flowering or at grain filling stage.
At intermediate stress level average grain yield is targeted to reduce by 15-30% of yields
expected under optimum conditions and the stress will be targeting grain filling. A yield
reduction of 30-60% of yields realised under optimum conditions is targeted for severe stress
levels and the stress affects both flowering and grain filling (Banziger et al., 2000). Under
severe stress, irrigation is scheduled such that drought stress coincides with anthesis and silk
emergence, but supplementary irrigation is applied 14 days after the end of pollen shedding
in order to facilitate adequate grain filling of the formed grain. In intermediate stress, drought
stress is timed to coincide with grain filling. It is important to ensure that irrigation is
uniformly applied before onset of stress as this will result in stress levels being uniform in all
genotypes, more constant plant performance and eventually improved breeding progress
(Banziger et al., 2000).

19
2.2.5 Effects of low nitrogen on maize performance
Grain yield in tropical maize is negatively affected by the abiotic stress factors drought and
low N (Pingali and Pandey, 2001). N is one of the major nutrients required by plants in large
quantities. N use efficiency is always low in dry areas and has become one of the most
limiting factors in crop yield improvement. It has been reported that besides providing
nutrients for crop development, application of N may possibly lead to improved drought
tolerance and enhanced yield (Zaman and Das, 1991; Xu et al., 2005). Kernel abortion in
maize has been seen to be aggravated by N stress thereby leading to reduced grain number
(Lemcoff and Loomis, 1986; Pearson and Jacobs, 1987; Uhart and Andrade, 1995a; b).
Roughly 85% of kernel abortion takes place in the course of the first 20 days after silk
emergence (Monneveux et al., 2005). N plays a crucial role in the anti-oxidant defence
enzyme and lipid peroxidation metabolism under stress environments (Sun et al., 2001;
Saneoka et al., 2004). N deficiency has also been reported to negatively affect leaf
expansion, emergence rate, radiation interception radiation use efficiency and assimilate
distribution amongst vegetative and reproductive organs (Uhart and Andrade, 1995a).
Reduction in kernel number and number of ears due to low N has been reported by Lemcoff
and Loomis (1986); Pearson and Jacobs (1987); Uhart and Andrade (1995a; b) and
Monneveux et al. (2005). Prolonged anthesis silking interval has also been reported as a
result of N deficiency (Jacobs and Pearson, 1991) and accelerated senescence (Moll et al.,
1994). Low N affects maize growth throughout the life cycle of the crop compared to drought
that occurs at any time during the growth of the crop (Banziger and Araus, 2007).

2.2.6 Breeding for low nitrogen tolerance in maize

Maize breeders have made enormous efforts towards screening maize germplasm for
tolerance to low N conditions. However, headway in screening for low N tolerance has been
slown down by pronounced genotype x season and genotype x location interactions (Ribaut
et al., 2007). Lafitte and Edmeades (1988) reported that N unavailability is projected to be
the major limiting factor in more than 20% of arable land. The use of inorganic fertiliser in
sub-Saharan Africa is on the lower side mainly as a result of unavailability of fertiliser and
ever escalating costs. Cultivars tolerant to low N have been found to be efficient in utilising

20
available N, mainly as a result of increased uptake capacity or their ability to efficiently
utilise absorbed N in grain production (Lafitte and Edmeades, 1994).

As a result of low heritability estimates for grain yield under low N environments, the
utilisation of secondary traits in the selection process has frequently been recommended
(Lafitte et al., 2003). Anthesis silking interval, leaf senescence and ears per plant have been
suggested as ideal secondary traits to select for when improving maize genotypes for low N
conditions (Banziger and Lafitte, 1997; Banziger et al., 2000). Selection indices centered
around these traits were established and significantly improved the selection efficacy under
low N stress environments (Banziger and Lafitte, 1997). It is simpler to breed for low N than
for drought conditions mainly because unavailability of N affects plant growth in a more
even manner unlike drought periods that occur randomly (Banziger et al., 2000). It has been
found that screening germplasm under severe low N conditions should be adequate to infer
low N stress tolerance for different levels of N deficiency. Maize germplasm has been
selected for both drought and low N tolerance using combining ability effects. Testers known
to be drought or low N tolerant are crossed with selected lines and the progeny is evaluated
for tolerance to the two stresses. Ideal parents are identified as having good general
combining ability (GCA) under these environments; therefore it is critical to consider looking
at combining ability effects of lines under evaluation in the current study.

2.3 Combining ability and gene action

Combining ability cannot be predicted from the parental phenotype and is assessed by
progeny testing from controlled pollinations. It was initially a broad notion considered jointly
for categorising an inbred line in relation to its hybrid performance, but was later refined as
GCA and SCA. This had a major influence on inbred line evaluation and population
development in maize breeding (Sprague and Tatum, 1942). Sprague and Tatum (1942)
defined GCA as the average performance of a line in hybrid combination and SCA as those
cases where some hybrid combinations are either superior or inferior than would be
estimated on the average performance of the parental inbred lines.

21
SCA effects should be used in combination with hybrid performance and GCA of particular
parents for selection (Shukla and Pandey, 2008). The GCA component is primarily a function
of additive gene action while SCA variance is mainly a function of dominance variance.
Evaluation of GCA effects of hybrid parents is essential to critic their appropriateness for
hybrid development, since the mean performance of parental lines does not always translate
to their GCA effects. GCA and SCA effects are important tools used by breeders in selecting
superior parents for developing crosses (Shukla and Pandey, 2008). Favourable alleles are
combined through hybrid combination with high per se performance with good SCA
estimates and having at least one of the parents with high GCA (Marilia et al., 2001). In
choosing ideal parents and crosses and to estimate the combining abilities of parents in early
generations, plant breeders have used the line x tester analysis method. Line x tester analysis
provides an efficient approach for identification of suitable parents and crosses exhibiting
good performance in traits under consideration (Ahuja and Dhayal, 2007).

GCA and SCA variances are used to deduce gene action. It is defined as the way genes
express themselves in which case GCA effects represent additive gene action, whilst SCA
effects represent non-additive gene action. Several mating designs have been used for
estimating gene action amongst them the North Carolina Design II (NCDII) (Comstock and
Robinson, 1948). The NCDII crosses are used for defining the cumulative gene effects of
breeding populations and for estimating GCA and SCA effects. In the current study NCDII
was used to form hybrids among 23 inbred lines that were to be evaluated for drought and
low N tolerance and the method was chosen because the objective was mainly to estimate
GCA and SCA variances. It is critical to understand gene action for grain yield and other
secondary traits so that effective breeding strategies are developed for stress tolerance
breeding without compromising on yield. Betran et al. (2003) reported additive gene action
being more important under drought conditions and non-additive gene action being important
under low N conditions. Their results suggested potential benefits of incorporating drought
tolerance in both parental inbred lines in order to enhance performance of hybrids under
drought conditions. A higher SCA variance compared to GCA was reported by Devi and
Singh (2011), suggesting that non-additive gene action played a more important role in
determining grain yield. Amount of heterosis realised in a cross depends among other things

22
on the differences of gene frequency between the crossed lines (Falconer, 1981). Once ideal
parents are identified it is important to assess the levels of heterosis and possible heterotic
relationships among these lines.

2.4 Heterosis and genetic diversity

2.4.1 Heterosis
The first step in the search for heterosis in crop improvement is a full characterisation of
available genetic diversity, which forms the basis for the analysis of combining ability of
inbred lines (Verbitskaya et al., 1999; Diniz et al., 2005). According to Falconer (1981)
heterosis is the product of directional dominance and square of differences in gene frequency
in parents. It is therefore obvious that the presence of both dominance and initial differences
in gene frequency in parental lines causes heterosis in F1 progeny. Heterosis will be
significant when alleles in both parents are in a homozygous state (Falconer, 1981). Success
of the formal seed sector involving maize and other crops has been attributed to
heterosis/hybrid vigour. Manipulating heterosis in breeding facilitates yield improvement and
helps augment many other necessary quantitative and qualitative traits in crops. Heterosis is
observed in a situation where offspring obtained from crossing two inbred lines or
populations perform above the mean of the two populations or lines for the trait under
consideration. It is usually seen as increased growth rate, size, yield and other traits in F1
individuals produced after crossing inbred lines (Melchinger and Gumber, 1998; Tollenaar et
al., 2004). Remarkable maize yield increases observed in the USA between the 1930s and
1970s came about through manipulation of heterosis (Duvick, 2001).

The magnitude of heterosis between lines or populations has assisted in determining the level
of genetic diversity. It provides a base for choosing germplasm to be used as parents for
developing segregating populations to be used in a reciprocal recurrent selection breeding
programme (Ortiz et al., 2008). Crosses produced from distantly related parents normally
exhibit higher levels of heterosis compared to crosses produced from closely related parents
and this can be partly explained by the level of genetic diversity that exists between the
parents (Fabrizius et al., 1998). The success of any hybrid breeding programme depends on

23
the existence of reasonable levels of heterosis and a favourable environment for economic
hybrid seed production (Shukla and Pandey, 2008). The effective manipulation of intra- and
inter-sub specific heterosis hinges on the genetic diversity of the parents, gene action, linked
hybrid vigour and the biological achievability of hybrid seed production (Shukla and Pandey,
2008).

Results from studies of heterosis for grain yield in maize until 1979 were summarised by
Hallauer and Miranda (1988) and mid-parent heterosis (MPH) ranged between -3.6% and
72.0%, whilst high-parent heterosis (HPH) ranged between -9.9% and 43.0%. Recently
Betran et al. (2003) reported average MPH of 171.0% and HPH of 132.0% and heterosis
increased as the severity of drought stress increased. In a study by Dhliwayo et al. (2009)
average MPH of 5.14% was reported. George et al. (2011) reported MPH values of 114.0%
and 130.0% and HPH values of 84.0% and 92.0% under high and low phosphorus soils,
respectively. Maize hybrids normally yield two to three times better than their inbred parents,
but ideal hybrids from the farmer’s perspective are not necessarily the ones with high
heterosis under optimum conditions but the hybrids with higher yield advantage under
drought conditions (Duvick, 1997). Better performing hybrids in terms of yield owe their
yield advantage not only to heterosis but also to other heritable elements that are not
necessarily affected by heterosis.

2.4.2 Heterotic groups

Identification of heterotic groups and heterotic patterns is an important undertaking in hybrid
maize breeding and contributes to substantial improvements in yield performance.
Establishment of heterotic groups have assisted in improving breeding progress because
hybrid performance can be determined without production of crosses or carrying out field
experiments (Tracy and Chandler, 2006; Troyer, 2006). New inbred lines are established
through selection of germplasm within the same heterotic group and the inbred lines are then
evaluated by crossing them to an inbred line belonging to a different heterotic group. Exotic
germplasm can be effectively exploited when the heterotic relationships among populations
are understood. Heterotic patterns used in main maize production areas worldwide have been
reviewed by several authors (Wellhausen, 1978; Ron Parra and Hallauer, 1997). Heterosis

24
can be exploited in maize breeding through classification of inbred lines into different
heterotic groups thereby leading to improved hybrid performance (Bhatnagar et al., 2004).

Heterotic groups were not identified until extensive yield test data of different combinations
of inbred lines in double crosses became available (Hallauer, 1997). Scientists from
CIMMYT have conducted several combining ability studies to enable them to identify
heterotic patterns among several maize populations and gene pools (Beck et al., 1990; 1991;
Crossa et al., 1990; Vasal et al., 1992). Initially groups were identified by how lines
performed in crosses i.e A x B crosses were superior to either A x A or B x B crosses where
A and B represent different germplasm sources (Hallauer, 1997). Lines in genetically
different heterotic groups are usually identified by positive SCA effects between them (Vasal
et al., 1992). Inbred lines in the same heterotic group have a tendency to exhibit negative
SCA effects when crossed (Vasal et al., 1992). In eastern and southern Africa, the heterotic
groups are based on Southern Cross (SC), Salisbury White (N3), K64r/M162W and Natal
Potchefstroom Pear Elite Selection (NPPES) varieties. The varieties SC, N3 and NPPES
were developed from varieties imported from the USA, while K64r is a direct import from
the USA (Mickelson et al., 2001). CIMMYT has developed a number of heterotic groups
from some of the above broad groups to suit its lowland tropical, sub-tropical and highland
breeding programmes. In its programmes of southern and eastern Africa, there are two
heterotic groups, A and B. Group A includes the following germplasm: Tuxpeno, Reid
Yellow Dent and N3, whilst group B has ETO, Lancaster Sure Crop and SC germplasm
(Mickelson et al., 2001). Heterotic groups used in the National Breeding Programme in
Zimbabwe are N3 and SC and these are equivalent to CIMMYT groups A and B,
respectively. Lines in group A have been crossed with lines in group B and have shown high
levels of heterosis. Grouping lines into heterotic groups has resulted in production of hybrids
with high hybrid vigour and as a result high yielding varieties have been released into the
market. It is critical to understand genetic diversity that exists among inbred lines as it is the
basis for forming heterotic groups and helps in optimising the levels of heterosis during
hybrid formation. In cases where defined heterotic groups do not exist, marker-based genetic
distances can also be used to avoid making crosses between closely related lines. In the
current study maize germplasm from two sources, namely the National Programme and

25
CIMMYT, was used, so it is important to determine the level of diversity through genetic
characterisation of these lines.

2.4.3 Genetic diversity and characterisation

Genetic diversity is enhanced by germplasm collections from diverse sources and these
should be well characterised for improved management and effective utilisation.
Characterisation of elite material has received special attention since the late 1980s because
of its strategic interest for breeders’ rights protection and also because analyses of
homogeneous material (inbreds) are easier than using unfixed material (Charcosset and
Moreau, 2004). In order to optimise efficiency of hybrid combinations and to develop new
inbred lines, existing maize inbred lines from various sources must be properly characterised.
Germplasm organisation and variety protection in hybrid breeding programmes is enhanced
through the knowledge of genetic relationships among inbred lines (Melchinger et al., 1991;
Bernardo, 2002). A better understanding of genetic relationships amongst genotypes is
valuable in crop breeding programmes because it allows the organisation of germplasm, such
as elite lines, and affords well-organised parental selection.

Newly developed inbred lines are usually separated into different heterotic pools using
pedigree data (Messmer et al., 1993). Nonetheless, pedigree information tracing back to more
than two generations is difficult to find, therefore maize breeders now use genetic distance
(GD) evaluation as an alternative method for germplasm selection. Morphological,
biochemical and molecular analyses as well as heterosis and SCA revealed in different
crosses in maize have been used to quantify GD among germplasm (Vasal et al., 1992;
Ajmone-Marsan et al., 1998; Paterniani et al., 2000; Menkir et al., 2004; Laborda et al.,
2005). DNA marker-based diversity has given mixed results. Previous studies have
confirmed that a high similarity for molecular markers was always associated with a high co-
ancestry (Melchinger, 1999). Once possible parents have been identified based on an
amalgamation of marker data and a trait of interest, the next step is to decide which crosses
should be made to develop new breeding populations or hybrids (Charcosset and Moreau,
2004). Molecular markers are used to investigate the level of genetic diversity amongst maize
inbred lines and this is mainly done through fingerprinting. Since homozygous lines were

26
used in this study, fingerprinting is most appropriate in assessing genetic diversity amongst
them. The fingerprinting data can then be used in estimating genetic distances among inbred
lines.

2.4.4 Molecular markers

A molecular marker is defined as a DNA sequence that is readily detected and whose
inheritance can easily be examined. A marker must be polymorphic and occur in diverse
forms such that a chromosome carrying the mutant gene can be distinguished from the
chromosome with the normal gene by a marker it carries. Molecular markers can be used to
better document the organisation of genetic diversity between possible parental materials of
new breeding programmes and to establish the distinctiveness of new cultivars prior to their
registration (Charcosset and Gallais, 2002). Plant breeders can by-pass traditional phenotype-
based selection methods that include raising plants until they reach maturity and monitoring
their physical make-up in order to infer underlying genetic make-up through use of molecular
markers (Varshney et al., 2004).

Molecular markers evolved from hybridisation based markers namely restriction fragment
length polymorphism (RFLP) (Botstein et al., 1980). With the advent of the polymerase
chain reaction (PCR) a number of markers were developed namely random amplified
polymorphic DNA (RAPD) (Williams et al., 1990), amplified fragment length polymorphism
(AFLP), simple sequence repeat (SSR) and SNP. However, each of these markers has their
advantages and disadvantages. RFLP markers were successfully used in constructing linkage
maps for different crops including maize and wheat (Hoisington, 2001). The limitations that
RFLP markers had which included need for a suitable probe library, automation not being
possible and technically demanding among others, prompted the development of RAPD
markers. RAPD markers quickly gained popularity over RFLP markers because of their
simplicity and reduced assay costs. They were also found to require small amounts of DNA
compared to RFLPs and many possible primers were available. However, due to their
dominant nature and not being reproducible RAPDs became less popular with reseachers.
AFLPs were then developed and were found to combine specificity of restriction analysis
with PCR amplification (Vos et al., 1995). AFLPs have a higher efficiency in detecting

27
polymorphism compared to RAPDs and RFLPs (Garcia-Mas et al., 2000). However, the
AFLP technique was later found to be more labour intensive and time consuming compared
to RAPDs. SSR markers were then applied to overcome the disadvantages of RFLPs, RAPDs
and AFLPs. The SSR markers became popular mainly because of their co-dominant nature,
large number of polymorphisms, their random distribution in the entire genome and being
reproducible (Senior et al., 1993; Vos et al., 1995). It was also found that once SSR primers
were developed, screening became less expensive. However, when primers were not
available the organism had to be screened for SSRs first. The process of screening for SSRs
was found to be expensive and practically complex and at times only a small number of SSRs
could be detected. SNP markers were later developed and because of their abundance in
different genomes and high levels of automation they have become the most popular
markers.

Molecular markers have been widely used in assessing genetic diversity in maize
(Melchinger et al., 1992; Betran et al., 2003; Jones et al., 2007; Dhliwayo et al., 2009; Van
Inghelandt et al., 2010; Devi and Singh, 2011; George et al., 2011). Melchinger et al. (1992)
used RFLP markers to assess genetic diversity among six flint and six dent maize inbred
lines and narrow GDs were reported between flint x dent crosses (0.56-0.73) compared to
flint x flint (0.14-0.66) and dent x dent (0.23-0.62). RFLP markers were again used to assess
genetic diversity amongst tropical maize inbred lines and the GDs ranged from 0.20-0.84
(Betran et al., 2003). RAPD markers have been used to examine GD across varied species
comprising segregating lines of maize (Ajmone-Marson et al., 1993) and have also been used
to assess genetic diversity among homozygous maize inbred lines (Lanza et al., 1997; Devi
and Singh, 2011). The most widely used DNA marker for germplasm characterisation until
now was SSR markers. SSR markers are easy to use, relatively cheap when primers are
available and have a high degree of polymorphism provided by the large number of alleles
per locus (Vignal et al., 2002). PCR based SSR markers have been widely used in
fingerprinting of maize germplasm (Messmer et al., 1992; Dubreuil et al., 1996; Smith et al.,
1997; Senior et al., 1998; Dhliwayo et al., 2009; George et al., 2011). A total of 89 SSR
markers were found to perform better at clustering maize germplasm into populations than
did a set of 847 SNPs or 554 SNP haplotypes (Hamblin et al., 2007).

28
2.4.5 Choosing a marker
Given the innumerable DNA marker technologies available and the widespread range of
applications they can be used for, an obvious problem that has to be faced is how to select the
most ideal DNA marker for a particular analysis. A number of issues have to be considered
and these include the technology itself, the problem under investigation and the
circumstances of the investigator. Widely used markers have been categorised into different
groups and these include hybridisation based markers such as RFLPs (Helentjaris et al.,
1986) and PCR-based markers such as SSRs (Senior et al., 1993) and SNPs. A perfect
marker system has been defined as one that is highly informative, evenly dispersed across the
genome, co-dominant, accurate and have reproducible data that can be produced in a high-
throughput and economical manner (Jones et al., 2007; Yan et al., 2010). RFLP and SSR
possess the majority of these aspects but they have high developmental costs. SSR markers
have been a marker of choice for the majority of crops but there have been complications in
their use, especially challenges in accurately sizing SSR alleles due to PCR and
electrophoresis artifacts (Hatcher et al., 1993; Jones et al., 1997; Bovo et al., 1998; Fernando
et al., 2001; Heckenberger et al., 2002; Davison and Chilba, 2003).

2.4.6 Single nucleotide polymorphism (SNP)

A SNP is a single change in the sequence of a section of DNA. It may come about as a result
of a substitution of one nucleotide for another at the polymorphic site. A SNP can also be a
single base insertion or deletion variant referred to as an indel. The majority of SNPs are
biallelic; however they can also be tri- or tetra-allelic. Tri allelic SNPs involve the presence
of three different nucleotides for the specific SNP whilst tetra-allelic involve all four
nucleotides. Molecular markers based on SNPs are plenty, evenly dispersed throughout the
entire genome and adequate to discriminate individuals in a population. SNP markers have
been found to possess almost all attributes of an ideal marker, which include the potential for
high throughput and low cost genotyping and they have since become the marker system of
choice (Yan et al., 2010). SNP markers received much consideration because of higher
frequencies in the genome compared to SSR markers (Van Inghelandt et al., 2010). SNP
markers are usually biallelic and tend to be less informative than SSRs on a single marker
basis, nevertheless this weakness can be compensated for by their abundance and the

29
capacity to utilise SNP haplotypes (Gupta et al., 2001; Ching et al., 2002; Rafalski, 2002).
SNP markers have become the most widely used markers because they target single
nucleotide differences between genotypes, showing more polymorphism compared to other
types of markers (Jung et al., 2010).

According to Lu et al. (2009) a large number of SNP markers would be required to substitute
extremely polymorphic SSR markers in studies of diversity and relatedness. Jones et al.
(2007) found a clear advantage for SNP markers when evaluating the repeatability of
genotyping results and proportion of missing data for SSR and SNP markers. In a study by
Van Inghelandt et al. (2010) the average number of alleles per SSR locus was higher than
that for the SNP markers and this was due mainly to the fact that SNP markers are usually
biallelic (Vignal et al., 2002). The use of SNP markers in plants has been limited firstly by
high developing and implementation costs and secondly because the development of SNP
markers in plants is complex due to the presence (in polyploid plants) of homologues i.e.
non-allelic versions of genes residing on homoelogous chromosomes. There are various SNP
genotyping methods available but most of them are too costly for low- to medium-throughput
academic laboratories and breeding programmes (Comai et al., 2004; Lin et al., 2009). These
methods include the Illumina GoldenGate® genotyping and KBiosciences Allele Specific
PCR technologies. It is also worth mentioning that the use of SNP markers produces a simple
binary output that is appropriate for automatic data collection systems and therefore their use
is gaining more popularity (Rostoks et al., 2005; Varshney et al., 2007).

SNP markers, specifically for cultivar identification, have in recent years been developed for
use by the commercial sector following demand by the commercial sector (Reale et al., 2006;
Shirasawa et al., 2006; Yoon et al., 2007). Numerous SNP markers, mostly developed from
DNA sequences of known genes, are now available for use in maize. Consequently SNP
markers have become the marker of choice for various tasks in maize improvement that
include genetic diversity analysis, linkage map construction, marker trait association or
quantitative trait locus (QTL) mapping and marker-assisted selection (MAS) (Lu et al.,
2009). Ching et al. (2002) investigated the frequency of SNPs and distribution of DNA
polymorphisms at 18 maize genes using 36 maize inbred lines. Tenaillon et al. (2001)

30
reported a SNP every 104 base pairs (bp) in coding regions, whilst Ching et al. (2002)
reported a SNP every 31 bp in non-coding regions and a SNP every 124 bp in coding regions.
In another study by Rafalsaki (2002) a SNP every 48 bp in non-coding regions and every 130
bp in coding regions was reported. SNP frequency in maize has been found to be high
compared to other crops for example rice has a SNP frequency of 0.5-0.78%, wheat 0.5%
and soybean 0.36% (Vroh et al., 2006).

2.4.7 Correlation between genetic distance and heterosis

GD has been widely used to predict the level of heterosis in maize. The level of heterosis in
F1 populations is correlated with GD of the parental lines, and with more divergent parents
the level of heterosis is higher and vice versa (Prasad and Singh, 1986; Duvick, 1999). In
crops such as maize, oat, wheat and rice, GD based on molecular markers has been broadly
correlated with heterosis but with diverse outcomes (George et al., 2011). Some studies
based on RFLP and SSR markers have concluded that heterosis is significantly correlated
with GDs (Lee et al., 1989; Smith et al., 1990; Melchinger et al., 1992; George et al., 2011),
whilst a low correlation was reported by other researchers (Godshalk et al., 1990; Melchinger
et al., 1990). Boppenmeier et al. (1992), Ajmone-Marson et al. (1998), Menkir et al. (2004),
Balestre et al. (2008), Legesse et al. (2008) and Dhliwayo et al. (2009) reported no
significant correlation between GD among inbred lines and the agronomic performance of
their F1 hybrids. In general, linear correlation between DNA marker-based genetic distance
and F1 MPH seems to be higher among related and intra-group than among unrelated and
inter-heterotic group lines (Melchinger, 1999). Conclusive results on the application of
molecular based GD in predicting hybrid performance in maize have not been reported.
Betran et al. (2003) found that drought significantly depressed correlations between GD and
F1 grain yield, MPH, HPH and SCA. Performance of genotypes is also influenced by G x E
interaction so it is critical that the magnitude and pattern of G x E is examined. Genotypes
will tend to perform differently in different environments resulting in changes in genotype
rankings.

31
2.5 Genotype by environment interaction and assessment of stability
G x E interaction is the different performances of genotypes in different environments and
consists of the following types of interaction: i) crossover interaction (COI) or genotypic
rank changes across environments, the most crucial interaction in crop improvement and
production (Baker, 1988; 1990), ii) non-COI or scale changes among environments and iii) a
combination of both. G x E interactions may result from differences in manifestation and
severity of moisture stress in different environments, differences in flowering time and
mineral nutrient deficiencies and toxicities whose manifestation and rigorousness interrelate
with moisture stress (Cooper et al., 1999; Banziger and Cooper 2001). G x E interactions are
critical only if they involve significant COI (Baker, 1988; 1990). G x E interaction plays an
important role in the performance of different genotypes in various environments; therefore
the choice of genotypes and test sites determines the level of stability estimates (Robert and
Denis, 1996; Simic et al., 2003). Genotypes are generally evaluated in several environments
to select the best ones.

G x E interaction is a key concern in plant breeding mainly because it reduces progress from
selection due to the build-up effects of the three components of interaction between genotype
and environment, namely genotype x location, genotype x year and genotype x location x
year. It also poses challenges in cultivar recommendation because it is statistically difficult to
deduce the main effects (Kang and Magari, 1996). Relationships between phenotypic and
genotypic variances is reduced by G x E interaction and this leads to best performing
genotypes in one environment to perform poorly in another, compelling breeders to look at
genotypic stability. Consequently large G x E interactions hinders advancement from
selection and has important repercussions for testing and cultivar release. Once G x E
interaction is identified as being significant it is important to then assess the response of
varieties in different environments as well as assess their overall stability. This can be done
through use of various statistical models some of which include additive main effects and
multiplicative interaction (AMMI) and genotype and genotype by environment interaction
(GGE) biplot analysis.

32
2.5.1 Additive main effects and multiplicative interaction
Cultivar responses in multi-environment trials have been predicted using various statistical
models. Models such as AMMI, genotype site regression model (SREG) (Cornelius et al.,
1996) and the model that combines genotype, environment and attribute variables in
regression models (GEAR) (Moreno-Gonzalez and Crossa, 1998) generally provide better
estimates of a genotype’s performance in specific environments than genotype-environment
combination means. The AMMI model combines additive and multivariate methodologies
(Nurminiemi et al., 2002; Pinnschmidt and Hovmoller, 2002).

AMMI was found suitable to handle both the main effects and G x E interactions in multi-
location yield trials more effectively and efficiently than any other statistical model (Gauch,
1993). The amalgamation of analysis of variance (ANOVA) and principal component
analysis (PCA) in the AMMI model alongside with prediction assessment is an important
tool in understanding G x E interaction and obtaining better yields. AMMI is represented by
the following model:

Yij = µ + Gi + Ej + ∑n k=1 λk αik γ jk + ℮ ij

Where Yij is the yield of the ith genotype in the jth environment, µ is the grand mean, Gi and Ei
are the genotype and environment deviations from the grand mean, respectively, λk is the
eigen value of the PCA analysis axis k, αik and γjk are the genotype and the environmental
principal component scores for axis k, n is the number of principal components retained in
the model and ℮ij is the error term.

G x E interaction is partitioned into several orthogonal axes namely interaction principal

component analysis (IPCA) using PCA of AMMI. IPCA1 and IPCA2 are usually used in the
construction of a biplot because higher IPCA axes are subject to noise and have no predictive
value (Van Eeuwijk, 1995). Wallace et al. (1993) concluded that AMMI statistical analysis
can separate and quantify G x E interaction effects on yield and other measured traits in
different environments.

33
2.5.2 Genotype and genotype by environment interaction biplot analysis
GGE biplot methodology for graphical analysis of multi-environment trial data (MET) was
developed by Yan et al. (2000). GGE denotes genotypic main effect (G) plus G x E
interaction, and these are two main sources of variation that are relevant to cultivar
assessment. The GGE biplot displays the GGE of MET data. The biplot is constructed by
plotting the first two principal components (PC1 and PC2) and these are also referred to as
primary and secondary effects respectively. The PC1 and PC2 values are derived from
singular value decomposition (SVD) of the environment-centered data. GGE biplot analysis
is used to identify some of the least discriminating locations and representative test locations
(Fan et al., 2006). The same researchers implied that the GGE biplot methodology was a
valuable tool for categorising sites that lead to optimum cultivar performance and efficient
utilisation of limited resources available for the testing programmes.

2.6 Conclusions
As a result of climate change drought was predicted to occur frequently and Zimbabwe will
not be spurred from these droughts. There is a shift from focusing on breeding varieties for
optimum conditions to breeding stress tolerant varieties. Maize inbred lines developed in the
past through Zimbabwe’s National Breeding Programme were not bred for drought and low
N stress tolerance. Generally, literature review revealed that there is very little published
information for research conducted in Zimbabwe pertaining to germplasm tolerant to biotic
and abiotic stress factors. There is a huge gap that still exists between maize yield potential
and the actual yield in Zimbabwe, indicating that opportunities for grain yield improvement
do exist. Therefore breeding for tolerance to drought and other stresses would contribute
towards raising the national maize yield average. Yield losses due to drought are largest at
flowering stage due to poor pollen-silk synchronisation and this further aggravates poor ear
and kernel development. It is suggested that grain yield can be improved by selecting for
short anthesis-silking interval and high number of ears per plant under stress environments. A
few studies reviewed revealed that additive gene action was important in inheritance of grain
yield under drought conditions. Literature therefore reveals that gene introgression in
parental-inbred lines is bound to improve hybrid performance. Reviewed literature showed
that G x E interaction was of great concern to plant breeders as it caused distortions in

34
performance of varieties in multi-location trials. Large G x E interaction reduces progress
from selection and has important repercussions for testing and cultivar release. This suggests
that there is need to evaluate hybrids in multi-location trials and identify the high yielding
and stable ones before recommending them for release. Genetic diversity in a breeding
programme can be enhanced by germplasm collections from divergent sources. However in
order to maximise utilisation of new introductions there is need to characterise germplasm to
determine genetic relationships among inbred lines. Heterosis is critical in a hybrid breeding
programme and the amount of heterosis between lines has assisted in determining the level of
genetic diversity. This suggests that heterosis can be enhanced by determining heterotic
groupings such that crosses for hybrid development are made amongst lines in divergent
groups.

2.7 References
Ahuja, S.L. and L.S. Dhayal. 2007. Combining ability estimates for yield and fibre quality
traits in 4x13 line x tester crosses of Gossypium hirsutum. Euphytica 153: 87-98.
Ajmone-Marsan, P., G. Egidy, G. Monfredini, S. di Silvestro and M. Motto. 1993. RAPD
markers in maize genetic analysis. Maydica 38: 259-264.
Ajmone-Marsan, P., P. Castiglion, F. Fusari, M. Kuiper and M. Motto. 1998. Genetic
diversity and its relationship to hybrid performance in maize as revealed by RFLP
and AFLP markers. Theoretical and Applied Genetics 96: 219-243.
Baker, R.J. 1988. Analysis of genotype-environmental interactions in crops. IST Atlas of
Science. Animal and Plant Sciences 1: 1-4.
Baker, R.J. 1990. Crossover genotype-environmental interaction in spring wheat. In: Kang,
M.S. (ed) Genotype x Environment Interaction and Plant Breeding. Lousiana State
University, Baton Rogue LA. pp 42-51.
Balestre, M., R.G. Von Pinho, J.C. Souza and J.L. Lima. 2008. Comparison of maize
similarity and dissimilarity genetic coefficients based on microsatellite markers.
Genetics and Molecular Research 7: 695-705.
Banziger, M. and H.R. Lafitte. 1997. Efficiency of secondary traits for improving maize for
low nitrogen target environments. Crop Science 37: 1110-1117.

35
Banziger, M. and J.L. Araus. 2007. In: Jenks, M.A. (ed) Advances in Molecular Breeding:
Toward Drought and Salt Tolerant Crops. Springer. pp 587-601.
Banziger, M. and M.E. Cooper. 2001. Breeding for low-input conditions and consequences
for participatory plant breeding-examples from tropical maize and wheat. Euphytica
122: 503-519.
Banziger, M., G.O. Edmeades, D. Beck and M. Bellon. 2000. Breeding for Drought
Tolerance in Maize. CIMMYT Special Publication. Mexico. D.F: CIMMYT. p 68.
Bassetti, P. and M.E. Westgate. 1993. Emergence, elongation and senescence of maize silks.
Crop Science 33: 271-275.
Beck, D.L., S.K. Vasal and J. Crossa. 1990. Heterosis and combining ability of CIMMYT’s
tropical early and intermediate maturity maize (Zea mays L.) germplasm. Maydica
35: 279-285.
Beck, D.L., S.K. Vasal and J. Crossa. 1991. Heterosis and combining ability among
subtropical and temperate intermediate maturity maize germplasm. Crop Science 31:
68-73.
Beebe, S., I.M. Rao, C. Cajio and M. Grajales. 2008. Selection for drought resistance in
common bean also improves yield in phosphorus limited and favorable environments.
Crop Science 48: 582-592.
Bernardo, R. 2002. Breeding for Quantitative Traits in Plants. Stemma Press. Woodbury.
Betran, F.J., D. Beck, M. Banziger and G.O. Edmeades. 2003. Genetic analysis of inbred and
hybrid grain yield under stress and non-stress environments in tropical maize. Crop
Science 43: 807-817.
Bhatnagar, S., F.J. Betran and L.W. Rooney. 2004. Combining abilities of quality protein
maize inbreds. Crop Science 44: 1997-2005.
Bolanos, J. and G.O. Edmeades. 1993a. Eight cycles of selection for drought tolerance in
lowland tropical maize. I. Responses in grain yield, biomass and radiation utilization.
Field Crops Research 31: 233-252.
Bolanos, J. and G.O. Edmeades. 1993b. Eight cycles of selection for drought tolerance in
tropical maize. II. Responses in reproductive behavior. Field Crops Research 31:
253-268.

36
Bolanos, J. and G.O. Edmeades. 1996. The importance of anthesis-silking interval in
breeding for drought tolerance in tropical maize. Field Crops Research 48: 65-80.
Bolanos, J., G.O. Edmeades and L. Martinez. 1993. Eight cycles of selection for drought
tolerance in tropical maize. III. Responses in drought-adaptive physiological and
morphological traits. Field Crops Research 31: 269-286.
Boppenmeier, J., A.E. Melchinger, E. Brunklaus-Jung, H.H. Geiger and R.G. Herrmann.
1992. Genetic diversity for RFLPs in European maize inbreds: I. Relation to
performance of flint x dent crosses for forage traits. Crop Science 32: 895-902.
Botstein, D., R.L. White, M. Skolnick and R.W. Davis. 1980. Construction of a genetic
linkage map in man using restriction fragment length polymorphism. American
Journal of Human Genetics 32: 314-331.
Bovo, D., M. Rugge and Y.H. Shiao. 1998. Origin of spurious multiple bands in the
amplification of microsatellite sequences. Molecular Pathology 52: 50-51.
Bruce, W.B., G.O. Edmeades and T.C. Barker. 2002 Molecular and physiological approaches
to maize improvement for drought tolerance. Journal of Experimental Botany 53: 13-
25.
Campos, H., M. Cooper, J.E. Habben, G.O. Edmeades and J.R. Schussler. 2004. Improving
drought tolerance in maize: A view from industry. Field Crops Research 90: 19-34.
Chapman, S.C. and G.O. Edmeades. 1999. Selection improves drought tolerance in tropical
maize populations. II. Direct and correlated responses among secondary traits. Crop
Science 39: 1315-1324.
Charcosset, A. and A. Gallais. 2002. Application of markers in selection. In: de Vienne, D.
(ed) Molecular Markers in Plant Genetics and Biotechnology. Science Publishers pp
153-176.
Charcosset, A. and L. Moreau. 2004. Use of molecular markers for the development of new
cultivars and the evaluation of genetic diversity. Euphytica 137: 81-94.
Ching, A., K.S. Caldwell, M. Jung, M. Dolan, O.S. Smith, S. Tingey, M. Morgante and A.J.
Rafalski. 2002. SNP frequency, haplotype structure and linkage disequilibrium in
elite maize inbred lines. BioMed Central Genomics 3: 1-14.

37
Comai, L., K. Yound, B.J. Till, S.H. Reynolds, E.A. Greene, C.A. Codomo, L.C. Enss, J.E.
Johnson, C. Burtner, A.R. Odden and S. Henikoff. 2004. Efficient discovery of DNA
polymorphisms in natural populations by ecotilling. Plant Journal 37: 778-786.
Comstock, C.C. and H.F. Robinson. 1948. The component of genetic variance of populations
of biparental progenies and their use in estimating the average degree of dominance.
Biometrics 4: 254-266.
Cook, E.R., R. Seager, M.A. Cane and D.W. Stahle. 2007. North American droughts.
Reconstructios, causes and consequences. Earth Science Research 81: 93-134.
Cooper, M., D.W. Podlich and S. Fukai. 1999. Combining information from multi-
environment trials and molecular markers to select adaptive traits for yield
improvement of rice in water-limited environments. In: Ito, O., J. O’Toole and B.
Hardy (eds) Genetic Improvement of Rice for Water-limited Environments, IRRI
Makati CITY. pp 13-33.
Cooper, M., F.A. van Eeuwijk, S.C. Chapman, D.W. Podlich and C. Loffler. 2006. Genotype
by environment interactions under water-limited conditions. In: Ribaut, J.M. (ed)
Drought Adaptation in Cereals. Haworth Press Inc. pp 51-96.
Cornelius, P.L., J. Crossa and M.S. Seyedsadr. 1996. Statistical tests and estimators of
multiplicative models for genotype by environment interaction. In: Kang, M.S. and
H.G. Gauch (eds) Genotype by Environment Interaction. CRC Press, Boca Raton FL.
pp 199-234.
Crossa, J., R.W. Zobel and H.G. Gauch. 1990. Additive and multiplicative interaction
analysis of two international maize cultivar trials. Crop Science 30: 493-500.
Davison, A. and S. Chilba. 2003. Laboratory temperature variation is a previously
unrecognized source of genotyping error during capillary electrophoresis. Molecular
Ecology Notes 3: 321-323.
Devi, P. and N.K. Singh. 2011. Heterosis, molecular diversity, combining ability and their
inter-relationships in short duration maize (Zea mays L.) across the environments.
Euphytica 178: 71-81.
Dhliwayo, T., K. Pixley, A. Menkir and M. Warbuton. 2009. Combining ability, genetic
distances and heterosis among elite CIMMYT and IITA tropical maize inbred lines.
Crop Science 49: 1201-1210.

38
Diniz, L.E.C., C.F. Ruas, V.P. Carvalho, F.M. Torres, E.A. Ruas, M.O. Santos, T. Sera and
P.M. Ruas. 2005. Genetic diversity among forty coffee varieties assessed by RAPD
markers associated with restriction digestion. Brazilian Archives of Biology and
Technology 48: 511-521.
Dubreuil, P., P. Dufour, E. Krejci, M. Cusse, D. de Vienne, A. Gallais and A. Charcosset.
1996. Organisation of RFLP diversity among inbred lines of maize representing the
most significant heterotic groups Crop Science 36: 790-799.
Du Plessis, D.P. and F.J. Dijkhuis. 1967. The influence of time lag between pollen shedding
and silking on the yield of maize. South African Journal of Agricultural Science 10:
667-674.
Duvick, D.N. 1997. Genetic rates of gain in hybrid maize yields during the past 40 years.
Maydica 22: 187-197.
Duvick, D.N. 1999. Heterosis. Feeding people and protecting resources. In: Coors, J.G. and
S. Pandey (eds) The Genetics and Exploitation of Heterosis in Crops. ASSA/CSSA,
SSA, Madison, WI. pp 19-29.
Duvick, D.N. 2001. Biotechnology in the 1930s: the development of hybrid maize. Nature
Reviews of Genetics 2: 69-74.
Edmeades, G.O. and T.B. Daynard. 1979. The relationship between final grain yield and
photosynthesis at flowering in individual maize plants. Canadian Journal of Plant
Science 59: 585-601.
Edmeades, G.O., J. Bolanos, M. Banziger, J.M. Ribaut, J.W. White, M.P. Reynolds and H.R.
Lafitte. 1998. Improving crop yields under water deficits in the tropics. In: Chopra,
V.L., R.B. Singh and A. Varma (eds) The Second International Crop Science
Congress, Oxford and IBH, New Delhi. pp 437-451.
Edmeades, G.O., J. Bolanos, M. Hernandez and S. Bello. 1993. Causes for silk delay in
lowland tropical maize. Crop Science 33: 1029-1035.
Edmeades, G.O., J. Bolanos, S.C. Chapman, H.R. Lafitte and M. Banziger. 1999. Selection
improves drought tolerance in tropical maize populations: I. Gains in biomass, grain
yield and harvest index. Crop Science 39: 1306-1315.
Edmeades, G.O., S.C. Chapman, M. Banziger and H.R. Lafitte. 1994. Recent evaluations of
progress in selection for drought tolerance in tropical maize. Paper presented at the

39
 Fourth Eastern, Central and Southern African Regional Maize Conference, Harare,
Zimbabwe. 28 March-1 April 1994.
Fabrizius, M.A., R.H. Busch, K. Khan and L. Huckle. 1998. Genetic diversity and heterosis
of spring wheat crosses. Crop Science 38: 1108-1112.
Falconer, D.S. 1981. Introduction to Quantitative Genetics. 2nd edition. Longman, London.
Fan, J.B., K.L. Gunderson, M. Bibikova, J.M. Yeakley, J. Chen, E. Wickham Garcia, L.L.
Lebruska, M. Laurent, R. Shen and D. Barker. 2006. Illumina Universal Bead Arrays.
Methods in Enzymology 410: 57-73.
FAO. 2011. Food and Agriculture Organisation of the United Nations. FAO Statistical
Database 2010. http//faostat.fao.org. Accessed on line 10 June 2012.
Fernando, P., B.J. Evans, J.C. Morales and D.J. Melnick. 2001. Electrophoresis artifacts-A
previously unrecognized cause of error in microsatellite analysis. Molecular Ecology
Notes 1: 235-328.
Fu, J. and B. Huang. 2001. Involvement of antioxidants and lipid peroxidation in the
adaptation of two cool-season grasses to localized drought stress. Environmental and
Experimental Botany 45: 105-112.
Garcia-Mas, J., M. Oliver, H. Gomez-Paniagua and M.C. Devicente. 2000. Comparing
AFLP, RAPD and RFLP markers for measuring genetic diversity in melon.
Theoretical and Applied Genetics 101: 860-864.
Gauch, H.G. 1993. Matmodel Version 2.0: AMMI and related analysis for two-way data
matrices. Micro computer Power, Ithaca, New York, USA.
George, M.L.C., F. Salazar, M. Warbuton, L. Narro and F.A. Vallejo. 2011. Genetic distance
and hybrid value in tropical maize under P stress and non-stress conditions in acid
soils. Euphytica 178: 99-109.
Godshalk, E.B., M. Lee and K.R. Lamkey. 1990. Relationship of restriction length
polymorphisms to single-cross hybrid performance of maize. Theoretical and Applied
Genetics 80: 273-280.
Grant, R.F., B.S. Jackson, J.R. Kiniry and G.F. Arkin. 1989. Water deficit timing effects on
yield components in maize. Agronomy Journal 81: 61-65.

40
Gupta, P.K., J.K. Roy and M. Prasad. 2001. Single nucleotide polymorphism: A paradigm for
molecular marker technology and DNA polymorphism detection with emphasis on
their use in plants. Current Science 80: 524-535.
Hallauer, A.R. 1997. Maize breeder, geneticist and administrator. Plant Breeding Journal 14:
1-11.
Hallauer, A.R. and J.B. Miranda. 1988. Quantitative Genetics in Maize Breeding. 2nd edition
Iowa State University Press. Ames.
Hamblin, M.T., M.L. Warburton and E.S. Buckler. 2007. Empirical comparison of simple
sequence repeats and single nucleotide polymorphisms in assessing of maize diversity
and relatedness. PLoS ONE 2: 1367.
Hatcher, S.L., Q.T. Lambert, L.T. Raymond and J.R. Carslon. 1993. Heteroduplex formation:
A potential source of errors from PCR products. Prenatal Diagnosis 13: 171-177.
Heckenberger, M., M. Bohn, J.S. Zieyle, L.K. Joe, J.D. Hauser, M. Hutton and A.E.
Melchinger. 2002. Variation of DNA fingerprints among accessions within maize
inbred lines and implications for identification of essentially derived varieties. I.
Genetic and technical sources of variation in SSR data. Molecular Breeding 10: 181-
191.
Heisey, P.W.A. and G.O. Edmeades. 1999. Maize production in drought- stressed
environments: Technical options and research resource allocation Part 1 of CIMMYT
1997/1998 World Facts and Trends; Maize production in drought-stressed
environments: Technical options and research resource allocation, Mexico, D.F:
CIMMYT.
Helentjaris, T., M. Scolum and S. Wright. 1986. Construction of genetic linkage maps in
maize and tomato using restriction fragment length polymorphisms. Theoretical and
Applied Genetics 72: 761-769.
Hoisington, D. 2001. Application of biotechnology to maize improvement: past, present and
future prospects. Seventh Eastern and Southern Africa Regional Maize Conference.
CIMMYT. pp 7-11.
Jacobs, B.C. and C.J. Pearson. 1991. Potential yield of maize determined by rates of growth
and development of ears. Field Crops Research 27: 281-298.

41
Jones, C.J., K.J. Edwards, S. Castaglione, M.O. Winfield, F. Sala, C. Van de Wiel, G.
Bredemejer, B. Vosman, M. Matthes, A. Daly, R. Brettschneider, P. Bettini, M.
Buiatti, E. Maestri, A. Malcevschi, N. Marmiroli, R. Aert, G. Volckaert, J. Rueda, R.
Linacero, A. Vasquez and A. Karp. 1997. Reproducibility testing of RAPD, AFLP
and SSR markers in plants by a network of European Laboratories. Molecular
Breeding 3: 381-390.
Jones, E.S., H. Sullivan, D. Bhattramakki and J.S.C. Smith. 2007. A comparison of simple
sequence repeat and single nucleotide polymorphism marker technologies for the
genotypic analysis of maize (Zea mays L.) Theoretical and Applied Genetics 115:
361-371.
Jung, J.K., S.W. Park, W.Y. Liu and B.C. Kang. 2010. Discovery of single nucleotide
polymorphisms in Capsicum and SNP markers for cultivar identification. Euphytica
175: 91-107.
Kang, M.S. and R. Magari. 1996. New developments in selecting for phenotype stability in
crop breeding. In: Kang, M.S. and H.G. Gauch Jr (eds) Genotype x Environment
interaction. CRC Press: Boca Raton, FL pp 1-14.
Kim, J.Y., A. Mahe, J. Brangeon and J.L. Prioul. 2000. A maize vacuolar invertase, IVR2 is
induced by water stress. Organ tissue specificity and diurnal modulation of
expression. Plant Physiology 124: 71-84.
Laborda, P.R., K.M. Oliveira, A.A.F. Garda, M.E. Paterniani and A.P. Souza. 2005. Tropical
maize germplasm: what can we say about its genetic diversity in the light of
molecular markers. Theoretical and Applied Genetics 111: 1288-1299.
Lafitte, H.R. and G.O. Edmeades. 1988. An update on selection under stress: Selection
criteria. In: Gelaw, B. (ed) Towards Self-Sufficiency. Proceedings of the Second
Eastern, Central and Southern African Regional Maize Workshop. pp 309-333.
Lafitte, H.R. and G.O. Edmeades. 1994. Improvement for tolerance to low soil nitrogen in
tropical maize. I. Selection criteria. Field Crops Research 39: 1-14.
Lafitte, R., A. Blum and G. Atlin. 2003. Using secondary traits to help identify drought
tolerant genotypes. In: Fischer, K.S., R. Lafitte, S. Fukai, G. Atlin and B. Hardy (eds)
Breeding Rice for Drought-Prone Environments. The International Rice Research
Institute. Los Banos. pp 14-22.

42
Lanza, L.L.B., C.L. de Souza Jr, L.M.M. Ottoboni, M.L.C. Vieira and A.P. de Souza. 1997.
Genetic distance of inbred lines and prediction of maize single-cross performance
using RAPD markers. Theoretical and Applied Genetics 94: 1023-1030.
Lee, M., E.B. Godshalk, K.R. Lamkey and W.W. Woodman. 1989. Association of restriction
fragment length polymorphisms among maize inbreds with agronomic performance
of their crosses. Crop Science 29: 1067-1071.
Legesse, B.W., A.A. Meyberg, K.V. Pixley and A.M. Botha. 2008. Relationship between
hybrid performance and AFLP based genetic distance in highland maize inbred lines.
Euphytica 162: 313-323.
Lemcoff, J.H. and R.S. Loomis. 1986. Nitrogen influence on yield determination in maize.
Crop Science 26: 1017-1022.
Lin, C.H., J.M. Yeakley, T.K. McDaniel and R. Shen. 2009. Medium to high-throughput
SNP genotyping using VeraCode microbeads. In: Bugert, P. (ed) DNA and RNA
Profiling in Human Blood: Methods and Protocols. Humana Press. pp 129-142.
Liu, F., C.R. Jensen and M.N. Andersen. 2004. Drought stress effect on carbohydrate
concentration in soybean leaves and pods during early reproductive development: its
implication in altering pod set. Field Crops Research 86: 1-13.
Loggini, B., A. Scartazza, E. Brugnoli and F. Navari-Izzo. 1999. Antioxidative defense
system. Pigment composition and photosynthetic efficiency in two wheat cultivars
subjected to drought. Plant Physiology 119: 1091-1100.
Lu, Y., J. Yan, C.T. Guimaraes, S. Taba. Z. Hao, S. Gao, S. Chen, J. Li, S. Zhang, B.S.
Vivek, C. Magorokosho, S. Mugo, D. Makumbi, S.N. Parentoni, T. Shar, T. Rong,
J.H. Crouch and Y. Xu. 2009. Molecular characterization of global maize breeding
germplasm based on genome-wide single nucleotide polymorphisms. Theoretical and
Applied Genetics 120: 93-115.
Malosetti, M., J.M. Ribaut, M. Vargas, J.Crossa, M.P. Boer and F.A. Van Eeuwijk. 2007.
Multi-trait multi-environment QTL modeling for drought stress adaptation in maize.
In: Spiertz, J.H.J, P.C. Strulk and H.H. Van Laar (eds) Scale and Complexity in Plant
Systems Research, Gene-Plant Crop Relations, Springer. pp 25-36.

43
Marilia, C.F., T.C. Servio, O.R. Valter, V. Clibas and T.M. Svi. 2001. Combining ability for
nodulation in common bean (Phaseolus vulgaris L.) genotypes from Andean and
middle American gene pools. Euphytica 118: 265-270.
Melchinger, A.E. 1999. Genetic diversity and heterosis. In: Coors, J.G. and S. Pandey (eds)
The Genetics and Exploitation of Heterosis in Crops. ASA, CSSA, and SSSA,
Madison. pp 99-118.
Melchinger, A.E. and R.K. Gumber. 1998. Overview of heterosis and heterotic groups in
agronomic crops. In: Lamkey, K.R. and J.E. Staub (eds) Concepts and Breeding of
Heterosis in Crop Plants. CSSA Special Publication 25. CSSA, Madison. pp 29-44.
Melchinger, A.E., J. Boppenmaier, B.S. Dhillon, W.G. Pollmer and R.G. Herrman. 1992.
Genetic diversity for RFLPs in European maize inbreds. II Relation to performance of
hybrids within versus between heterotic group for forage traits. Theoretical and
Applied Genetics 84: 672-681.
Melchinger, A.E., M. Lee., K.R. Lamkey and W.L. Woodman. 1990. Genetic diversity for
restriction fragment length polymorphisms: relation to estimated genetic effects in
maize inbreds. Crop Science 30: 1033-1040.
Melchinger, A.E., M.M. Messmer, M. Lee, W.L. Woodmana and K.R. Lamkey. 1991.
Diversity and relationships among U.S. maize inbreds revealed by restriction
fragment length polymorphisms. Crop Science 31: 669-678.
Menkir, A., A. Melake-Berhan, C. The, I. Ingelbrecht and A. Adeboju. 2004. Grouping of
tropical and mid-altitude maize inbred lines on the basis of yield data and molecular
markers. Theoretical and Applied Genetics 108: 1582-1590.
Messmer, M.M., A.E. Melchinger, J. Boppenmaier, E. Brunklaus-Jung and R.G. Herrmann.
1992. Relationships among early European maize (Zea mays L.) inbreds: I Genetic
diversity among flint and dent lines revealed by RFLPs. Crop Science 32: 1301-1309.
Messmer, M.M., A.E. Melchinger, R.G. Hermann and J. Boppenmeier. 1993. Relationship
among early European maize inbreds. II. Comparison of pedigree and RFLP data.
Crop Science 33: 944-950.
Mhike, X., P. Okori, C. Magorokosho and T. Ndhlela. 2011. Validation of the use of
secondary traits and selection indices for drought tolerance in tropical maize (Zea
mays L.) African Journal of Plant Science 5: 96-102.

44
Mickelson, H.R., H. Cordova, K.V. Pixley and M.S. Bjarnason. 2001. Heterotic relationships
among nine temperate and subtropical maize populations. Crop Science 41: 1012-
1020.
Moll, R.H., W.A. Jackson and R.L. Mikkelsen. 1994. Recurrent selection for maize: Grain
yield and dry matter nitrogen accumulation and partitioning changes. Crop Science
34: 874-881.
Monakhova, O.F. and I.I. Cheryadév. 2004. Effects of cytokinin preparations on the stability
of the photosynthetic apparatus of two wheat cultivars experiencing water deficiency.
Applied Biochemistry and Microbiology 40: 659-667.
Monneveux, P., P.H. Zaidi and C. Sanchez. 2005. High density and low nitrogen affects yield
and associated traits in tropical maize. Crop Science 45: 535-545.
Moreno-Gonzalez, J. and J. Crossa. 1998. Combining genotype, environment and attribute
variables in regression models for predicting cell means of multi-environment cultivar
trials. Theoretical and Applied Genetics 96: 803-811.
Nurminiemi, M., S. Madsen, O.A. Rognli, A. Bjornstad and R. Ortiz. 2002. Analysis of the
genotype by environment interaction of spring barley tested in the Nordic region of
Europe: relationships among stability statistics for grain yield. Euphytica 127: 123-
132.
Ortiz, R., W.W. Wagoire, O. Stolen, G. Alvarado and J. Crossa. 2008. Combining ability and
heterosis under pest epidemics in a broad based global wheat-breeding population.
Plant Breeding 127: 222-227.
Paterniani, M.E.A.G.Z., E. Sawazaki, C. Dudienas, A.P. Duarte and P.B. Gallo. 2000. Diallel
crossing among maize lines with emphasis on resistance to foliar diseases. Genetics
and Molecular Biology 23: 381-385.
Pearson, C.J. and B.C. Jacobs. 1987. Yield components and nitrogen partitioning of maize in
response to nitrogen before and after anthesis. Australian Journal of Agricultural
Research 38: 1001-1009.
Pingali, P.L. and S. Pandey. 2001. Meeting world maize needs: technological opportunities
and priorities for the public sector. In: Pingali, P.L. (ed) 1999/2000 World Maize
Facts and Trends. CIMMYT, Mexico. pp 1-24.

45
Pinnschmidt, O.H. and S.M. Hovmoller. 2002. Genotype x environment interactions in the
expression of net blotch resistance in spring and winter barley varieties. Euphytica
125: 227-243.
Prasad, S.K. and T.P. Singh. 1986. Heterosis in relation to genetic divergence in maize (Zea
mays L.). Euphytica 35: 919-924.
Prine, G.M. 1971. A critical period for ear development in maize. Crop Science 11: 782-786.
Rafalski, J.A. 2002. Application of single nucleotide polymorphisms in crop genetics.
Current Opinion in Plant Biology 5: 94-100.
Reale, S., S. Doveri, A. Diaz, A. Angiolillo, L. Lucentini, F. Pilla, A. Martin, P. Donini and
D. Lee. 2006. SNP-based markers for discriminating olive (Olea europaea L.)
cultivars. Genome 49: 1193-1205.
Ribaut, J.M., M. Banziger, J. Betran, C. Jiang, G.O. Edemeades, K. Dreher and D.
Hoisington. 2002. In: Kang, M.S. (ed) Use of Molecular Markers in Plant Breeding:
Drought Tolerance Improvement in Tropical Maize. CAB International. Quantitative
Genetics, Genomics and Plant Breeding. pp 85-99.
Ribaut, J.M., Y. Fracheboud, P. Monneveux, M. Banziger, M. Vargas and C. Jiang. 2007.
Quantitative trait loci for yield and correlated traits under high and low soil nitrogen
conditions in tropical maize. Molecular Breeding 20: 15-29.
Robert, N. and J.B. Denis. 1996. Stability of baking quality in bread wheat using several
statistical parameters. Theoretical and Applied Genetics 8: 172-178.
Ron Parra, J. and A.R. Hallauer. 1997. Utilisation of exotic maize germplasm. Plant
Breeding Revolution 14: 165-187.
Rostoks, N., S. Mudie, L. Cardle, J. Russell, L. Ramsay, A. Booth, J.T. Svensson, S.I.
Wanamaker, H. Walia, E.M. Rodriguez, P.E. Hedley, H. Liu, J. Morris, T.J. Close,
D.F. Marshall and R. Waugh. 2005. Genomewide SNP discovery and linkage analysis
in barley based on genes responsive to abiotic stress. Molecular Genetics and
Genomics 274: 515-527.
Saini, H.S. and M.E. Westgate. 2000. Reproductive development in grain crops during
drought. Advances in Agronomy 68: 59-96.

46
Saneoka, H., R.E.A. Moghaieb, G.S. Premachangra and K. Fujita. 2004. Nitrogen nutrition
and water stress effects on cell membrane stability and leaf water relations in Agrostis
polustris. Experimental Botany 52: 131-138.
Schussler, J.R. and M.E. Westgate. 1991. Maize kernel set at low water potential. II.
Sensitivity to reduced assimilates at pollination. Crop Science 31: 1196-1203.
Senior, L., M. Lymm and M. Heun. 1993. Mapping maize microsatellites and polymerase
chain reaction confirmation of the targeted repeats using CT primer. Genome 36: 884-
889.
Senior, M.L., J.P. Murphy, M.M. Goodman and C.W. Stuber. 1998. Utility of SSRs for
determining genetic similarities and relationships in maize using an agarose gel
system. Crop Science 38: 1088-1098.
Shirasawa, K., S. Shiokai, M. Yamaguchi, S. Kishitani and T. Nishio. 2006. Dot Blot SNP
analysis for practical plant breeding and cultivar identification in rice. Theoretical
and Applied Genetics 113: 147-155.
Shukla, S.K. and M.P. Pandey. 2008. Combining ability and heterosis over environments for
yield and yield components in two-line hybrids involving thermosensitive genic male
sterile lines in rice (Oryza sativa L.) Plant Breeding 127: 28-32.
Simic, D., J. Gunjaca, Z. Zdunic, I. Brkie and J. Kovacevic. 2003. Biometrical
characterization of test sites for maize breeding. Poljoprivreda 9: 18-24.
Sinclair, T.R. and R.C. Muchow. 2001. System analysis of plant traits to increase grain yield
on limited water supplies. Agronomy Journal 93: 263-270.
Smith, J.S.C., E.C.L. Chin, H. Shu, O.S. Smith, S.J. Wall, M.L. Senior, S.E. Mitchell, S.
Kresovich and J. Ziegle. 1997. An evaluation of the utility of SSR loci as molecular
markers in maize (Zea mays L.): Comparison with data from RFLPs and pedigree.
Theoretical and Applied Genetics 95: 163-173.
Smith, O.S., J.S.C. Smith, S.L. Bowen, R.A. Tenborg and S.J. Wall. 1990. Similarities
among a group of elite maize inbreds as measured by pedigree, F1 grain yield,
heterosis and RFLPs. Theoretical and Applied Genetics 80: 833-840.
Sprague, G.F. and L.A. Tatum. 1942. General vs specific combining ability in single crosses
of corn. Journal of American Society of Agronomy 34: 923-932.

47
Stevens, R. 2008. Review: Prospects for using marker assisted breeding to improve maize
production in Africa. Journal of the Science of Food and Agriculture 88: 745-755.
Sun, Q., Z.S. Liang, W.L. Wang, X.J. Li and F.S. Zhang. 2001. Effects of mechanism lipid
peroxidation and photosynthesis rate on water stress under different nitrogen levels in
maize. Acta Agricultural Boreal-Occident sin 1: 7-10.
Tenaillon, M.I., M.C. Sawkins, A.D. Long, B. Gaut, J.F. Doebley and S. Brandon. 2001.
Patterns of DNA sequence polymorphism along chromosome 1 of maize (Zea mays
L.) Proceedings of the National Academy of Sciences of the United States of America.
98: 9161-9166.
Tollenaar, M., A. Ahmadzadeh and E.A. Lee. 2004. Physiological basis for grain yield
improvement in maize. Crop Science 44: 2086-2094.
Tollenaar, M., L.M. Dwyer and D.W. Stewart. 1992. Ear and kernel formation in maize
hybrids representing three decades of grain yield improvement in Ontanio. Crop
Science 32: 432-438.
Tracy, W.F. and M.A. Chandler. 2006. The historical and biological basis of the concept of
heterotic patterns in corn belt dent maize. In: Lamkey, K.R. and M. Lee (eds) Plant
Breeding: The Arnel R Hallauer International Symposium. Blackwell Publishing,
Ames, IA. pp 219-233.
Troyer, A.F. 2006. Adaptedness and heterosis in corn and mule hybrids. Crop Science 46:
528-543.
Uhart, S.A. and F.H. Andrade. 1995a. Nitrogen deficiency in maize. I. Effects on crop
growth, development, dry matter partitioning and kernel set. Crop Science 35: 1376-
1383.
Uhart, S.A. and F.H. Andrade. 1995b. Nitrogen deficiency in maize. II. Carbon-nitrogen
interaction effects on kernel number and grain yield. Crop Science 35: 1384-1389.
Van Eeuwijk, F.A. 1995. Linear and bi-linear models for the analysis of multi-environment
trials: An inventory of models. Euphytica 84: 1-7.
Van Inghelandt, D.A., E. Melchinger, C. Lebreton and B. Stich. 2010. Population structure
and genetic diversity in a commercial maize breeding program assessed with SSR and
SNP markers. Theoretical and Applied Genetics 120: 1289-1299.

48
Varshney, R.K., U. Beier, E.K. Khlestkina, R. Kota, V. Korzun, A. Graner and A. Borner.
2007. Single nucleotide polymorphisms in rye (Secale cereale L.): Discovery,
frequency and applications for genome mapping and diversity studies. Theoretical
and Applied Genetics 114: 1105-1116.
Varshney, R.K., V. Korzun and A. Borner. 2004. Molecular marker maps in cereals:
Methodology and progress. In: Gupta, P.K. and R.K. Varshney (eds) Cereal
Genomics. Kluwer Academic Publishers. The Netherlands. pp 35-82.
Vasal, S.K., G. Srinivasan, G.C. Han and F. Gonzalez. 1992. Heterotic patterns of eighty-
eight white subtropical CIMMYT maize lines. Maydica 37: 319-327.
Verbitskaya, T.G., Y.M. Sivolap, D.V. Guzhva and V.M. Sokolov. 1999. Analysis of the
relation between the genetic divergence of the DNA of corn lines and hybrid
productivity. Russian Agricultural Sciences 5: 28-33.
Vignal, A., D. Milan, M. SanCristobal and A. Eggen. 2002. A review of SNP and other types
of molecular markers and their use in animal genetics. Genetics Selection and
Evolution Journal 34: 275-305.
Vos, P., R. Hogers, M. Bleeker, M. Reijans, T. van de Lee, M. Hornes, A. Frijters, J. Pot, J.
Peleman, M. Kuiper and M. Zabeau. 1995. AFLP: A new technique for DNA
fingerprinting. Nucleic Acids Research 23: 4407-4414.
Vroh, B.I., M.D. McMullen, H. Sanchez-Villeda, S. Schroeder, J. Gardiner, M. Polacco, C.
Soderlund, R. Wing, Z. Fang and E.H. Coe. 2006. Single nucleotide polymorphisms
and insertion-deletions for genetic markers and anchoring the maize finger print
Contig physical map. Crop Science 46: 12-21.
Wallace, D.H., J.P. Baudoin, J. Beaver, D.P. Coyne, D.E. Halseth, P.N. Masaya, H.M.
Munger, J.R. Myers, M. Silbernagel, K.S. Yourstone and R.W. Zobel. 1993.
Improving efficiency of breeding for higher crop yield. Theoretical and Applied
Genetics 86: 27-40.
Wellhausen, E.J. 1978. Recent developments in maize breeding in the tropics. In: Walden,
D.B. (ed) Maize Breeding and Genetics. John Wiley & Sons Inc. New York. pp 59-
84.
Westgate, M.E. and J.S. Boyer. 1986. Reproduction at low silk and pollen water potentials in
maize. Crop Science 26: 951-956.

49
Williams, J.K.F., A.R. Kubelik, K.G. Livak, J.A. Rafalski and S.V. Tingey. 1990. DNA
polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic
Acids Research 18: 6531-6535.
World Food Programme. 2003. World Food Programme 2002 Annual Report.
http://www.wfp.org. Accessed on line 9 June 2012.
World Food Programme. 2009. The horn of Africa: A look at the current situation.
http://www.wfp.org Accessed on line 10 June 2012.
World Food Programme. 2011. Hunger needs expected to rise in horn of Africa.
http://www.wfp.org. Accessed on line 10 June 2012.
Xu, M., L. Zhu, H.X. Shou and P. Wu. 2005. A Pin1 family gene, OsPin1 involved in auxin-
dependent adventitious root emergence and tillering in rice. Plant Cell Physiology 46:
1674-1681.
Yan, J., X. Yang, T. Shah, H. Sanchez-Villeda, J. Li., M. Warbuton, Y. Zhou, J.H. Crouch
and Y. Xu. 2010. High-throughput SNP genotyping with the Golden Gate assay in
maize. Molecular Breeding 25: 441-451.
Yan, W., L.A. Hunt, Q. Sheng and Z. Szlavnics. 2000. Cultivar evaluation and mega-
environment investigation based on the GGE biplot. Crop Science 40: 597-605.
Yoon, M.S., Q.J. Song, I.Y. Choi, J.E. Specht, D.L. Hyten and P.B. Cregan. 2007. BARC
Soy SNP 23: a panel of 23 selected SNPs for soybean cultivar identification.
Theoretical and Applied Genetics 114: 885-899.
Zaman, A. and P.K. Das. 1991. Effect of irrigation and nitrogen on yield and quality of
safflower. Indian Journal of Agronomy 36: 177-179.
Zinselmeier, C., J.E. Habben, M.E. Westgate and J.S. Boyer. 2000. Carbohydrate metabolism
in setting and aborting maize ovaries. In: Westgate, M.E. and K.J. Boote (eds)
Physiology and Modeling Kernel Set in Maize. CSSA Special Publication 29.
Madison. WI CSSA. pp 1-13.

50
 CHAPTER 3
Combining ability between Zimbabwean and CIMMYT maize inbred lines
under stress and non-stress conditions

Abstract
Drought and low nitrogen (N) remain some of the major factors limiting maize production in
Zimbabwe. It is therefore crucial for the National Breeding Programme to continue assessing
the breeding values of potential stress tolerant parents for developing new and locally
adapted hybrids. In this study 10 DR&SS and 13 CIMMYT parental lines were crossed using
a NCDII mating scheme. The resultant 72 single cross hybrids together with eight local
checks were evaluated under non-stress, drought and low N conditions using a 0.1 alpha
lattice design with two replications across seven sites in the 2009/10 and 2010/11 seasons.
The objective of this study was to estimate the combining ability effects among the
CIMMYT and DR&SS elite white maize inbred lines. Significant GCA and SCA effects
(P≤0.001) for grain yield, anthesis days and anthesis silking interval across all environments
suggested the importance of both additive and non-additive gene effects respectively in the
expression of these traits. However, additive gene action assumed a more important role as
shown by higher GCA variances for most traits. Non-additive gene action was found to be
important in the expression of grain yield, anthesis silking interval, ears per plant and
senescence under stress environments. Larger general combining ability due to females
(GCAf) than general combining ability due to males (GCAm) for grain yield under drought
and for anthesis silking interval, ears per plant and senescence under both drought and low N
suggested the importance of maternal effects in the expression of these traits. Tester
identification was based on good GCA for grain yield and stability under diverse
environments. Lines RS61P, NAW5885, CML444, CML539, CML442, CML537 and
CML548 showed desirable GCA effects under both drought and low N conditions. In the SC
heterotic group the single cross RS61P/CML444 and in the N3 heterotic group
2N3d/CML548 was identified as potential testers. The study therefore identified superior
germplasm that will be put into use in germplasm improvement for stress environments.

51
3.1 Introduction
Maize accounts for roughly 15% of the daily intake of calories all over sub-Saharan Africa,
though this varies from region to region as some countries obtain up to 50% of their every
day calories from maize (Stevens, 2008). Temperatures are expected to rise and rainfall
distribution to change in key production regions as a result of global climate change and this
is anticipated to lead to significant yield losses in maize, mainly as a result of drought
(African Press Agency, 2007). Significant yearly yield losses of between 9.3-15.5% are
expected to be accountable to moisture stress (Wang et al., 2006). Maize ranks first in terms
of the number of producers, area grown and total cereal production in Zimbabwe. Drought
reduced maize production in Zimbabwe by about 70% between 1981 and 1982 (Rukuni et
al., 2006) and in 1991-1993 the country again registered the worst season for maize
production. In 2001-2003, the drought left about seven million people malnourished and the
nation imported more than two million tons of maize, hence the need to develop drought
tolerant varieties (Rukuni et al., 2006).

Hybrid maize breeding has become the major driving force behind the success of seed
systems worldwide, mainly due to good adaptation and superior yield performance of
hybrids. As a result of hybrid breeding sufficient infrastructure has been developed mainly
for seed production and seed supply chain linkages have been improved (Devi and Singh,
2011). It is essential for any breeding programme to evaluate the breeding value of
prospective parental lines to be used for developing new, locally or extensively adapted
varieties. NCDII crosses are used for assessing gene action in breeding populations and this
is done by estimating GCA and SCA variances. GCA represents additive gene action and it is
the average performance of a line in different hybrids, whilst SCA represents non-additive
gene action and it is measured as the deviation of hybrid performance from parental
performance. GCA and SCA are powerful tools used by breeders in selecting best parents for
further crosses. Studies on combining ability help breeders in identifying parental lines with
good GCA and in detecting hybrids with good SCA. The best performing genotypes ought to
show steady performance across environments, as evaluated through multi-environment trials
(Devi and Singh, 2011).

52
Heritability estimates allow breeders to develop more efficient selection strategies and to
predict gain from selection (Allard, 1999). Variance components and heritability estimates
have been extensively used by plant breeders in selection of promising genotypes and in
prediction of percentage heritability of desirable traits (Morakinyo, 1996). While it is useful
to have an estimate of the total genetic effects on a particular trait, such as broad-sense
heritability, narrow-sense heritability provides a better estimate of the breeding value (Allard,
1999). It is important to realise that heritability is not only influenced by the trait under
consideration but is also influenced by the population and environmental conditions which
individuals are exposed to as well as the method of data collection used (Falconer and
Mackay, 1996). It is advisable when selecting for grain yield of hybrids to do indirect
selection for other yield related traits that show close correlation with yield and exhibit high
heritability because yield is considered a polygenic trait.

Germplasm improvement for drought tolerance will remain a high priority for the Zimbabwe
National Breeding Programme mainly because most of the maize in the country is produced
under rain-fed conditions and is a major enterprise on small scale farms, where drought is
considered to be the chief abiotic limitation to production. The Zimbabwe National Breeding
Programme is one of the partners that have worked closely with CIMMYT in producing
drought tolerant maize varieties. However, the predominant maize inbred lines used by the
National Breeding Programme in the most successful and current commercial hybrids and
their derivatives were developed in the last century and as such they were not screened for
drought tolerance. To this effect the National Breeding Programme has acquired a number of
inbred lines from CIMMYT for use in improvement of existing inbred lines, development of
new inbred lines as well as constituting of new hybrids. However, little is known about the
heterotic relationships between CIMMYT drought tolerant maize donors and Zimbabwe
National Breeding Programme maize lines. In order to determine the best parents to use in
the national breeding programme, it is important to understand the heterotic relationships
between the CIMMYT and National Breeding Programme lines with a view of selecting
good parents to initiate crosses for pedigree, backcross and potential marker assisted
recurrent selection (MARS) populations for line extraction. In addition, the classification of
the inbred lines into heterotic groups will facilitate exploitation of heterosis which can

53
contribute to hybrid performance. The objectives of this study were to estimate combining
ability and heterotic patterns among 23 elite CIMMYT and DR&SS inbred lines for grain
yield and other agronomic traits under optimal, low N and drought conditions.

3.2 Materials and methods

3.2.1 Germplasm
Ten DR&SS elite inbred lines susceptible to drought and diseases were crossed to 13
CIMMYT elite drought tolerant and disease resistant lines (Table 3.1) using a NCDII mating
scheme. DR&SS lines were used as females and CIMMYT lines were used as males.
Seventy-two single cross hybrids were successfully produced out of a potential 130 hybrids
and the hybrids are presented in Appendix 1. The crosses were evaluated at six sites in the
2009/10 summer season and one managed drought site in the winter of 2010. The same
evaluations were undertaken in the 2010/11 summer season and the winter of 2011.

Table 3.1 Maize germplasm used to produce the single cross hybrids
DR&SS
Heterotic group CIMMYT lines Heterotic group
lines
N3.2.3.3. N3 CML312-B A
SC5522 SC CML395-B B
2Kba SC CML442-B A
K64r N3 CML444-B-B B
NAW5885 N3 CML536 A
SV1P SC CML537 A
WCOBY1P SC CML538 A
2N3d N3 CML539 A
RS61P SC CML548 A
RA214P N3 CML545 A
CML544 B
CZL052 B
CZL03007 B

3.2.2 Testing environments

Sites included one low N site, one mid-altitude site, four non-drought and high N sites and
one managed drought site. The sites were the Agricultural Research Trust Farm (ART Farm)

54
(17°26'S, 31.5°E, 1 480 masl), Rattray Arnold Research Station (RARS) (17°40'S, 1 308
masl), Harare Research Station (HRS) (17.13°S, 31°E, 1 406 masl), Kadoma Research
Institute (KRI) (18.32°S, 30.90°E, 1 155 masl), Chiredzi Research Station (CRS) (21.02°S,
31.58°E, 433 masl), and Chisumbanje Research Station (20°S, 33°E, 455 masl). The
managed drought trial was planted at CRS during the winter. Trials were conducted in each
site in the 2009/10 and 2010/11 summer seasons and 2010 and 2011 winter seasons at CRS
only. ART Farm, RARS, CRS and Chisumbanje were non-stress sites, whilst Kadoma was
non-stress within the mid-altitude and HRS was the low N site.

3.2.3 Management
General maize cultural practices were applied at all sites. Weeds were mostly controlled
using herbicides at all sites, but hand weeding was also done when necessary. Managed
drought trials were done in winter under irrigation, which was terminated two weeks before
flowering to target stress during flowering. The total amount of rainfall received in all the
sites in the 2009/10 and 2010/11 and the amounts of irrigation applied to managed drought
trials are presented in Table 3.2. The level of stress applied was projected to achieve 15-20%
(1-2 t ha-1) of yields achieved under well watered conditions. This stress level delays silking
and causes ear abortion in non-stress tolerant genotypes (Banziger et al., 2000). Such stress
levels achieve an anthesis silking interval of between 4-8 days and 0.3-0.7 ears per plant
(Banziger et al., 2000). The low N site used had already been depleted of N and this was
achieved through growing summer maize and irrigated winter wheat continuously for six
years. According to the soil analysis results the soil had the capacity to supply N since it
contained 7 ppm in the top 30 cm of the soil and 7 ppm in the soil depth 30-60 cm. In terms
of kg ha-1 this translates to 54 kg ha-1. The 7 ppm was therefore considered as low N.
Optimum sites constituted of the crop grown during the rainy season under rain-fed
conditions with different N rates being applied to the crop in different sites. Maize fert (N-8,
P-16, K-8) was applied as basal dressing with ART farm, Rattray Arnold, Chiredzi and
Chisumbanje receiving 400 kg ha-1 and Kadoma 350 kg ha-1. Different rates of ammonium
nitrate (AN) were also applied with ART farm and Rattray Arnold receiving two split
applications of 200 kg ha-1, Chiredzi and Chisumbanje two split applications of 120 kg ha-1
and Kadoma two split applications of 100 kg ha-1.

55
Table 3.2 Amount of rainfall received and irrigation applied in the 2009/10 and 2010/11
seasons

Site Amount of rainfall(mm) Irrigation applied (mm)

2009/10 2010/11 2009/10 2010/11
ART farm 714 529
RARS 1074 907.5
Harare 563 533.5
Kadoma 593 500.5
Chiredzi 332.3 341.3 220 220
Chisumbanje 335.2 466.5
ART farm=Agricultural Research Trust farm; RARS=Rattray Arnold Research Station

3.2.4 Experimental design and data collection

Each site consisted of one hybrid and one parental trial and these were planted side by side.
The hybrid trial consisted of 72 hybrids and eight hybrid checks whilst the line trial consisted
of 23 lines and two inbred checks. Trials were planted in six sites during summer and one site
during winter. Trials were planted using a 0.1 alpha lattice design with two replications for
both the hybrid and line trials. The hybrid trials were one row each whilst the line trials
constituted of two rows. The rows were 4 m long with an inter-row spacing of 0.75 m and in-
row spacing of 0.25 m. A maximum of four seeds were planted per planting station and
plants were thinned to one plant per station at three weeks after emergence to achieve the
targeted plant population of 53 000 plants ha-1. Different traits were measured and derived as
described in Table 3.3. The data were collected following the standard practise at CIMMYT.

3.2.5 Statistical analysis

ANOVA for each environment and combined ANOVA were computed using the PROC
MIXED procedure of SAS (SAS Institute, 2002). Genotypes were considered fixed effects
whilst replications and incomplete blocks were considered random effects. The SAS
programme for the line x tester analysis was used to compute the GCA and SCA effects
following the procedure presented by Singh and Chaudhary (1977). The following line x
tester model was used:

56
Yijk = µ + gi + gj + sij + rk + eijk
Where:
Yijk = mean value of a character measured on cross i x j in kth replication
gi = GCA effect of ith parent
gj = GCA effect of the parent j
sij = SCA effect of cross i x j
rk = replication effect
eijk = environmental effect peculiar to (ijk)th individual
µ = population mean effect

Estimation of GCA effects:

Lines: gi = xi…/tr – y…/lrt
Testers: gt = x.j./lr – x…/ltr

Estimation of SCA effects:

sij = xij./r – xi…/tr – x.j./lr – x…/ltr
Where:
l = number of lines
t = number of testers
r = number of replications

Estimation of standard errors:

S.E. (GCA for line) = (Me/r x t)½
S.E. (GCA for tester) = (Me/r x l)½
S.E. (SCA effects) = (Me/r)½
S.E. (gi – gj) line = (2Me/r x t)½
S.E. (gi – gj) tester = (2Me/r x l)½
S.E. (sij – skl) = (2Me/r)½

For each cross combination (P1 x P2) MPH was calculated as the difference between the F1
hybrid mean and the average of its parents (Falconer and Mackay, 1996) as follows:

57
MPH = [(F1-MP)/MP] x 100
Where F1 is the mean of the F1 hybrid performance and MP = (P1 + P2)/2 in which P1 and P2
are the means of the inbred parents, respectively.

Heritability of all the traits was calculated using the narrow sense formula according to
Hallauer and Miranda (1988).
h2 = gca/ gca + sca + error
Where: h2 = narrow sense heritability
gca = General combining ability
sca = Specific combining ability

3.3 Results

3.3.1 Analysis of variance and hybrid mean performance across all environments
The means for different traits presented in tables in this chapter are for best 10 and poorest 10
hybrids in terms of grain yield and the complete data for all the entries is found in
Appendices 2-5. The data for GCA effects for lines and testers presented in figures is also
presented in Appendices 6 and 7. ANOVA and means for grain yield and other agronomic
traits across the 14 sites in the 2009/10 and 2010/11 seasons are presented in Tables 3.4 and
3.5. Sites were highly significantly different for all traits (Table 3.4, P≤0.001). Entries were
highly significantly different for grain yield, anthesis date, anthesis silking interval, plant
height and ear height (P≤0.001) and significantly different for ears per plant (P≤0.05). The
general combining ability attributable to females (GCAf) mean squares were highly
significant (P≤0.001) for grain yield, anthesis days, anthesis silking interval, plant height and
ear height, whilst ears per plant were only significant at P≤0.05. GCAf*site was highly
significant for grain yield, anthesis silking interval, plant height and ear height. The GCAm
was highly significant (P≤0.001) for grain yield, anthesis days, anthesis silking interval, plant
height and ear height and was not significant for ears per plant.

58
Table 3.3 Agronomic traits that were measured and derived
Trait Procedure
Anthesis days (AD) /silking date (SD) Taken as number of days after planting to
when 50% of plants started shedding pollen
or had extruded silks of at least 5cm
Anthesis silking interval (ASI) Derived from anthesis date and silking date
as follows: ASI=SD-AD
Ears per plant (EPP) Calculated as a ratio of the number of ears
with at least one fully developed grain
divided by the number of harvested plants
Plant height (PH) Measured as the height between the base of
a plant and the insertion of the first tassel
branch
Ear height (EH) Measured as the height between the base of
a plant to the insertion of the top ear
Ear position (EPO) Calculated as EH divided by PH
Root lodging (RL) Measured as a percentage of plants that
showed lodging by being inclined 45°
Stem lodging (SL) Measured as a percentage of plants that
were broken below the ear
Leaf senescence (SEN) Measured on a score of 1-10 with 1 having
no signs of senescence and 10 100%
senescence
Ear rot (ER) Measured as the number of ears affected
then converted to percentage
Ear aspect (EA) Measured as the appearance of the ear with
1 being excellent and 5 being very poor
Grain texture (TEX) Measured on a scale of 1-5 with 1 being
flint and 5 being dent
Grain yield (GYD) Calculated from shelled grain weight per
plot adjusted to 12.5% grain moisture
Foliar diseases Measured on a scale 0f 1-5 where 1 is
completely free from disease and 5 severely
affected

59
The GCAf and GCAm mean squares were greater than SCA mean squares for all traits. SCA
was significant for grain yield, anthesis silking interval and plant height (P≤0.001) and
significant for anthesis days (P≤0.05). SCA*site was highly significant (P≤0.001) for
anthesis days, plant height (P≤0.01) and anthesis silking interval (P≤0.05). The mean values
for grain yield, anthesis days, anthesis silking interval, ears per plant, plant height and ear
height across 14 sites were 4.07 t ha-1, 70.6 days, 1.4 days, 0.84 ears, 238.5 cm and 122.4 cm
respectively (Table 3.6).

Senescence is another important trait especially under drought and low N conditions. Entries
were significantly different for this trait (P≤0.001) as well as for grey leaf spot, common rust
and leaf blight turcicum (Table 3.5). The sites were significantly different for senescence, ear
rot, grey leaf spot, common rust and leaf blight turcicum. The GCAf mean squares were
larger than GCAm mean squares for senescence, grey leaf spot and leaf blight turcicum and
on the other hand, the overall GCA mean squares were larger than the SCA mean squares. A
check variety, entry 74 (SC727), out-yielded all experimental hybrids as well as the check
varieties and had a mean yield of 5.62 t ha-1 (Table 3.6). The second best performer was entry
61 (RS61P/ CML548) with a mean yield of 5.31 t ha-1 and it performed above the mean. The
poorest performing hybrid was entry 17 (2Kba/CML444) with a mean yield of 2.29 t ha-1.
Generally the hybrids had good disease scores for GLS, RUST and ET except for entry 48
(2N3d/CML444), which had a GLS score of 3.9 (Table 3.6).

60
 Table 3.4 Combined analysis of variance of 14 sites in the 2009/10 and 2010/11 seasons for grain yield and other
agronomic traits
Source DF GYD AD ASI EPP PH EH EPO

Site 13 547.24*** 20561.34*** 558.42*** 2.12*** 207100.97*** 86941.07*** 0.617***

Entry 69 9.70*** 192.96*** 16.61*** 0.15* 4165.09*** 2417.32*** 0.012***
GCAf 7 48.42*** 887.08*** 62.34*** 0.23* 16331.75*** 8475.14*** 0.038***
GCAm 11 18.93*** 646.52*** 25.18*** 0.17 12050.78*** 9060.36*** 0.029***
GCAf*site 79 2.90*** 0.00 5.30*** 0.15* 706.29*** 430.18*** 0.006
GCAm*site 119 0.29 0.00 7.16*** 0.13* 661.57*** 344.137*** 0.006**
Site*entry 765 1.35* 0.65 4.25*** 0.12 448.98*** 246.32** 0.004
SCA 52 2.69*** 5.87** 8.46*** 0.13 785.61*** 185.72 0.005
SCA* site 566 1.33 5.52*** 3.48* 0.11 368.55** 197.25 0.004
Error 843 1.34 3.24 2.92 0.11 301.78 192.6 0.004
Heritability (narrow sense) % 60 73 40 66 73 93 74
***P≤0.001; **P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear height; EPO=ear position;
DF=degrees of freedom; GCAf=general combining ability attributable to females; GCAm=general combining ability attributable to males; SCA=specific combining ability.

61
Table 3.5 Combined analysis of variance across 14 sites for senescence and diseases in
the 2009/10 and 2010/11 seasons
Source DF SEN DF ER DF GLS RUST ET
Site 4 1207.22*** 11 5552.69*** 2 70.71*** 1.57*** 4.20***
Entry 69 1.26*** 69 105.95** 69 1.71*** 0.18*** 0.37***
Site*entry 254 0.12 627 80.71* 94 0.68*** 0.10 0.24*
GCAf 7 5.32*** 7 201.76** 7 7.59*** 0.28** 0.86***
GCAm 11 3.21*** 11 108.16 11 3.28*** 0.51*** 0.85***
GCAf*site 26 0.43 65 123.85** 10 0.14 0.09 0.52**
GCAm*site 40 0.00 97 89.94* 14 1.86*** 0.28*** 0.29*
SCA 52 0.65*** 52 102.12* 52 0.57** 0.10 0.19
Site*line*tester 187 0.08 464 71.61 69 0.52** 0.06 0.19
Error 328 0.30 704 69.21 166 0.33 0.10 0.18
Heritability (narrow
59 49 43 67 92
sense) %
***P≤0.001; **P≤0.01; *P≤0.05; SEN=senescence; ER=ear rot; GLS=grey leaf spot; RUST=common rust; ET=leaf blight turcicum;
DF=degrees of freedom; GCAf=general combining ability attributable to females; GCAm=general combining ability attributable to
males; SCA=specific combining ability.

Combined ANOVA across six optimum sites and the hybrid mean performances are
presented in Table 3.7. Entry 74 (SC727) was the best performer (7.86 t ha-1) across optimum
conditions followed by entry 68 (RA214P/CML538) (Table 3.7). Entry 61 (RS61P/CML548)
was still among the top 10 best performing hybrids and had an anthesis silking interval of -
0.1 together with entry 23 CML545/2Kba-B) and entry 30 (K64r/CML442). Entries 57
(RS61P/CML444), 48 (2N3d/CML444) and 5 (N3.2.3.3/CML444) had the best ears per plant
values of 1.00, 1.08 and 1.01 respectively. Entry 57 was the third best hybrid across all sites
and maintained the same position under optimum conditions. Entries 74, 61, 57, 48, 52, 79
and 68 were amongst the best ten hybrids across all sites and under optimum conditions they
were also amongst the best 10 hybrids, although rankings varied.

The mean performance for grain yield and other agronomic traits under managed drought
conditions are presented in Table 3.8. The data presented is for the best 10 hybrids and the
poorest 10 hybrids. The best ten hybrids performed well above the mean (2.09 t ha-1). Entry
52 (CML548/2N3d) was the best performing hybrid under managed drought over two
seasons with a mean yield of 3.26 t ha-1.

62
Table 3.6 Performance of hybrids for grain yield and other agronomic traits across 14 sites in the 2009/10 and 2010/11
seasons
Entry GYD AD ASI PH EH RL SL EPP ER GLS RUST ET SEN
t ha-1 d d cm cm % % # % 1-5 1-5 1-5 1-10
Best 10 hybrids 74 5.62 75.3 1.4 266.7 138.6 5.7 5.0 0.80 8.4 1.5 1.3 2.2 1.9
61 5.31 69.7 0.6 230.7 119.5 11.0 3.7 0.92 5.0 2.7 1.0 1.5 2.1
57 4.94 72.4 1.5 240.5 133.3 6.9 6.7 0.91 5.2 1.6 1.0 2.6 1.9
54 4.89 70.2 1.6 242.2 138.7 8.3 4.6 0.87 6.9 2.1 1.0 1.8 1.9
48 4.88 73.0 1.1 267.9 147.0 7.7 3.7 0.96 8.1 3.9 1.4 1.1 1.8
63 4.78 69.8 1.2 227.3 121.1 19.4 6.7 0.85 4.7 1.3 1.0 1.6 2.1
52 4.76 71.0 1.0 244.4 125.3 9.4 1.8 0.86 12.7 3.1 1.3 2.0 2.2
79 4.74 75.1 0.9 243.3 135.6 5.4 5.5 0.83 4.4 1.9 1.0 1.5 1.8
45 4.74 70.2 1.9 239.6 122.5 7.7 7.9 0.84 6.3 2.7 1.0 1.5 2.1
68 4.69 70.7 1.5 235.5 114.6 10.0 0.6 0.89 7.4 1.8 1.0 1.5 2.3

Poorest 10 hybrids 69 3.48 71.7 1.6 191.4 91.6 6.2 20.8 0.72 7.3 2.6 1.0 1.0 2.4
18 3.45 71.0 1.7 249.4 133.7 1.9 6.6 0.86 6.4 2.3 1.4 1.9 2.3
39 3.41 71.4 0.7 251.1 132.2 6.0 0.7 0.87 8.2 3.3 1.5 1.0 1.7
50 3.41 70.7 1.5 232.2 116.2 14.7 3.3 0.69 12.7 3.1 1.3 1.6 2.2
14 3.28 75.6 2.6 257.7 142.8 7.0 11.4 0.74 6.1 1.6 1.2 1.8 1.9
37 3.07 68.4 0.4 222.6 107.8 17.3 14.5 0.70 9.0 1.7 1.1 1.8 2.2
73 2.91 72.3 4.2 250.8 136.2 20.7 2.9 0.55 13.2 2.7 1.2 2.0 2.3
12 2.79 73.7 2.4 251.5 134.7 16.2 4.7 0.70 7.1 2.9 1.0 2.1 2.2
29 2.37 71.2 3.3 238.3 127.3 18.6 21.5 0.74 9.8 2.8 1.3 2.7 2.0
17 2.29 74.4 2.8 239.8 124.9 16.3 11.9 0.67 5.9 1.9 1.4 1.2 2.0
Mean 4.07 70.6 1.4 238.5 122.4 10.3 6.1 0.84 7.6 2.5 1.1 1.8 2.1
LSD 0.68 0.9 0.9 10.1 8.0 9.2 11.0 0.11 4.4 0.9 0.5 0.9 0.4
MSE 1.04 2.9 2.1 285 163.4 129.1 60.6 0.02 35.0 0.5 0.1 0.2 0.1
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging; SL=stem lodging; EPP=ears per plant; ER=ear rot; GLS=grey
leaf spot; RUST=common rust; ET=leaf blight turcicum; SEN=senescence; LSD=least significant difference; MSE=mean square error; DF=degrees of freedom.

63
Table 3.7 Performance of hybrids for grain yield and other agronomic traits across six optimum sites in the 2009/10 and
2010/11 seasons
Entry GYD AD ASI PH EH RL SL EPP HC ER GLS RUST ET
t ha-1 d d cm cm % % # % % 1-5 1-5 1-5
Best 10 hybrids 74 7.86 73.1 0.6 296.5 148.5 4.9 5.0 0.97 22.4 4.5 1.5 1.3 2.2
68 6.79 67.8 0.1 257.1 120.6 10.0 0.6 0.95 5.2 7.3 1.8 1.0 1.5
57 6.63 70.0 0.5 262.5 142.5 11.1 6.7 1.00 11.1 2.7 1.6 1.0 2.6
48 6.63 69.2 0.5 291.1 151.6 6.0 3.7 1.08 8.7 9.2 3.9 1.4 1.1
61 6.62 67.5 -0.1 247.8 120.5 13.1 3.7 0.98 19.3 3.7 2.7 1.0 1.5
7 6.61 67.5 1.0 266.5 122.7 9.8 5.6 0.86 8.8 6.5 1.8 0.9 1.8
79 6.55 71.5 0.5 264.6 141.0 9.0 5.5 0.96 3.3 3.4 1.9 1.0 1.5
5 6.54 69.6 0.7 272.4 142.7 10.8 13.1 1.01 13.1 2.9 3.4 1.1 2.3
51 6.52 67.1 -0.9 265.6 116.9 8.0 7.6 0.98 16.0 7.3 3.6 1.0 1.6
52 6.45 68.5 0.2 261.9 125.6 6.0 1.8 0.95 24.8 4.4 3.1 1.3 2.0

Poorest 10 hybrids 13 4.49 70.2 1.6 276.6 136.1 20.3 6.7 0.85 5.8 4.8 2.4 1.9 2.8
23 4.46 63.8 -0.1 249.9 119.5 4.5 9.0 0.92 16.1 12.9 2.1 1.3 2.3
30 4.44 65.4 -0.1 246.1 110.1 6.9 2.0 0.86 16.1 5.9 3.8 1.2 2.5
22 4.33 64.3 0.1 250.0 120.2 12.1 4.4 0.92 16.5 3.0 3.0 1.0 1.0
73 4.26 68.9 3.0 278.2 142.7 23.7 2.9 0.66 9.0 7.7 2.7 1.2 2.0
18 4.17 68.8 0.6 270.5 134.1 -1.1 6.6 0.98 22.1 6.8 2.3 1.4 1.9
37 3.67 65.3 0.1 240.9 111.6 25.2 14.5 0.87 7.9 6.7 1.7 1.1 1.8
12 3.54 70.4 0.9 282.9 137.1 17.3 4.7 0.78 3.7 4.1 2.9 1.0 2.1
17 3.10 71.2 1.6 270.4 125.2 14.0 11.9 0.90 19.5 8.2 1.9 1.4 1.2
29 2.91 67.8 2.1 257.7 129.1 17.9 21.5 0.81 7.0 6.1 2.8 1.3 2.7
MEAN 5.39 67.7 0.6 260.2 125.5 9.9 6.1 0.94 13.0 5.3 2.5 1.1 1.8
LSD 0.99 1.2 1.0 14.5 9.4 14.0 11.0 0.12 9.6 4.2 0.9 0.5 0.9
MSE 1.24 2.1 1.4 263.7 112.6 147.6 60.6 0.01 92.1 13.4 0.5 0.1 0.2

GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging; SL=stem lodging; EPP=ears per plant; HC=husk cover;
ER=ear rot; GLS=grey leaf spot; RUST; common rust; ET=leaf blight turcicum; LSD=least significant difference; MSE=mean square error; DF=degrees of freedom.

64
Table 3.8 Performance of hybrids for grain yield and other agronomic traits
across two managed drought sites in the 2009/10 and 2010/11 seasons

Entry GYD AD ASI PH EH RL EPP SEN

t ha-1 d d cm cm % # 1-10
Best 10 hybrids 52 3.26 98.0 1.1 236.9 121.6 5.5 0.95 3.7
27 3.07 95.8 0.9 216.6 100.7 0.2 1.07 3.3
19 3.01 91.7 1.9 230.5 105.7 0.7 0.90 3.2
38 3.01 94.6 3.0 222.9 110.3 7.5 0.92 2.9
59 2.84 94.8 -0.1 210.4 102.3 3.5 0.91 3.8
36 2.79 90.6 -0.7 215.9 98.2 0.6 0.96 3.6
61 2.79 94.1 1.9 230.1 114.4 2.3 0.97 3.6
66 2.77 98.7 4.4 220.8 105.7 5.2 0.84 3.9
3 2.74 97.1 3.0 228.5 119.4 31.4 0.79 3.7
30 2.71 93.7 1.3 226.4 112.9 2.1 0.96 3.5

Poorest 10 hybrids 17 1.40 102.3 4.5 214.2 115.9 19.4 0.30 3.5
68 1.40 96.8 4.0 219.3 102.1 -1.1 0.76 4.1
65 1.36 102.8 3.9 238.7 120.6 0.4 0.19 3.8
43 1.36 97.7 3.6 232.4 111.7 5.6 0.67 3.9
13 1.35 97.7 0.1 238.2 112.6 4.5 0.10 3.8
14 1.18 104.0 1.2 237.9 114.1 6.4 0.43 3.3
51 1.17 97.2 1.0 227.6 104.3 9.5 0.68 3.5
73 1.15 99.2 5.8 247.4 124.2 -0.7 0.21 4.0
56 1.04 99.3 2.8 209.3 96.1 23.3 0.44 3.7
69 0.59 95.6 2.9 166.9 66.4 -3.1 0.30 4.1
Mean 2.09 96.4 1.7 223.7 109.3 4.5 0.77 3.6
LSD 0.92 2.6 2.3 20.9 23.0 15.7 0.45 0.73
MSE 0.42 3.3 2.8 220 267.5 62.5 0.05 0.3
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging;
EPP=ears per pant; SEN=senescence; LSD=least significant difference; MSE=mean square error; DF=degrees of freedom.

Entry 61 (RS61P/ CML548) and entry 52 (CML548/2N3d-B) were amongst the best 10
hybrids across sites (Table 3.6) as well as under managed drought (Table 3.8). Entry 59
(CML538/RS61P-B) and entry 36 CML545/K64r-B) had good anthesis silking interval
values of -0.1 and -0.7 respectively, which are desirable under drought conditions. Entry
69 (CML539/RA214P-B) was the poorest amongst all hybrids with a mean yield of 0.59 t
ha-1. Entry 27 (K64r-B/CML536) had 1.07 ears per plant under managed drought and was
the second best performing hybrid in terms of grain yield. The number of days to anthesis
increased under managed drought conditions with a mean of 96.4 days compared to 67.7
days under optimum conditions.

65
Mean hybrid performance for grain yield and other secondary traits under low N
conditions are presented in Table 3.8. Entry 76 (CZH0829), an experimental hybrid from
CIMMYT, was the best performing hybrid, followed by entry 16 CZL03007/SC5522-B).
Generally the mean yields were low under low N conditions with a mean of 0.47 t ha-1.
Entries 61, 54 and 57 remained within the ten best performing hybrids. Plant heights were
reduced under low N conditions with a mean of 217.0 cm. The days to anthesis increased
with anthesis-silking interval with mean values of 72.9 and 7.3 days, respectively. The
ear rot values were also high with entry 2 recording 59.9% ear rot. Ears per plant values
were also lower than under low N with a mean of 0.67 compared to 0.77 under managed
drought.

3.3.2 Performance per se of inbred lines

In across site analysis the inbred line CML548 was the best performer (2.91 t ha-1) and
the poorest performer was SC5522 (0.51 t ha-1) (Table 3.10). CML548 was also the best
performer under optimum conditions and the poorest performer was again SC5522.
Under managed drought 2Kba was the best yielder (3.57 t ha-1) followed by K64r (3.12 t
ha-1). The CIMMYT inbred lines generally displayed better performance across all sites
compared to the DR&SS inbred lines.

3.3.3 Combining ability and heritability

GCA variances had predominance over the SCA variances for all traits. In ears per plant
SCA variance was 52% of the GCA variance. Ear height had the highest heritability
value (93%) and anthesis-silking interval had the lowest heritability value (40%) (Table
3.11). Grain yield had a narrow sense heritability of 60%. Three traits anthesis-silking,
ear rot and GLS had heritability estimates below 50%.

66
Table 3.9 Performance of hybrids across two low nitrogen sites in the 2009/10 and
2010/11 seasons
Entry GYD AD ASI PH RL EPP ER
t ha-1 d d cm % # %
Best 10 hybrids 76 0.81 71.4 4.7 210.9 2.6 0.76 8.8
16 0.77 73.4 5.9 228.0 1.9 0.56 12.6
3 0.76 71.5 10.3 217.7 2.0 0.76 10.1
1 0.75 73.4 4.6 222.3 0.8 0.79 11.7
54 0.73 73.7 4.1 210.4 5.0 0.70 24.6
61 0.73 70.7 2.3 177.7 7.2 0.76 9.3
8 0.72 70.3 6.8 217.1 1.0 0.78 12.5
78 0.71 74.1 7.9 192.3 6.3 0.65 27.9
57 0.68 73.6 8.6 221.9 1.7 0.79 16.2
26 0.67 70.4 6.3 239.7 3.1 0.71 26.0

Poorest 10 hybrids 30 0.26 69.7 8.8 185.1 7.1 0.86 26.8

70 0.22 77.2 17.6 205.8 10.4 0.36 43.9
73 0.22 74.8 12.2 215.0 15.2 0.59 29.6
45 0.21 71.4 11.4 220.8 6.2 0.66 14.4
2 0.2 74.6 8.6 184.9 4.7 0.44 59.9
50 0.16 74.4 7.8 207.6 6.5 0.47 38.7
37 0.15 73.7 9.5 185.7 8.3 0.38 30.6
25 0.14 72.5 12.4 205.3 0.3 0.44 21.3
11 0.10 71.9 12.6 234.5 31.6 0.67 16.8
69 0.08 75.5 8.8 208.3 6.0 0.51 24.3

Mean 0.47 72.9 7.3 217.0 5.0 0.67 19.3

LSD 0.34 3.9 5.0 46.6 22.3 0.22 21.8
MSE 0.03 7.5 6.3 546.9 125 0.02 119.6
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; RL=root lodging; EPP=ears per plant;
ER=ear rot; LSD=least significant difference; MSE=mean square error; DF=degrees of freedom.

3.3.4 Correlation between grain yield and secondary traits

Grain yield was significantly negatively correlated (P≤0.01) with the secondary traits
anthesis date (r=-0.314), anthesis silking interval (r=-0.299) and positively correlated with
ears per plant (r=0.57) (Table 3.12). There was significant correlation (P≤0.01) of
anthesis date with anthesis silking interval, plant height and ear height. Anthesis silking
interval was negatively but significantly correlated with ears per plant. Grain yield was
negatively correlated with senescence although the correlation was not significant.

67
Table 3.10 Performance of inbred parents for grain yield (t ha-1) across different environments in the 2009/10 and 2010/11
seasons
Inbred line Across Optimum Managed drought Low N
N3233-B 1.52 1.94 2.98 0.83
SC5522-B 0.51 0.63 1.26 0.28
2Kba-B 1.17 1.55 3.57 0.35
K64r-B 1.37 1.77 3.12 0.02
NAW5885-B 1.30 1.67 2.61 0.02
SV1P-B 1.05 1.21 1.50 0.65
WCOBY1P-B 1.19 1.91 1.21 0.54
2N3d-B 1.98 2.65 2.29 1.11
RS61P-B 2.53 3.32 2.77 0.53
RA214P-B 1.75 2.31 1.58 0.78
Group Mean 1.44 1.90 2.29 0.51
CML312-B 1.68 2.25 2.41 0.80
CML395-B 2.39 3.20 2.48 0.71
CML442-B 2.27 2.91 2.34 0.76
CML444-B-B 2.58 2.60 1.29 1.43
CML536 1.81 2.37 1.94 0.65
CML537 2.32 3.16 0.94 0.52
CML538 2.37 2.98 2.85 1.21
CML539 1.99 2.54 2.02 1.15
CML548 2.91 3.87 1.51 1.34
CML545 2.58 3.04 2.99 2.30
CML544 1.95 2.42 1.92 1.53
CZL052 1.52 1.99 2.69 0.44
CZL03007 1.91 2.49 2.43 0.47
CML448 1.60 2.08 2.29 0.49
CML449 2.83 3.44 2.86 1.74
Group Mean 2.18 2.75 2.54 1.10
Mean 1.88 2.41 2.63 0.83
LSD 0.49 0.62 1.22 1.32
MSE 0.34 0.36 0.51 0.41
Min 0.51 0.63 0.94 0.02
Max 2.91 3.87 3.57 2.3

68
Table 3.11 General and specific combining ability variances and heritability
estimates for the measured traits across sites
Trait GCA variance SCA variance Heritability %

GYD 75.72 10.76 60

AD 3548.32 23.48 73
ASI 249.36 13.92 40
EPP 0.92 0.52 66
PH 48203.12 3142.44 73
EH 33900.56 742.88 93
SEN 21.32 2.60 74
ER 807.04 408.48 59
GLS 30.36 2.28 49
RUST 2.04 0.4 67
ET 3.44 0.76 92
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear height;
SEN=senescence; ER=ear rot; GLS= grey leaf spot; RUST=common rust; ET=leaf blight turcicum; GCA=general combining
ability; SCA=specific combining ability.

Table 3.12 Correlation coefficients between grain yield and other secondary traits
under managed drought conditions
GYD AD ASI PH EH RL EPP

AD -0.31**
ASI -0.30** 0.30**
PH 0.00 0.56** 0.14
EH 0.07 0.51** 0.04 0.84
RL -0.68 0.24* 0.17 0.11 0.09
EPP 0.57** -0.47 -0.42** -0.17 -0.09 -0.22
SEN -0.20 -0.21 0.13 -0.12 -0.23* -0.03 -0.12
** P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height;
RL=root lodging; EPP=ears per plant; SEN=senescence.

Grain yield was negatively but significantly correlated (P≤0.01) with anthesis silking
interval (r=-0.538), root lodging (r=-0.346) and ear rot (r=-0.434) under low N conditions
(Table 3.13). It was also highly and positively correlated with ears per plant. Anthesis
days were significantly (P≤0.01) and positively correlated with anthesis silking interval
and negatively correlated with ears per plant. Anthesis silking interval was correlated
with root lodging (P≤0.05), ears per plant and ear rot (P≤0.01). The correlation with ears

69
per plant and root lodging was negative. As expected, ears per plant were significantly
and negatively correlated with ear rots.

Grain yield was significantly (P≤0.05) and negatively correlated with root lodging (r=-
0.27) and stem lodging (r=-0.28) under optimum conditions (Table 3.14). It was also
significantly (P≤0.01) and positively correlated with ears per plant (r=0.48). Anthesis
days were significantly (P≤0.01) and positively correlated with anthesis silking interval
and ear and plant height. Plant height was highly and positively correlated with ear height
(r=0.85).

3.3.5 Relative contribution of general combining ability and specific combining

ability sums of squares to variation
GCA sum of squares contributed a larger percentage of variation to grain yield across
sites, and for optimum and low N conditions, whilst under managed drought conditions
SCA contributed a higher percentage of the variation (Table 3.15). Again GCA made a
larger contribution in all sites for anthesis days, plant height and ear height.

Table 3.13 Correlation coefficients between grain yield and other secondary traits
under low nitrogen conditions
GYD AD ASI PH RL EPP

AD -0.20
ASI -0.54** 0.30**
PH 0.13 0.28* 0.07
RL -0.35** 0.03 0.27* -0.20
EPP 0.47** -0.38** -0.38** 0.22 -0.36**
ER -0.43** 0.07 0.35** -0.17 0.12 -0.37**
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; RL=root lodging;
EPP=ears per plant; ER=ear rot.

70
Table 3.14 Correlation coefficients between grain yield and other secondary traits
under optimum conditions
GYD AD ASI PH EH RL SL
AD 0.01
ASI -0.21 0.49**
PH 0.01 0.69** 0.50**
EH 0.15 0.70** 0.49** 0.85**
RL -0.27* 0.03 0.28* 0.05 0.07
SL -0.28* 0.07 0.14 -0.11 -0.01 0.25*
EPP 0.48** 0.01 -0.31** -0.02 0.03 -0.34** -0.10
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height;
RL=root lodging; SL=stem lodging; EPP=ears per plant.

SCA contributed more to variation for ears per plant (prolificacy) in all sites as well as
for anthesis silking interval under drought and low N sites. GCA was partitioned into
GCAf and GCAm (Table 3.15). GCAf dominated GCAm for grain yield across sites (52 vs
31%), under optimum conditions (53 vs 28%) and under low N conditions (44 vs 28%),
whereas under drought they contributed the same percentage (22%). GCAm was generally
higher than GCAf for anthesis days. Anthesis silking interval was controlled by GCAf
across sites and under low N sites, whilst under drought and optimum conditions it was
controlled by GCAm.

3.3.6 Importance of maternal and paternal effects

In across site analysis GCAf mean squares were higher than GCAm mean squares for
grain yield, anthesis days, anthesis silking interval, plant height, ears per plant,
senescence, ear position, GLS and ear rot (Table 3.16). Under optimum conditions GCAf
mean squares were higher than GCAm mean squares for grain yield, anthesis days,
anthesis silking interval, plant height and GLS. GCAf and GCAm mean squares for grain
yield under drought conditions were both not significant. GCAf mean squares were again
higher for grain yield under low N conditions as well as for anthesis silking interval. The
mean squares for senescence were both not significant under drought conditions.

71
Table 3.15 Relative contribution (∑SS) for GCA and SCA across environments
% Sum of Squares
Trait Environment Female GCA Male GCA SCA
GYD Across 52 31 17
Optimum 53 28 19
Drought 22 22 56
Low N 44 28 28
AD Across 26 53 21
Optimum 38 55 7
Drought 39 51 10
Low N 30 47 23
ASI Across 38 24 38
Optimum 34 36 30
Drought 12 21 67
Low N 29 22 49
EPP Across 15 18 67
Optimum 15 30 55
Drought 24 15 61
Low N 19 19 62
SEN Drought 20 25 55
PH Across 40 46 14
Optimum 43 47 10
EH Across 34 60 6
Optimum 32 59 9

GCA=general combining ability; SCA=specific combining ability; GYD=grain yield; AD=anthesis days; ASI=anthesis silking
interval; EPP=ears per plant; SEN=senescence; PH=plant height; EH=ear height.

3.3.7 General combining ability effects across environments

The inbred line GCA effects across sites for grain yield, anthesis days, anthesis silking
interval, ears per plant, plant height, ear height, root lodging, stem lodging and
senescence are presented in Figures 3.1, 3.2 and Table 3.17.

The line with the best GCA value for grain yield (0.72) (Figure 3.1) and ears per plant
(0.064) (Table 3.17) across all environments was RS61P. SC5522 had the poorest GCA
effect for grain yield (-0.95) (Figure 3.1). The ideal line GCA values for anthesis days
would be the negative values and in this case the line with the best GCA value for
anthesis days was 2Kba (-1.99) (Figure 3.2).

72
Table 3.16 Mean squares for general combining ability due to female and male
effects under different environments

Trait Across Optimum Managed drought Low N

GCAf GCAm GCAf GCAm GCAf GCAm GCAf GCAm
GYD 48.42*** 18.93** 37.88*** 12.76*** 3.21 2.02 3.98*** 1.62***
AD 887.08*** 646.52*** 213.55*** 194.83*** 210.91*** 177.88*** 76.78*** 76.98***
ASI 62.34*** 25.18*** 22.68*** 15.39*** 10.38** 11.46*** 48.91*** 23.14**
PH 16331.75*** 12050.78*** 11775.29*** 8195.45*** 1880.71*** 1632.11*** 3563.03*** 1941.45***
EH 8475.14*** 9060.36*** 4574.82*** 5346.24*** 1366.65*** 1399.11*** 541.46* 577.16**
EPP 0.23* 0.17 0.048* 0.062*** 0.18** 0.07 0.12*** 0.07**
SEN 5.32*** 3.21*** nm nm 0.74 0.6 nm nm
EPO 0.038*** 0.029*** 0.022*** 0.018*** 0.009* 0.007* 0.005 0.006
GLS 7.59*** 3.28*** 7.58** 3.27*** nm nm nm nm
RUST 0.28** 0.51*** 0.28* 0.51*** nm nm nm nm
ET 0.86** 0.85*** 0.86*** 0.84*** nm nm nm nm
ER 201.76** 108.16 61.66** 96.74*** nm nm nm nm
***P≤0.001; **P≤0.01; *P≤0.05; GCAf=female general combining ability; GCAm=male general combining ability; GYD=grain
yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPP=ears per plant;
SEN=senescence; EPO=ear position; GLS=grey leaf spot; RUST=common rust; ET=leaf blight turcicum; ER=ear rot; nm=not
measured.

SC5522 had the highest positive GCA effect for anthesis days (5.19) which is an
indication that it is a very late line. K64r had the best GCA value for anthesis silking
interval (-0.69), plant height (-9.5) and ear height (-7.7) (Table 3.17).

The tester GCA effects for grain yield are presented in Figure 3.3. The tester with the best
GCA effect for grain yield was tester 11 (CML548). Testers with the poorest GCA effect
for grain yield were testers 1 (CML395) and 12 (CZL052) (Figure 3.3). Testers 3 and 7
had good anthesis days values whilst tester 10 had the highest positive GCA value for
anthesis days, indicating that it is very late (Figure 3.4). The tester with the best GCA
effect for anthesis silking interval was tester 12, while tester 1 had the poorest GCA
effect for anthesis silking interval. The tester GCA effects for other agronomic traits are
presented in Table 3.18.

73
 1

0.8

0.6

0.4

0.2
Line GCA

0
K64r N3.2.3.3 RS61P NAW5885 2N3d 2Kba SC5522 RA214P
-0.2

-0.4

-0.6

-0.8

-1

-1.2
Lines

Figure 3.1 Line general combining ability (GCA) values for grain yield for all
environments.

Table 3.17 Line general combining ability values for other agronomic traits for all
environments
Line ASI EPP PH EH RL SL SEN

K64r -0.69* 0.001 -9.5** -7.7** 1.8 1.2 0.4*

N3.2.3.3 0.34 -0.026 6.2* 5.1* 0.5 -0.3 1.2***

RS61P -0.57* 0.064 -6.6* 1.7 -0.9 -5.6* -1.9***
NAW5885 0.71** 0.015 2.3 -2.7 1.0 1.3 0.2*
2N3d -0.29 -0.011 14.0 7.2** -6.7* 2.8 1.5***
2Kba -0.05 -0.009 -5.6* -3.9 2.4 -0.1 -0.5*
SC5522 0.70** -0.062 20.9*** 16.8*** -1.5 0.1 -0.8*
RA214P 0.38 -0.016 0.7 -1.0 -2.3 0.3 0.1
LSD (0.05) 0.015 0.00056 1.54 0.98 1.43 4.33 0.004
SE 0.348 0.067 3.545 2.83 3.28 4.19 0.111
***P≤0.001; **P≤0.01; *P≤0.05; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear height;
RL=root lodging; SL=stem lodging; SEN=senescence; LSD=least significant difference.

74
 6

3
Line GCA

0
K64r N3.2.3.3 RS61P NAW5885 2N3d 2Kba SC5522 RA214P
-1

-2

-3
Lines

Figure 3.2 Line general combining ability (GCA) values for anthesis days for all
environments.

0.6

0.4

0.2

0
Tester GCA

1 2 3 4 5 6 7 8 9 10 11 12
-0.2

-0.4

-0.6

-0.8

-1
Testers

Figure 3.3 Tester general combining ability (GCA) effects for grain yield for all
environments.

75
 6

3
Tester GCA

0
1 2 3 4 5 6 7 8 9 10 11 12
-1

-2

-3
Testers

Figure 3.4 Tester general combining ability (GCA) effects for anthesis days for
all environments.

Table 3.18 Tester general combining ability effects for other agronomic traits under
all environments
Tester ASI EPP PH EH RL SL SEN

1 1.00** -0.078 19.0*** 15.7*** 2.2 1.8 0.5**

2 -0.50* 0.022 -9.3** -6.0* -8.4* -1.8 -0.3*
3 -0.20 0.005 -13.2*** -11.1*** 0.1 0.4 -0.3*
4 0.10 0.003 -0.3 -3.3 -0.8 -1.5 -0.7**
5 0.10 0.000 2.4 -0.1 4.0 -4.0 0.7**
6 -0.10 -0.005 -6.0* -3.8 2.3 3.8 -1.6***
7 -0.60* 0.034 -4.0 -1.0 -1.1 -2.2 1.1***
8 -0.20 -0.032 -3.2 -4.9 -1.0 -2.4 0.3*
9 0.30 -0.026 11.3*** 11.7*** -1.0 -1.8 0.2
10 -0.10 0.027 20.0*** 15.6*** -0.1 14.7*** -0.7**
11 0.20 0.058 -2.3 -1.6 -1.5 -0.9 0.8**
12 -1.00** 0.026 -6.4* 0.2 -9.8** 1.4 -3.3***
LSD (0.05) 0.02 0.00056 1.54 0.98 1.43 4.33 0.004
SE 0.43 0.082 4.34 3.47 4.02 5.13 0.136
***P≤0.001; **P≤0.01; *P≤0.05; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear height;
RL=root lodging; SL=stem lodging; SEN=senescence; LSD=least significant difference.

76
3.3.8 General combining ability effects under optimum conditions
The line GCA effects for grain yield were significant (P≤0.05). The line with the best
positive GCA effect for grain yield was RS61P (0.64) followed by N3.2.3.3 (0.59)
(Figure 3.5). Four lines K64r, 2N3d, 2Kba and SC5522 had negative GCA effects for
grain yield under optimum conditions with SC5522 having the poorest GCA effect of -
1.32. GCA effects for the other agronomic traits under optimum conditions are presented
in Table 3.19. There were five lines with negative GCA effect for anthesis days, which is
an indication of their earliness. K64r recorded a GCA effect for anthesis days of -1.68
followed by 2Kba with a GCA effect of -1.43. SC5522 had the highest positive GCA
effect for anthesis days of 3.75. Anthesis silking interval is another important trait to
consider in selection and lines K64r, RS61P, 2N3d and 2Kba recorded negative GCA
effects for anthesis silking interval of -0.64, -0.42, -0.29 and -0.14 respectively.
Generally RS61P displayed good GCA effects for traits such as plant height, ears per
plant, root and stalk lodging and diseases with the exception of ear height. SC5522 had
the highest positive GCA effect for plant height (25.49), ear height (18.24) and root
lodging (5.59). K64r had positive GCA effects for all diseases recorded.

0.5

0
Line GCA

K64r N3.2.3.3 RS61P NAW5885 2N3d 2Kba SC5522 RA214P

-0.5

-1

-1.5
Lines

Figure 3.5 Line general combining ability (GCA) effects for grain yield under
optimum conditions.

77
Table 3.19 Line general combining ability effects for anthesis days and other agronomic traits under optimum conditions

Line AD ASI EPP PH EH RL SL ER ET GLS RUST

K64r -1.68*** -0.64** -0.014 -12.76*** -9.60*** 0.38 -0.95 1.19*** 0.20** 0.01 0.05
N3.2.3.3 0.45* 0.40* 0.009 11.39*** 8.30*** 2.55 0.62 -0.42 0.03 0.39* 0.01
RS61P -0.64** -0.42* 0.039 -11.13*** 1.38 -1.13 -1.73 -1.30*** -0.01 -0.38* -0.12*
NAW5885 -0.11 0.64** -0.017 4.99* -1.33 0.06 0.46 0.08 -0.23** 0.05 0.02
2N3d 1.58*** -0.29 -0.001 14.32*** 4.07** 0.97 0.36 -0.17 -0.03 1.36*** 0.15*
2Kba -1.43*** -0.14 -0.013 -6.53* -4.90** -0.64 1.32 0.79* 0.04 -0.11 0.00
SC5522 3.75*** 0.96*** -0.054 25.49*** 18.24*** 5.59* -0.26 -4.78*** 0.24** -0.10 0.22*
RA214P 1.83*** 0.15 0.014 1.78 0.07 -4.57* 0.13 0.47 -0.09 -0.39* -0.06
LSD (0.05) 0.027 0.017 0.00019 2.64 1.57 2.51 1.35 0.21 0.004 0.0078 0.002
SE 0.309 0.242 0.0258 3.03 2.34 2.95 1.76 0.69 0.084 0.1153 0.061
***P≤0.001; **P≤0.01; *P≤0.05; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear height; RL=root lodging; SL=stem lodging; ER=ear
rot; ET=leaf blight turcicum; GLS=grey leaf spot; RUST=common rust; LSD=least significant difference.

78
Tester GCA effects for grain yield and other agronomic traits are presented in Table 3.20.
Tester 11 (CML548) had the highest positive GCA effect for grain yield and the tester with
the poorest GCA effect was tester 10 (CML536) with a negative GCA effect. Tester 2
(CML312) had the best GCA effects for both anthesis days and anthesis silking interval,
whilst tester 10 (CML536) had the highest positive GCA effect for anthesis days.

3.3.9 General combining ability effects under managed drought conditions

Lines K64r and RS61P had the highest significant (P≤0.05) positive GCA effects for grain
yield under drought conditions of 0.37 followed by NAW5885 with a GCA effect of 0.12
(Figure 3.6). The rest of the lines had negative GCA effects for grain yield under drought
conditions with SC5522 having the poorest GCA effect. The GCA effects for the secondary
traits anthesis days, anthesis silking interval, ears per plant and senescence are presented in
Table 3.21. Two lines K64r and RS61P, had negative GCA effects for both anthesis days and
anthesis silking interval. These two lines also recorded good GCA effects for ears per plant
and senescence. Lines SC5522 and RA214P had poor GCA effects for anthesis days and
other secondary traits.

3.3.10 General combining ability effects under low nitrogen conditions

Three lines had significant positive GCA effects for grain yield under low N conditions with
RS61P having the best GCA effect of 0.58 (Figure 3.7). Line 2N3d had the poorest GCA
effect of -0.70 followed by SC5522 with a GCA effect of -0.63. Tester 3 (CML 539) had the
best GCA effect for grain under low N conditions (0.30) while tester 1 (CML 395) had the
lowest GCA effect (-0.68) (Figure 3.8). Lines 2Kba and RS61P showed significant negative
GCA effects for anthesis days of -1.6 and -1.4 respectively whilst lines 2N3d and SC5522
showed significant positive GCA effects of 2.7 and 4.1 respectively (Table 3.22). Line
RS61P had a significant negative GCA effect for anthesis silking interval. RS61P and
N3.2.3.3 had positive significant GCA effects for ears per plant. Tester GCA effects for other
traits are presented in Table 3.23.

79
Table 3.20 Tester general combining ability effects for grain yield and other agronomic traits under optimum conditions
Tester GYD AD ASI PH EH RL SL EPP ER ET GLS RUST

1 -1.19*** 2.7*** 1.4*** 20.4*** 15.2*** 9.7*** -2.2 -0.073 -4.9*** 0.6** 0.3* 0.4*
2 -0.97** -3.3*** -1.2*** -21.5*** -12.2*** -9.6*** 9.5*** 0.004 -3.3*** -0.3* -0.6** -0.1
3 0.22 -1.6** -0.4* -15.7*** -10.7*** -4.1 2.6* -0.001 -0.8* -0.1 0.3* -0.1
4 -0.02 -0.4 0.2 -1.0 -4.8* -1.7 -1.9 -0.009 -0.1 0.1 0.5** 0.0
5 0.16 0.5 0.2 2.9* -0.7 3.6 -1.8 0.029 0.4 -0.1 0.3* -0.1
6 0.02 -0.6 -0.2 -4.6* -3.2* 3.2 1.6 -0.011 -1.7** -0.1 -0.5** 0.0
7 0.10 -2.2*** -0.6* -6.3* -3.2* -5.9* -1.3 0.040 2.7** 0.0 -0.1 -0.1
8 0.25 -0.4 -0.2 -4.6* -5.9* -1.3 -2.9* -0.048 1.4** 0.1 -0.4* -0.1
9 0.01 3.0*** 0.2 15.9**** 16.7*** -0.4 4.3** 0.025 1.4** 0.0 -0.1 0.2*
10 -1.34*** 3.6*** -0.2 29.6*** 22.7*** 13.2*** -3.7* 0.016 -3.5*** 0.2* -0.1 0.3*
11 0.48* 0.1 0.3 -0.1 -1.1 -3.9 -1.0 0.003 -0.9* -0.2* 0.0 0.0
12 -1.10*** -3.0*** -1.0*** -23.1*** -4.7* -3.5 -1.6 0.004 -3.3*** 0.6** -0.3* 0.2*
LSD (0.05) 0.024 0.027 0.017 2.64 1.57 2.51 1.35 0.00019 0.27 0.004 0.0078 0.002
SE 0.25 0.379 0.297 3.71 2.86 3.61 2.16 0.03162 0.87 0.1030 0.1414 0.075
***P≤0.001; **P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging; SL=stalk lodging; EPP=ears
per plant; ER=ear rot; ET= leaf blight turcicum; GLS=grey leaf spot; RUST=common rust; LSD=least significant difference.

80
Tester 3 (CML539) had the best significant negative GCA effect for anthesis days (-2.2),
whilst tester 12 (CZL052) had a significant negative GCA effect for anthesis silking interval
(-1.7). Tester 4 (CML 442) had the best GCA effect for ears per plant under low N conditions
(Table 3.23), whilst tester 5 (CML 537) had the poorest GCA effect. Testers 1 (CML395), 2
(CML 312), 6 ([CML445/ZM621B]-2-1-2-3-1-B*8), 10 (CML536) and 11 (ZM523A-16-2-
1-1-B*5) showed good GCA effects for both plant height and ear height.

0.5

0.4

0.3

0.2

0.1
Line GCA

0
K64r N3.2.3.3 RS61P NAW5885 2N3d 2Kba SC5522 RA214P
-0.1

-0.2

-0.3

-0.4

-0.5

-0.6
Lines

Figure 3.6 Line general combining ability (GCA) effects for grain yield under
managed drought conditions.

3.3.11 Specific combining ability effects across all environments

The SCA effects for crosses are presented in Table 3.24. The line and tester GCA effects are
also presented in the same table in order to determine whether lines with good GCA effects
also produced crosses with good SCA effects or vice versa.

81
Table 3.21 Line general combining ability effects for anthesis days and other secondary
traits under managed drought conditions
Line AD ASI EPP SEN

K64r -2.0*** -0.7* 0.07 -0.02

N3.2.3.3 1.4** 0.4 -0.06 0.13
RS61P -2.2*** -0.4 0.06 -0.21*
NAW5885 0.3 0.4 0.07 -0.23*
2N3d 2.8*** -0.5 -0.01 0.01
2Kba -3.4*** 0.3 0.01 0.09
SC5522 2.3*** 0.1 -0.13 0.09
RA214P 2.7*** 0.7* -0.07 0.15
LSD (0.05) 0.12 0.13 0.0019 0.01
SE 0.373 0.39 0.046 0.117
***P≤0.001; **P≤0.01; *P≤0.05; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; SEN=senescence;
LSD=least significant difference.

0.8

0.6

0.4

0.2
Line GCA

0
K64r N3.2.3.3 RS61P NAW5885 2N3d 2Kba SC5522 RA214P
-0.2

-0.4

-0.6

-0.8
Lines

Figure 3.7 Line general combining ability (GCA) effects for grain yield under low
nitrogen conditions.

82
 0.4

0.2

0
Tester GCA

1 2 3 4 5 6 7 8 9 10 11 12
-0.2

-0.4

-0.6

-0.8
Testers

Figure 3.8 Tester general combining ability (GCA) effects for grain yield under low
nitrogen conditions.

Table 3.22 Line general combining ability effects of other agronomic traits under low
nitrogen conditions
Line AD ASI EPP PH EH

K64r -0.4 -0.9* -0.0296 -6.9 -3.5

N3.2.3.3 0.2 0.8* 0.0230 3.3 1.2
RS61P -1.4** -2.1*** 0.1314 8.8* 6.8*
NAW5885 -0.7 0.3 -0.0335 8.5* 1.0
2N3d 2.7*** 1.5** 0.0087 -30.0*** -7.2**
2Kba -1.6** -0.4 -0.0298 4.3 2.1
SC5522 4.1*** 1.9** -0.0711 -16.0*** -7.5**
RA214P 1.4** 1.3** -0.0427 2.4 -1.5
LSD (0.05) 0.26 0.34 0.00089 24.05 8.04
SE 0.55 0.63 0.032 5.26 3.05
***P≤0.001; **P≤0.01; *P≤0.05; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear
height; LSD=least significant difference.

83
Table 3.23 Tester general combining ability effects of other agronomic traits under low
nitrogen conditions
Tester AD ASI EPP PH EH

1 3.6*** 1.9** -0.0972 -23.6*** -9.5**

2 -1.8** 1.1* 0.0849 -26.5*** -8.3**
3 -2.2*** -0.1 0.0703 7.2 -2.7
4 -1.3* -1.3* 0.0972 3.0 -0.1
5 1.1* -0.2 -0.0005 5.1 4.9*
6 -0.8 1.6** -0.0793 -8.9* -5.1*
7 -1.5* -1.4** 0.0278 4.2 4.7*
8 -0.7 -0.4 -0.0093 2.0 -0.9
9 2.5*** 0.5 -0.0283 10.1* 8.0**
10 5.5*** -0.7 0.0566 -25.3*** -12.8***
11 -0.2 0.7 -0.0534 -1.6 -1.1
12 -1.1* -1.7** 0.0274 -11.2** 0.4
LSD (0.05) 0.26 0.34 0.00089 24.05 8.04
SE 0.67 0.77 0.03924 6.45 3.73
***P≤0.001; **P≤0.01; *P≤0.05; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; PH=plant height; EH=ear
height; LSD=least significant difference.

The cross with the highest SCA (0.79) for grain yield was between line 3 (RS61P) and tester
9 (CML444) whilst the poorest (-0.76) was between line 8 (RA214P) and tester 3 (CML539).
RS61P (line 3) had the best positive GCA effect and CML444 (tester 9) had negative GCA.
The second best cross was between line 6 (2Kba) and tester 11 (CML548). Both the line and
the tester had positive GCA effects. The poorest cross was between line 6 (2Kba) and tester 9
(CML444) and both these lines had negative GCA effects for grain yield. SC5522 had
significant positive SCA effects of 0.77 and 0.75 respectively (data not shown) with
CML395 and CML 536 of 0.77 despite the fact that it had the poorest GCA effect for grain
yield.

Line 6 (2Kba) and tester 11 (CML548) produced a cross with the best SCA effects for
anthesis days (-5.73) (Table 3.25). 2Kba (line 6) had a negative GCA effect (-1.99) for
anthesis days whilst CML548 (tester 11) also had a negative GCA effect (-0.29) for anthesis
days. The same line 2Kba produced a cross with a positive SCA effect (2.66) with tester 9
(CML444), which had a positive GCA effect of 2.86. Line 4 (NAW5885) with a negative

84
Table 3.24 Specific combining ability effects for grain yield across all environments
Line
Tester 1 2 3 4 5 6 8 GCA

3 0.50* 0.15 -0.06 -0.52* 0.07 0.35 -0.76** 0.16

4 0.02 -0.23 -0.65* 0.51* 0.06 0.18 -0.17 0.19
5 0.36 -0.09 -0.26 -0.32 -0.45 0.27 0.03 0.24
7 0.14 0.50* -0.20 -0.19 -0.50* 0.12 0.13 0.01
8 -0.09 0.03 -0.07 -0.17 -0.26 -0.22 0.36 0.13
9 0.35 0.32 0.79** -0.30 0.33 -0.92** -0.27 -0.11
11 0.03 -0.44 0.08 -0.50* -0.16 0.68* 0.19 0.34
GCA -0.12 0.17 0.72 0.13 -0.9 -0.27 0.18
Lines 1=K64r; 2=N3.2.3.3; 3=RS61P; 4=NAW5885; 5=2N3d; 6=2Kba; 8=RA214P; Testers 3=CML539; 4=CML442; 5=CML537;
7=CML545; 8=CML538; 9=CML444; 11=CZL052; GCA=general combining ability. SE=0.82; LSDs:V(sij)=0.014* and 0.018** for
testing significance of SCA effects from zero; *P≤0.05; **P≤0.01.

Table 3.25 Specific combining ability effects for anthesis days across all environments

Line

Tester 1 2 3 4 5 6 8 GCA
3 -0.60 0.87 -0.56 0.03 0.81 -0.16 1.31* -2.11
4 -0.32 -0.34 1.61* -0.44 0.55 -0.02 0.55 -0.78
5 -0.47 -0.08 -0.71 1.52* 0.57 -0.03 0.80 0.51
7 -0.30 0.42 0.28 0.23 0.02 -0.14 -0.96 -2.13
8 -0.05 -0.51 0.35 1.48* 0.93 0.08 -0.53 -0.91
9 -0.54 -0.96 -0.25 -1.97* -1.16 2.66*** -0.20 2.86
11 0.15 -0.49 0.40 2.07** -0.57 -5.73*** 0.88 -0.29
GCA -1.44 0.24 -1.12 -0.50 3.60 -1.99 1.40
Lines 1=K64r; 2=N3.2.3.3; 3=RS61P; 4=NAW5885; 5=2N3d; 6=2Kba; 8=RA214P; Testers 3=CML539; 4=CML442; 5=CML537;
7=CML545; 8=CML538; 9=CML444; 11=CZL052; GCA=general combining ability. SE=1.27; LSDs:V(sij)=2.10* and 2.96** for
testing significance of SCA effects from zero; *P≤0.05; **P≤0.01; ***P≤0.001.

GCA effect (-0.50) produced three crosses with high positive SCA effects of 2.07, 1.52 and
1.48 with testers 11(CML548), 5 (CML537) and 8 (CML538) respectively.

SCA effects for anthesis silking interval are presented in Table 3.26. Crosses with negative
SCA effects are ideal. The cross with the best SCA effect (-1.53) was between line 4
(NAW5885) and tester 9 (CML444) and both the line and the tester had positive GCA effects
of 0.71 and 0.26 respectively. The same cross had the best GCA effect for anthesis days. The
poorest cross (1.31) was between line 6 (2Kba) and tester 9 (CML444) with GCA effects of -
0.05 and 0.26, respectively. The second best cross was between line 2 (N3.2.3.3) and tester

85
11 CML548 with a SCA effect of -0.66 and the two parents had positive GCA effects of 0.34
and 0.26, respectively.

Table 3.26 Specific combining ability for anthesis silking interval across all
environments

Line
Tester 1 2 3 4 5 6 8 GCA
3 -0.27 0.01 0.35 0.46 -0.15 -0.60 0.25 -0.16
4 -0.25 0.54 -0.36 0.33 -0.48 -0.23 0.29 0.12
5 -0.14 -0.13 0.14 0.03 0.30 -0.04 0.05 0.08
7 -0.17 -0.09 0.11 1.21* -0.52 -0.06 -0.60 -0.60
8 -0.01 0.40 -0.58 -0.26 0.65 0.39 -0.22 -0.16
9 0.20 -0.30 0.25 -1.53* -0.30 1.31* -0.36 0.26
11 0.28 -0.66 -0.41 0.30 -0.23 -0.05 0.74 0.24
GCA -0.69 0.34 -0.57 0.71 -0.29 -0.05 0.38
Lines 1=K64r; 2=N3.2.3.3; 3=RS61P; 4=NAW5885; 5=2N3d; 6=2Kba; 8=RA214P; Testers 3=CML539; 4=CML442; 5=CML537;
7=CML545; 8=CML538; 9=CML444; 11=CZL052; GCA=general combining ability. SE=1.21; LSDs: V(sij)=1.99* and 2.82** for
testing significance of SCA effects from zero. *P≤0.05

3.3.12 Specific combining ability effects under optimum conditions

The cross between line 3 (RS61P) and tester 9 (CML444) had the highest positive SCA
effect for grain yield under optimum conditions (0.99) and both parents had positive GCA
effects for grain yield (Table 3.27). The second best cross was between line 1 (K64r) and
tester 3 (CML 539) and one of the parents K64r had a negative GCA effect for grain yield (-
0.38). The cross between line 3 (RS61P) and tester 4 (CML 442) had the highest negative
SCA effect (-1.10). Line 6 (2Kba) and tester 4 (CML 442) produced a cross with a positive
SCA effect (0.47) when they both had negative GCA effects for grain yield of -0.70 and -
0.02 respectively. Line 4 (NAW5885) had positive GCA effect for grain yield but it produced
crosses with negative SCA effects with testers 3 (CML 539), 5 (CML 537), 7 (CML545), 8
(CML 538), 9 (CML 444) and 11 (CML548) (Table 3.26).

3.3.13 Specific combining ability effects under managed drought conditions

Line 5 (2N3d) and tester 9 (CML 444) had negative GCA effects of grain yield of -0.21 and -
0.42 respectively but they produced a cross with the highest significant positive SCA effect
under managed drought (1.16) (Table 3.28). Tester 4 (CML 442) had the highest GCA effect
but had crosses with negative SCA effects of -0.80 with line 5 (2N3d) and -0.74 with line 3
(RS61P) which also had a positive GCA effect. The parental lines with positive GCA effects

86
sometimes produced hybrids with negative SCA, whilst on the other hand some parental lines
with negative GCA effects produced hybrids with positive GCA effects.

Table 3.27 Specific combining ability for grain yield under optimum conditions

Line
Tester 1 2 3 4 5 6 8 GCA

3 0.94** 0.16 -0.01 -0.44 0.35 -0.19 -1.05** 0.22

4 -0.33 -0.07 -1.10** 0.57 0.42 0.47 -0.13 -0.02
5 0.38 -0.35 -0.05 -0.18 -0.26 0.11 -0.19 0.16
7 0.13 0.37 0.12 -0.26 -0.71 -0.07 0.13 0.10
8 -0.37 0.51 -0.49 -0.57 -0.18 -0.17 0.76* 0.25
9 0.62 0.67 0.99** -0.68 -0.06 -0.86* -0.54 0.01
11 -0.05 -0.48 0.12 -0.36 -0.55 0.72* -0.07 0.48
GCA -0.38 0.59 0.64 0.36 -0.92 -0.70 0.37
Lines 1=K64r; 2=N3.2.3.3; 3=RS61P; 4=NAW5885; 5=2N3d; 6=2Kba; 8=RA214P; Testers 3=CML539; 4=CML442; 5=CML537;
7=CML545; 8=CML538; 9=CML444; 11=CZL052; GCA=general combining ability. SE=0.71; LSDs: (Vsij)=1.17* and 1.65** for
testing significance of SCA effects from zero; *P≤0.05; **P≤0.01.

Table 3.28 Specific combining ability effects under managed drought conditions
Line
Tester 1 2 3 4 5 6 8 GCA
3 0.32 0.32 0.78** -0.46 -0.06 -0.41 -0.77** -0.13
4 0.11 -0.13 -0.74** 0.66* -0.80** 0.50* 0.25 0.59
5 -0.34 -0.87** 0.17 0.79** -0.51* 0.03 0.58* -0.04
7 0.50* 0.62* -0.56* -0.59* -0.34 0.64* 0.06 -0.16
8 0.17 -0.60* 0.20 0.53* -0.18 -0.20 -0.08 -0.10
9 -0.37 -0.36 0.06 0.24 1.16*** -0.69* 0.27 -0.42
GCA 0.37 -0.08 0.37 0.12 -0.21 -0.15 -0.26
Lines 1=K64r; 2=N3.2.3.3; 3=RS61P; 4=NAW5885; 5=2N3d; 6=2Kba; 8=RA214P; Testers 3=CML539; 4=CML442; 5=CML537;
7=CML545; 8=CML538; 9=CML444; 11=CZL052; GCA=general combining ability. SE=0.42; LSDs: V(sij)=0.69* and 0.97** for
testing significance of SCA effects from zero. *P≤0.05; **P≤0.01; ***P≤0.001.

3.3.14 Specific combining ability effects under low nitrogen conditions

Parental lines generally produced hybrids with negative SCA effects (Table 3.29). Line 1
(K64r) and tester 4 (CML442) produced a hybrid with the highest significant positive SCA
effect (0.73). Line 3 (RS61P) with the highest GCA effects produced hybrids with negative
SCA effects for grain yield with testers 4 (CML442), 5 (CML537) and 7 (CML545). Line 3
(RS61P) and tester 3(CML539) produced a cross with the second best (0.66) SCA effect.

87
Table 3.29 Specific combining ability effects under low nitrogen conditions

Line
Tester 1 2 3 4 5 6 8 GCA

3 -0.11 0.10 0.66** -0.15 -0.41** 0.16 -0.64** 0.30

4 0.73** -0.15 -0.55** 0.05 -0.12 -0.09 -0.11 0.27
5 0.13 0.43** -0.41** -0.57** -0.16 0.41** -0.09 0.18
7 0.18 -0.07 -0.28 0.23 0.34* -0.12 -0.01 -0.07
8 -0.30* -0.06 0.26 -0.56** -0.31* 0.21 0.43** -0.01
9 0.26 0.07 0.49** 0.06 -0.07 -0.53** -0.10 0.11
11 0.09 0.03 0.10 -0.16 -0.19 -0.18 -0.12 0.09
GCA -0.07 0.14 0.58 0.21 -0.70 -0.12 -0.21
Lines 1=K64r; 2=N3.2.3.3; 3=RS61P; 4=NAW5885; 5=2N3d; 6=2Kba; 8=RA214P; Testers 3=CML539; 4=CML442; 5=CML537;
7=CML545; 8=CML538; 9=CML444; 11=CZL052; GCA=general combining ability. SE=0.17; LSDs: V(sij)=0.28* and 0.40** for
testing significance of SCA effects from zero; *P≤0.05; **P≤0.01.

3.4 Discussion
The ANOVA revealed significant differences for grain yield and other agronomic traits
among the inbred parents and their single cross hybrids in different environments. The
environments under which evaluations were done were diverse and this was proved by the
significant differences among them. The differential performance of genotypes over
environments, as found in the current study, has implications for breeding, presenting the
question of whether to breed for specific or general adaptation. On the other hand such
information is useful in identifying a suitable genotype for specific environments. Entry 74
(SC727), a late maturing single cross hybrid, out-yielded all experimental hybrids as the
other hybrids were among the medium maturity group. Late maturing hybrids always tend to
give higher yields because when the season length is ideal they have more time to capture
and utilise the sun’s energy when forming their grain (Smith et al., 2009). Vasal et al. (1992)
reported mean grain yields of 4.59 t ha-1 under sub-tropical environments and 4.35 t ha-1
under temperate environments. The average grain yield reported in this study for all crosses,
across all environments, was 4.07 t ha-1 and this was again similar to the result reported by
George et al. (2011). The four hybrids (entries 54, 57, 61 and 63) that were amongst the top
10 best performing hybrids in across site analysis had one common parent, RS61P. The
highest yielding hybrid among the experimental hybrids, entry 61, was from a cross between
heterotic groups CIMMYT A x DR&SS SC (B) and the lowest yielding hybrid entry 17 was
between CIMMYT B x DR&SS SC (B) heterotic groups, while some of the best performing

88
hybrids (entries 54, 57 and 61) were among CIMMYT B x DR&SS SC (B). Results show
that there was no consistency in hybrid performance based on predefined heterotic groups.
Similar results were reported by Dhliwayo et al. (2009). Breeders at CIMMYT initially put
more emphasis on population improvement at the expense of hybrids and they are now
extracting lines from these broad populations (Kassa et al., 2012). Therefore heterotic
grouping used at CIMMYT is too broad and it is difficult to divide lines into heterotic groups
when the lines were developed from the same original pool without considering their racial
origin or heterotic pattern. Again the designation of A and B groups at CIMMYT might have
done in an arbitrary manner for the convenience of managing germplasm lines in hybrid-
oriented programmes.

Lines and testers differed significantly for anthesis days under different environments. The
number of days to anthesis was expressed differently by the lines across the tested
environments. Mungoma and Pollak (1988) and Betran et al. (2003) reported significant
differences for anthesis days under different environments and their results agree with
findings in this study. The mean days to anthesis were higher under stress conditions
(drought and low N) due to the negative effects the stress had on the growth of the maize
crop. Vasal et al. (1992) recorded mean days to anthesis of 71 under temperate environments
and 54 under sub-tropical environments. This has an implication on breeding, in that
selection for anthesis days has to be done under optimum conditions to cater especially for
seed production.

The trial mean yield of 2.09 t ha-1 reported in this study under drought was 61% lower than
the trial mean (5.39 t ha-1) under optimum conditions. According to Banziger et al. (2000) the
level of yield reduction observed in the current study was associated with severe drought and
results are within previously reported yield reduction ranges of drought stress levels to screen
maize genotypes (Camphos et al., 2006). Betran et al. (2003) reported yield reductions of 13
and 50% under intermediate and severe drought stress, respectively, in one site and
reductions of 5 and 48% in another site during the same season. Banziger et al. (2000)
however, reported yield reductions under moderate drought of 15-20%. Negative anthesis
silking-interval values reported in this study were indicative of ideal varieties under drought

89
conditions. This showed that the varieties were able to synchronise pollen shedding with silk
emergence. Edmeades et al. (1993) concluded that a reduced anthesis silking interval was a
sign of improved apportioning of assimilates to ears around flowering time. This scenario
assists drought tolerant selection cycles to reach silking earlier and have a better ear biomass
at anthesis. Entry 27 had an average number of ears per plant of 1.07 and this contributed to
its good performance.

Highly significant and negative correlations between grain yield, anthesis days and anthesis
silking interval under stress conditions were reported in this study and results are indicative
of the fact that an increase in any of these traits results in a corresponding decrease in grain
yield. These findings are consistent with findings by other investigators (Banziger et al.,
1997; Betran et al., 2003; Zaidi et al., 2004; Derera, 2005; Gissa, 2008; Pswarayi and Vivek,
2008). A significant correlation (r=-0.39) of anthesis silking interval with grain yield for
hybrids across all environments was found (data not shown) and this is consistent with
findings of Betran et al. (2003), Zaidi et al. (2004) and Derera (2005) of -0.33 to -0.45 under
moisture stress conditions. These results suggest that yield under stress can be improved by
selecting for early silk emergence. The correlation coefficients reported in this study under
stress conditions for anthesis days were -0.20 to -0.32 and for anthesis silking interval was -
0.39 to -0.54. Again these values are in agreement with findings of Zaidi et al. (2004) of -
0.22 to -0.56. Banziger et al. (1997) reported a correlation of -0.47 for anthesis silking
interval with grain yield under low N, yet in this study a correlation of -0.54 was reported.
However, Gissa (2008) reported a highly significant negative correlation between grain yield
and anthesis silking interval of -0.60. A highly significant and positive correlation of ears per
plant and grain yield reported in this study is consistent with findings by Banziger et al.
(1997) and Gissa (2008). These results show that an increase in ears per plant results in a
corresponding increase in grain yield under optimum, drought and low N conditions.
Anthesis-silking interval and ears per plant are considered important secondary traits used in
selection of drought tolerant materials (Bolanos and Edmeades, 1996; Vasal et al., 1997;
Banziger and Lafitte, 1997; Edmeades et al., 1997; Banziger et al., 2000, Mhike et al.,
2011b).

90
A number of investigators have reported variability in the performance of maize under low N
conditions (Banziger et al., 1997; 2000; Betran et al., 2003; Gissa, 2008; Mhike et al.,
2011a). The yield performance of hybrids in this study was 15% of the grain yield under
optimum conditions, however this was lower than what Banziger et al. (1997) recommended.
They reported that grain yield under low N stress should be between 25 and 35% of the
average yield under optimum conditions. Betran et al. (2003) reported values as high as 65%.
However some scientists indicated that varied magnitudes of grain loss can be expected
under low N conditions (Banziger et al., 1999; Smallberger and Toit, 2004; Monneveux et
al., 2005). Low yields under low N reported in the study might have been due to mid-season
dry spells that were experienced at the site.

GCA effects are associated with additive gene action while SCA effects are associated with
non-additive gene action. Significant GCA and SCA effects for traits such as grain yield,
anthesis days, anthesis silking interval, plant height, senescence and GLS across all
environments suggest the importance of both additive and non-additive gene action in the
inheritance of these traits. Gissa (2008) reported similar findings. GCA effects, however,
were more important than SCA effects. Betran et al. (2003) reported that additive genetic
effects across environments accounted for 61% of total genetic variation in grain yield and
they assumed importance over non-additive variances. GCA variances that are higher than
SCA variances indicate that additive genetic effects are more important (Gethi and Smith,
2004). This has an implication in breeding in that good parents can be identified using the
GCA effects and then crossed to produce high yielding hybrids. Early testing of inbred lines
becomes more effective and good hybrids can be identified in the early stages of breeding
using GCA effects (Melchinger et al., 1998). A high contribution of SCA to grain yield,
anthesis silking interval and ears per plant under stress environments in this study suggested
that non-additive gene action assumed a crucial role in the expression of these traits under
these conditions. These findings are contrary to the findings by Gissa (2008), where GCA
assumed a more important role in most of the traits under low N conditions. Expression of
grain yield, anthesis silking interval and ears per plant under stress environments being
controlled by non-additive genes in this study has an implication in breeding in that good
parents cannot be identified using these traits, instead good specific combiners are the ones

91
that can be selected for. Long et al. (2004) found both GCA and SCA effects to be significant
for grain yield but SCA effects were more important than GCA effects. The inbred lines used
in this study can therefore be exploited for grain yield under drought conditions through
targeting the SCA effects. In this study anthesis silking interval was mainly controlled by
non-additive gene effects.

GCAm and GCAf contributed differently to the expression of different traits in this study.
Derera et al. (2008) reported different contributions of GCAm and GCAf for grain yield under
drought conditions (44 versus 32%) and similar contributions under non-drought
environments (29 vs 31%). Results in this study are contrary to these findings. In this study
GCAf and GCAm for grain yield contributed similarly (22%) under drought conditions,
indicating that both parents made similar contributions to grain yield in hybrids under
drought. However, under non-drought environments GCAf dominated GCAm (53 vs 28%).
GCAm contributed more to anthesis days under optimum, drought and low N conditions,
indicating that inbred lines can be used in selecting for earliness or lateness in the maturity of
hybrids.

In this study GCAf mean squares for grain yield across all environments, under low N
conditions and under optimum conditions were higher than GCAm mean squares, suggesting
the importance of maternal effects in the expression of grain yield under these conditions.
The expression of grain yield under drought conditions was also affected by maternal effects.
Similar results were reported by Derera et al. (2008). The maternal effects were also
important in the expression of anthesis silking interval under low N and optimum conditions
and the paternal effects assumed an important role in the expression of ear height across all
trials. Senescence, an important secondary trait under stress conditions, was also influenced
by maternal effects. Prolificacy (ears per plant) under drought and low N conditions was
influenced by maternal effects. Khehra and Bhalla (1976) studied reciprocal dissimilarities
under optimum environments and reported that cytoplasmic effects were not significant for
grain yield, which is consistent with the observations in the current study. Largely maternal
effects, if unrestrained, would inflate GCA variance for yield and secondary traits; and as a
result heritability is overvalued which might deceive breeders in implementing a wrong

92
selection strategy (Derera et al., 2008). GLS and leaf blight turcicum diseases of economic
importance, especially under optimum conditions, were influenced by maternal effects,
whilst rust and ear rots were influenced by paternal effects.

GCAf x E, GCAm x E and SCA x E were significant for most traits in across site analysis in
this study and results were consistent with the findings by other scientists (Derera et al.,
2008; Gissa, 2008; Machida, 2008; Mhike et al., 2011a). It appears that G x E effects would
present challenges in the breeding of materials for different environments. Significant G x E
interactions highlights the need to use several environments in the estimation of genetic
effects. Crossover type of G x E was observed in the current study. Genetic component
estimates based on data from single environments were found to be unreliable. Results show
that GCA effects associated with the parents and SCA effects associated with crosses were
not consistent over environments. Banziger et al. (2000) reported that stress environments
produce high G x E interactions.

The lines and testers exhibited different GCA effects for different traits in this study. Line 3
(RS61P) showed consistency in its performance exhibiting good GCA effects for grain yield
across environments (0.72), optimum conditions (0.64), drought conditions (0.37) and under
low N conditions (0.58). This line is therefore a good general combiner for grain yield across
all environments and this was further confirmed by four of its hybrids that appeared among
the 10 best performing hybrids across all environments. This line was also a good general
combiner for other traits such as anthesis days, anthesis silking interval, plant height, root
and stalk lodging and senescence. A combination of a negative GCA effect for anthesis days
and anthesis silking interval would be good in the early maturing maize breeding programme.
On the other hand testers 9 (CML444) and 11 (CML548) were consistent in their GCA
effects, again proving to be good general combiners for grain yield and other agronomic
traits. The other lines and testers that showed good and consistent GCA effects were
NAW5885, CML539, CML537 and CML442.

The presence of significant SCA is a consequence of fluctuations in dominance relationships

among parents (Wassimi et al., 1986). Line 3 (RS61P) was a good specific combiner as

93
indicated by significant positive SCA effects for grain yield in different environments.
Hybrids RS61P/ CML444 and 2N3d/CML548 were among the best performing hybrids
across all environments, optimum conditions and 2N3d/CML548 was the overall best
performer under drought conditions with a mean yield of 3.26 t ha-1. However RS61P did not
combine well with other testers. Results from previous studies as well as the current study
indicate that a parent having a good GCA effect does not automatically produce better
hybrids all the time (Tyagi and Lal, 2005). A parent with poor GCA might produce better
hybrids (Tyagi and Lal, 2005) and this agrees with some of the findings in this study where
poor general combiners produced good hybrids with the testers. For example SC5522 had
poor GCA for grain yield but it produced hybrids with good SCA with CML395 (0.77) and
CML536 (0.75). The results suggest that SC5522 displayed a dominance effect whereby it
contributed non-additive genes towards expression of grain yield. Thus the role of dominance
in conferring heterosis was displayed by the results. This has an implication in breeding in
that lines should be selected based on both GCA and SCA effects. If GCA is insignificant it
is advisable to select lines based only on SCA (Narro et al., 2003).

Variance components and heritability estimates have been extensively used by plant breeders
in selection of promising genotypes and prediction of desirable traits (Morakinyo, 1996). The
relatively high heritability estimates in this study are an indication that the studied traits are
mainly controlled by additive genes. Hallauer and Miranda (1981) reported heritability
estimates that were lower in magnitude than the ones found in the current study. Mhike et al.
(2011a) reported heritability estimates that were above 50% for anthesis days, ear height,
anthesis silking interval and ear position and the rest were below 50%. A heritability estimate
of 68% for grain yield was reported in this study and this was higher than values reported by
Bolanos and Edmeades (1996) (60% under optimal and 40% under drought conditions) and
Mhike et al. (2011a) (21%). However, the magnitudes of heritability estimates are products
of the population being tested, environments within which the testing is done and traits being
measured (Falconer and Mackay, 1996). To this end therefore the differences in magnitudes
observed here is a manifestation of the differences in these three determinants of the
heritability estimates. It should, therefore, be understood that heritability values reported for

94
a given trait, are specific to a particular population under particular conditions (Hallauer and
Miranda, 1981).

3.5 Conclusions
Abiotic stress factors such as drought and low N, are among the critical factors affecting
maize production in Zimbabwe. Results from this study revealed that there is a high level of
genetic variability and there is a possibility of selecting good hybrids for grain yield and
other agronomic traits under both drought and low N conditions. The NCDII was effective
and ideal in the identification of lines with good GCA effects and potential single cross
testers. The lines identified as good potential parental lines in the stress breeding programme
include RS61P, NAW5885, CML444, CML539, CML442, CML537 and CML548. On the
other hand the single crosses RS61P/CML444 (SC) and 2N3d/CML548 (N) were identified
as the highest potential single cross testers, however further studies have to be done to
confirm validity of these testers. However RS61P performed best under all conditions. GCA
and SCA effects were significant for most traits across environments and this showed the
significance of both additive and non-additive genes in the expression of these traits. GCA
variances were, however, larger than SCA variances for the majority of traits, resulting in
high heritability estimates. The parental materials used in this study can therefore be used in
selecting good parents for future use in the development of stress tolerant genotypes as most
traits are controlled by additive genes. However, a higher SCA contribution to grain yield
under drought and ears per plant (prolificacy) and anthesis silking interval under both
drought and low N is an indication that non-additive gene action was important in the
expression of these traits. Under drought conditions specific hybrids with high mean yields
and high prolificacy can be selected. Maternal effects were important in modification of grain
yield and other traits in all the environments.

95
3.6 References
African Press Agency (APA), International Maize and Wheat Improvement Center. 2007.
Enhanced conditions: Drought tolerant maize to give African farmers options even
with global warming. Science/Daily, January 30.
Allard, R.W. 1999. Principles of Plant Breeding. 2nd edition. John Wiley & Sons, New York.
Banziger, M. and H.R. Lafitte. 1997. Efficiency of secondary traits for improving maize for
low-nitrogen target environments. Crop Science 37: 1110-1117.
Banziger, M., G.O. Edmeades and H.R. Lafitte. 1999. Selection for drought tolerance
increases maize yields across a range of nitrogen levels. Crop Science 39: 1035-1040.
Banziger, M., G.O. Edmeades and S. Quarrie. 1997. Drought stress at seedling stage: Are
there genetic solutions? In: Edmeades, G.O., M. Banziger, H.R. Mickelson and C.B.
Pena-Valvidia (eds) Developing Drought and Low N Tolerant Maize. Proceedings of
a Symposium, March 25-29, 1997, CIMMYT, El Batan, Mexico, D.F: CIMMYT. pp
348-354.
Banziger, M., G.O. Edmeades, D. Beck and M. Bellon. 2000. Breeding for drought and
nitrogen stress tolerance in maize. From theory to practise. Mexico, D.F: CIMMYT.
pp 23-31.
Betran, F.J., J.M. Ribaut, D. Beck and D. Gonzalez de Leon. 2003. Genetic diversity, specific
combining ability and heterosis in tropical maize under stress and non-stress
environments. Crop Science 43: 797-806.
Bolanos, J., and G.O. Edmeades. 1996. The importance of the anthesis-silking interval in
breeding for drought tolerance in tropical maize. Field Crops Research 48: 65-80.
Camphos, H., M. Cooper, G.O. Edmeades, C. Loffer, J.R. Schussler and M. Ibanez. 2006.
Changes in drought tolerance in maize associated with fifty years of breeding for
yield in the US Corn Belt. Maydica 51: 369-381.
Derera, J. 2005. Genetic effects and associations between grain yield potential, stress
tolerance and yield stability in Southern African maize (Zea mays L.) base
germplasm. PhD thesis. ACCI. Faculty of Science and Agriculture. University of
KwaZulu-Natal. South Africa.

96
Derera, J., P. Tongoona, B.S. Vivek and M.D. Laing. 2008. Gene action controlling grain
yield and secondary traits in southern African maize hybrids under drought and non-
drought conditions. Euphytica 162: 411-422.
Devi, P. and N.K. Singh. 2011. Heterosis, molecular diversity, combining ability and their
interrelationships in short duration maize (Zea mays L.) across the environments.
Euphtica 178: 71-81.
Dhliwayo, T., K. Pixley, A. Menkir and M. Warbuton. 2009. Combining ability, genetic
distances, and heterosis among elite CIMMYT and IITA tropical maize inbred lines.
Crop Science 49: 1201-1210.
Edmeades, G.O., J. Bolanos, M. Banziger, S.C. Chapman, A. Ortega, H.R. Latiffe, K.S.
Fischer and S. Pandey. 1997. Recurrent selection under managed drought stress
improves grain yields in tropical maize. In: Edmeades, G.O., M, Banziger, H.R.
Mickelson and C.B. Pena-Valdivia (eds) Developing Drought and Low N-Tolerant
Maize. Proceedings of a Symposium, March 25-29, 1997, CIMMYT, El Batan,
Mexico, D.F: CIMMYT. pp 415-425.
Edmeades, G.O., J. Bolanos, M. Hernandez and S. Bello. 1993. Causes for silk delay in a
lowland tropical maize population. Crop Science 33: 1029-1035.
Falconer, D.S. and T.F.C. Mackay. 1996. Introduction to Quantitative Genetics. 4th edition.
Longman. Essex. England.
George, M.L.C., F. Salazar, M. Warbuton, L. Narro and F.A. Vallejo. 2011. Genetic distance
and hybrid value in tropical maize under P stress and non-stress conditions in acid
soils. Euphytica 178: 99-109.
Gethi, J.G. and M.E. Smith. 2004. Genetic responses of single crosses of maize to Striga
hermonthica (DeL.) Benth. and Striga asiatica (L.) Kuntze. Crop Science 44: 2068-
2077.
Gissa, D.W. 2008. Genotypic variability and combining ability of quality protein maize
inbred lines under stress and optimal conditions. PhD Thesis. Department of Plant
Sciences, Faculty of Natural and Agricultural Sciences, University of the Free State,
South Africa.
Hallauer, A.R. and J.B. Miranda. 1981. Quantitative Genetics in Maize Breeding. 1st Edition.
Lowa State University Press. Ames Lowa. pp 267-292.

97
Khehra, A.S. and S.K. Bhalla. 1976. Cytoplasmic effects on quantitative characters in maize
(Zea mays L.). Theoretical and Applied Genetics 47: 271-274.
Long, J.K., M. Banziger and M.E. Smith. 2004. Diallel analysis of grain iron and zinc density
in Southern African-adapted maize inbreds. Crop Science 44: 2019-2026.

Machida, L. 2008. Quantitative genetic analysis of agronomic and kernel endosperm traits in
quality protein maize (QPM) and investigations of the putative nutritional value of
contaminated QPM crops. PhD thesis. ACCI. Faculty of Science and Agriculture.
University of KwaZulu-Natal. South Africa.
Melchinger, A., W. Schmit and H.H. Geiger. 1998. Comparison of testcrosses from F2 and
first backcross populations in maize. Crop Science 28: 743-749.
Mhike, X., D.M. Lungu and B. Vivek. 2011a Combining ability studies amongst AREX and
CIMMYT maize (Zea mays L.) inbred lines under stress and non-stress conditions.
African Journal of Agricultural Research 6: 1952-1957.
Mhike, X., P. Okori, C. Magorokosho and T. Ndhlela. 2011b. Validation of the use of
secondary traits and selection indices for drought tolerance in tropical maize (Zea
mays L.) African Journal of Plant Science 5: 96-102.
Monneveux, P., P.H. Zaidi and C. Sanchez. 2005. Population density and low nitrogen
affects yield-associated traits in tropical maize. Crop Science 45: 535-545.
Morakinyo, J.A. 1996. Heritability, correlation and expected responses to selection of some
yield components in grain sorghum, Sorghum bicolor L. Moench. Nigerian Journal of
Genetics 11: 48-54.
Mungoma, C. and L.M. Pollak. 1988. Heterotic patterns among ten Corn Belt and exotic
maize populations. Crop Science 28: 500-504.
Narro, L., S. Pandey, J. Crossa, C. De Leon and F. Salazar. 2003. Using line x tester
interaction for the formation of yellow maize synthetics tolerant to acid soils. Crop
Science 43: 1718-1728.
Pswarayi, A. and B.S. Vivek. 2008. Combining ability amongst CIMMYT’s early maturing
maize (Zea mays L.) germplasm under stress and non-stress conditions and
identification of testers. Euphytica 162: 353-362.

98
Rukuni, M., P. Tawonezvi and C. Eicher. 2006. Zimbabwe’s Agricultural Revolution
Revisited. University of Zimbabwe Publications. Harare. pp 119-140.
SAS Istitute. 2002. 2002-2008 by SAS Institute Inc., Cary, NC. USA.
Singh, R.K. and B.D. Chaudhary. 1977. Biometrical Methods in Quantitative Genetic
Analysis. Kalyani Publishers. Ludhiana.
Smallberger, S. and A.S. du Toit. 2004. Identification of maize cultivars tolerant to low soil
fertility in South Africa. In: Friesen, D.K. and A.F.E. Palmer (eds) Integrated
Approaches to Higher Maize Productivity in the New Millennium. Proceedings of the
7th Eastern and Southern Africa Regional Maize Conference. 5-11 February 2002,
CIMMYT/KARI, Nairobi, Kenya. pp 202-205.
Smith, M., R. Ericson, S. Norman, K. Payne and J. Singer. 2009. What’s Cropping Up: A
newsletter for New York Field Crops & Soils 19: 1-3.
SPSS 15.0 for Windows. 2006. www.spss.com SPSS Inc., Chicago III.
Stevens. R. 2008. Review: Prospects for using marker assisted breeding to improve maize
production in Africa. Journal of the Science of Food and Agriculture 88: 745-755.
Tyagi, A.P. and P. Lal. 2005. Line x tester analysis in sugar cane (Saccharum officinarum).
South Pacific Journal of Natural Science 23: 30-36.
Vasal, S.K., G Srinivasan, D.L Beck, J. Crossa, S. Pandey and C. De Leon. 1992. Heterosis
and combining ability of CIMMYT’s tropical late white maize germplasm. Maydica
37: 217-223.
Vasal, S.K., H. Cordova, D.L. Beck and G.O. Edmeades. 1997. Choices among breeding
procedures and strategies for developing stress tolerant maize germplasm. In:
Edmeades, G.O., M. Banziger, H.R. Michelson and C.B. Pena-Valdiva (eds)
Developing Drought and Low N-Tolerant Maize. Proceedings of a Symposium. 25-29
March 1997, Mexico, D.F: Mexico. pp 336-347.
Wang, P., W. Takabayashi, A. Endo, T. Shikanai, Y. Ji-Yu and M. Hualing. 2006.
Chloroplastic NAD(P)H dehydrogenase in tobacco leaves functions in alleviation of
oxidative damage caused by emperaure stress. Plant Physiology 141: 465-474.
Wassimi, N.N., T.G. Isleib and G.L. Hosfield. 1986. Fixed effect genetic analysis of a diallel
cross in dry beans (Phaseolus vulgaris L.). Theoretical and Applied Genetics 72: 449-
454.

99
Zaidi, P.H., G. Srinivasan, H.S. Cordova and C. Sanchez. 2004. Gains from improvement for
mid season drought tolerance in tropical maize (Zea mays L.) Field Crops Research
89: 135-152.

100
 CHAPTER 4

Genotype by environment interaction and stability analysis for grain yield

of single cross hybrids

Abstract
Maize is the most important cereal crop in Zimbabwe and is grown by both large and small
scale farmers who are located in different agro-ecological zones of the country. The
development and dissemination of adapted and high yielding maize varieties to these agro-
ecological zones involves conducting multi environment trials (METs). This study was
conducted with the objective of assessing genotype by environment (G x E) interaction and
stability of single cross hybrids grown in stress and non-stress environments. Yield data of 80
maize single cross hybrids tested across seven environments during the 2009/10 and 2010/11
seasons were analysed using AMMI and GGE biplot methods. However, the analysis was
later narrowed to 20 hybrids for better graphical visualisation and better conclusions. In the
combined ANOVA the environment (E) explained 85% of the total (G + E + G x E)
variation, whereas genotype (G) and G x E interaction captured 7.0% and 13.1%,
respectively. The largest proportion of the total variation was explained by environmental
effects as a result of inclusion of environments with varying stress conditions. Higher
yielding and adapted genotypes were determined by PC1 scores >0, whilst non-adapted and
lower yielding genotypes were determined by PC1 scores <0. PC2 scores of approximately
zero identified stable genotypes, whilst larger PC2 scores detected unstable genotypes. Good
hybrids had high PC1 scores (high yield) and small absolute PC2 scores (high stability). On
the other hand the environmental PC1 scores were related to non-crossover G x E interaction,
whilst PC2 scores were related to the crossover type. The two methods detected similar high
yielding and stable genotypes. The higher yielding and stable genotypes were G52
(CML548/2N3d-B) and G57 (CML444-BB/RS61P-B). Agricultural Research Trust farm was
the most powerful site in discriminating genotypes and the most representative environment.
Results showed that there were three mega environments within the test environments, which
can be utilised for multi-environment yield trials.

101
4.1 Introduction

The most important cereal crop in Africa is maize and in Zimbabwe, maize production
accounts for 80% of the total cereal production. It is grown by both large and small scale
farmers for both food and feed in different agro-ecological zones. Newly improved hybrid
maize cultivar candidates from different breeding programmes must be evaluated at many
sites and for a number of years before being recommended to be grown in given locations.
Evaluation of genotypic performance of hybrid maize cultivar candidates in many
environments generates valuable data to ascertain how stable and adapted genotypes are
(Crossa, 1990). The multi-location evaluation, however, result in G x E interactions that
often complicate the interpretation of results obtained and reduce efficiency in selecting the
best genotypes (Annicchiarico and Perenzin, 1994). The existence of G x E interaction may
mean that a superior variety in one location is not necessarily the best in another
environment. Kang et al. (1991) indicated that selection based on yield only may not always
be adequate when G x E is significant. The analysis of G x E interaction, thus, turns out to be
a significant strategy used by breeders for assessing varieties for adaptation and also for
selecting parents for base populations (Aina et al., 2007).

Plant breeders usually look for non-crossover G x E or rather the absence of G x E

interaction when selecting genotypes for general adaptation and crossover G x E interaction
for specific adaptation (Matus-Cadiz et al., 2003). ANOVA only defines if G x E is
significant or not, but it does not provide information as to which varieties or locations
contribute to the interaction (Samonte et al., 2005). Various approaches have been
established to reveal forms of G x E interaction and these include joint regression developed
by Finlay and Wilkinson (1963), Eberhart and Russel (1966) and Perkins and Jinks (1968),
AMMI developed by Gauch (1992) and GGE developed by Yan et al. (2000).

It is usually challenging to define how genotypes respond, without graphically presenting the
data, when many varieties are evaluated across many locations, seasons and years (Yan et al.,
2001). There are two types of biplots that have been extensively used to visualise G x E
interactions and these are the AMMI and GGE biplots (Gauch, 1988; Gauch and Zobel,
1997; Yan et al., 2000; Ma et al., 2004). Yan and Kang (2003) postulated that a GGE biplot

102
is an effective tool for mega environment analysis (“which-won-where”). GGE biplots have
also been found to be effective in depicting genotype mean performance and stability and the
ability of environments to distinguish genotypes in target locations. In the AMMI model,
ANOVA for varieties and environment main effects is combined with PCA of the G x E
interactions (Gauch and Zobel, 1996). Dissimilarities among the two approaches are that
GGE biplot analysis is founded on location centred PCA, whilst AMMI analysis is denoted
as double centred PCA. However it is not always easy to visualise “which-won-where” in
the AMMI graph, particularly when a number of varieties and locations are involved (Ebdon
and Gauch, 2002) and at times it could be deceptive (Yan et al., 2007).

Hence the AMMI graph is regarded as a better tool for presenting conclusions rather than as
a tool for determining “which-won-where”. In this study 80 genotypes were evaluated across
seven environments in the 2009/10 and 2010/11 seasons. Given the number of genotypes
used and number of environments there should be better a presentation of the G x E
interaction if both GGE and AMMI biplots are used and this should assist in reaching better
conclusions. The objectives of this study were therefore to i) analyse G x E interaction and
stability of single cross hybrids generated using CIMMYT elite drought tolerant lines and
DR&SS elite drought susceptible lines for grain yield across stress and non-stress
environments and ii) to observe the pattern of grouping of environments based on grain yield
responses of hybrids.

4.2 Materials and methods

4.2.1 Germplasm
Seventy two experimental single cross hybrids and eight check single cross hybrids were
evaluated in a total of 14 environments in the 2009/10 and 2010/11 seasons. The
experimental hybrids constituted of cross progeny of 10 DR&SS and 13 CIMMYT elite
inbred lines and their single cross hybrids. Details of the germplasm used in the study are
given in section 3.2.1. A total of 72 single cross hybrids out of a possible 130 hybrids were
used in the study because they were the successful ones and had enough seed for multi-
location trials.

103
4.2.2 Sites

The site details are given in the materials and methods section of Chapter 3 (section 3.2.2).
Details of annual average rainfall and soil type of these sites are shown in Table 4.1.

Table 4.1 Site annual average rainfall and soil type

Annual Latitude, longitude
Environment
Site Management average Soil type and altitude
code
rainfall (mm)
Rhodustalf greater group with 17°26'S, 31.5°E,
ART Farm E1 Optimum 891
texture code ICG 1 480 masl
Rhodustalf greater group with 17.13°S, 31°E, 1 406
Harare E2 Low N 820
texture code ICG masl
Haplustox code FRr14 and texture 18.32°S, 30.90°E,
Kadoma E3 Optimum 727
code DCE 1 155 masl
Haplustalf code LXh7 and texture 21.02°S, 31.58°E,
Chiredzi winter E4 Managed drought 450
code ICH 433 masl
Rattray Arnold E5 Optimum 820 Clay loam 17°40'S, 1 308 masl

Haplustalf code LXh7 and texture 21.02°S, 31.58°E,

Chiredzi E6 Random drought 450
code ICH 433 masl
Chisumbanje E7 Random drought 420 Deep vertisols
20°S, 33°E, 455 masl
ART=Agricultural Research Trust; Low N= low nitrogen.

4.2.3 Experimental design and data collected

Trials were planted in one row plots with 0.75 m inter-row spacing and 0.25 m in-row
spacing using a 0.1 alpha lattice design. Grain yield and other agronomic data such as
anthesis days, anthesis silking interval, plant height, ear height, root lodging, stem lodging,
ears per plant and senescence were collected during the growth period of the crop as
described in Table 3.2. Scores for diseases of economic importance in maize in Zimbabwe
were also taken and these include MSV, GLS, Exhohilium turcicum and Puccinia sorghi.

4.2.4 Statistical analysis

The data was first subjected to ANOVA in order to determine the effects of G x E interaction
for grain yield. ANOVA performed on plot basis and pooled over locations and seasons
using AGROBASE version II (2005). Broad sense heritability estimates were calculated for
the two seasons as described in section 3.2.5.

104
The AMMI statistical model in GenStat 14th Edition (2011) was used to analyse the yield
data.
The following AMMI model was used:
Yger = µ + αg + βe + ∑λn Ygn δen + ρge + Eger
Where:
Yger = Yield of genotype g in environment e for replicate r
µ = Grand mean
αg = Genotype mean deviations (genotype means minus grand mean)
βe = Environment mean deviation
n = Number of PCA axes retained in the model
λn = Singular value for PCA axis n
Ygn = Genotype eigenvector values for PCA axis n
δen = Environment eigenvector values for PCA axis n
ρge = Residuals
Eger = Error term

GGE biplot analysis was conducted using the “GGE biplot” software (Yan and Tinker,
2005). The model for a GGE biplot (Yan, 2002) based on SVD of the first two principal
components was used:
Yij – µ – βj = λ1ξi1ηj1 + εij

Where Yij = the measured mean (DBH) of genotype i in environment j

Μ = grand mean
βj = main effect of environment j,
µ + βj = the mean yield across all genotypes in environment j
λ1 and λ2 = singular values (SV) for the first and second principal component (PC1 and
PC2) respectively
ξi1 and ξi2 = eigenvectors of genotype I for PC1 and PC2 respectively
ηj1 and ηj2 = eigenvectors of environment j for PC1 and PC2 respectively
εij = residual associated with genotype i in environment j.

105
Spearman’s rank correlation (r) amongst environments was calculated from means across
environments and statistical calculations were executed with SPSS 15.0 version for Windows
(2006). Hierarchical cluster analysis of environments was done in NTSYSpc 2.21n software
(Rohlf, 1993) using trait means.

4.3 Results

4.3.1 Analysis of variance within years and across years

Results for ANOVA for within years and across years are presented in Tables 4.2 and 4.3. In
both the 2009/10 and 2010/11 seasons the environments, genotypes and G x E were highly
significant (P≤0.001). The broad sense heritability values for yield for the years were 65%
(2009/10) and 46% (2010/11) and the r values were 0.90 and 0.83 respectively.

Table 4.2 Analysis of variance for grain yield across environments in the 2009/10 and
2010/11 seasons
2009/10 2010/11
Source DF MS MS
Environments 6 1172.27*** 216.54***
Genotype 79 8.74*** 1.79***
GxE 474 3.03*** 0.97**
Residual 553 1.66 0.78
LSD 0.96 0.65
CV 28.2 34.2
R2 0.90 0.83
Herit. 0.65 0.46
***P≤0.001; **P≤0.01; G x E=genotype by environment interaction; LSD=Least significant difference; CV=coefficient of variation;
Herit=heritability; R2=coefficient of determination; DF=degrees of freedom; MS=mean square.

The combined ANOVA indicated that maize grain yields were significantly affected by
environment, which explained 85% of the total (G + E +G x E) variation, whilst genotype
and G x E explained 7.0% and 13.1% of variation respectively (Table 4.3). The mean squares
for years, E x Y, G x Y and G x E x Y were all highly significant.

106
Table 4.3 Combined analysis of variance for grain yield of 80 genotypes across seven
environments
Source DF SS MS
Enviroments 6 6073.676 1012.27***
Genotype 79 535.531 6.779***
Year 1 2237.381 2237.38***
GxE 474 1000.657 2.11***
ExY 6 2259.167 376.53***
GxY 79 295.793 3.74***
GxExY 474 893.205 1.88***
Residual 1114 1346.403 1.21
Total 2239 14893.493
CV 30.75
R2 0.91
***P≤0.001; G x E=genotype by environment interaction; E x Y=environment by year interaction; G x Y=genotype by year
interaction; G x E x Y=genotype by environment by year interaction; CV=coefficient of variation; R2=coefficient of determination;
DF=degrees of freedom; SS=sums of squares; MS=mean squares.

4.3.2 Additive main effect and multiplicative interaction analysis

The partitioning of G x E through AMMI analysis showed that PC1 and PC2 were significant
(P≤0.001) (Table 4.4), explaining 48.27% and 27.82% of the total variation and together they
explained 76.1% of G x E interaction (Table 4.5). The mean yields of genotypes in all
environments are presented in Appendix 8. The highest yielding genotypes for each
environment were as follows: G74 in Agricultural Research Trust farm, Harare low N and
Rattray Arnold Research Station, G61 in Kadoma, G52 in Chiredzi winter, G57 in Chiredzi
summer and G53 in Chisumbanje. In the AMMI biplot genotypes were clustered around the
zero point except for a few genotypes that were distinct (Figure 4.1). G74 proved that it was
high yielding but did not show good stability, whilst G59, G10 and G63 were high yielding
and more stable. Genotypes G29, G17, G73 and G12 were low yielding and more stable. G74
was mostly associated with high yielding environments.

Genotype means were plotted against the IPCA scores across all environments and G74 was
the highest yielder but very unstable, whilst G60 and G61 were high yielding and more stable
(Figure 4.2). G57 was also high yielding and relatively stable. G68, G4, G52, G5 and G63
were high yielding but very unstable. G73 was low yielding but very stable. Environments
Agricultural Research Trust farm (optimum), Chisumbanje (random drought), Rattray Arnold
Research Station (optimum) and Kadoma (optimum) were higher yielding; whilst

107
environments Harare low N, Chiredzi winter (managed drought) and Chiredzi (random
drought) were lower yielding (Figure 4.3).

4.3.3 Genotype and genotype by environment interaction biplot for all 80 genotypes
Positioning of genotypes on the GGE biplot are presented in Figure 4.4. The GGE biplot for
all 80 genotypes explained 61.7% of the genotype main effect and the G x E interaction
(Figure 4.4). Primary (PC1) and secondary (PC2) scores were significant and explained
47.5% and 14.2% of the genotype main effect and G x E interaction respectively. A moderate
percentage variability of GGE (61.7%) was explained by the biplot and this suggests some
strong and complex G x E interaction in the MET data. Genotypes G61, G52, G4, G57 and
G74 were mostly associated with Kadoma and Harare low N.

Table 4.4 Analysis of variance for additive main effects and multiplicative interaction
model for grain yield across seven environments for the 2009/10 and 2010/11
seasons
Source DF MS
Environments 13 912.61***
Genotype 79 13.80***
GxE 891 2.44***
IPCA1 91 9.59***
IPCA2 89 5.65***
IPCA3 87 2.99***
IPCA4 85 2.02**
IPCA error 539 0.68
Residual 972 1.19
***P≤0.001; **P≤0.01; IPCA=interaction principal components axes; G x E=genotype by environment interaction; DF=degrees of
freedom; MS=mean squares.

108
Table 4.5 Additive main effects and multiplicative interaction analysis of yield data of
80 maize genotypes tested across seven environments in the 2009/10 and
2010/11 seasons

IPCA Axis Eigenvalue % G x E explained Cumulative %

1 436.183 48.27 48.27

2 251.411 27.82 76.1

3 130.352 14.43 90.52

4 85.652 9.48 100

IPCA=interaction principal component axes; G x E=genotype by environment interaction.

G38, G59 and G27 were associated with Chiredzi, whilst G67 and G60 were associated with
Chisumbanje and Chiredzi winter. G74 was once again the highest yielding genotype. G48
was more closely associated with environments Rattray Arnold Research Station and
Agricultural Research Trust farm. Graphical presentation of stability of genotypes using
GGE biplot analysis is presented in Figure 4.5. Genotypes G5, G26, G62, G6, G57, G74 and
G45 were high yielding as well as stable since their absolute PC2 scores were near zero,
whereas genotypes G67, G60, G51, G7 and G68 were high yielding and unstable as they had
larger absolute PC2 scores (Figure 4.5). G74 was the highest yielding genotype (large PC1
score) but unstable in different environments (large PC2 score) (Figure 4.5). G17 and G37
were the poorest performing genotypes (low PC1 scores), whereas G73 and G21 were
amongst the poorest performing genotypes but very stable (near zero PC2 score). G61 and
G52 were very stable and their average yields were larger than all other genotypes except
G74. G74 is a commercial long season hybrid which was included in the trials as a control.

109
 2.0
E3

1.5

1.0
scores
res

G9
E4 G60
0.5 G38
IPCA sco

G25 G35
G50 G28G34
G42 G77
IPCA1

G33
G24G30
G19
G44 G51
G27 G4
G72
G29 G43
G40 G76
G36
G16 G5G45
G79
G49 G8 G61
E2 G12 E6G23
G37 G14 G1
G2
G7
G47 G52
G32
G66 G58 G57
0.0 G17 G73 G21 G59
G3G31 G26
G67
G69 G80 G10
G39G13 G46 G62 G63
G11G6
G71
G55
G53G65
G41
G56 E5G70
G78 G68
G15 G48
-0.5 Lower yielding G18 G75 G22
G74
Higher yielding
G54
G64 G20

E1
-1.0

E7

2 3 4 5 6 7

Genotype and environment means

Genotype & Environment means
Figure 4.1 Additive main effects and multiplicative interaction biplot for genotype
grain yield in seven environments for two seasons combined.
Genotypes are labelled with a G followed by a number; environments are labelled with an E followed by a number.
E1=Agricultural Research Trust farm; E2=Harare low nitrogen; E3=Kadoma; E4=Chiredzi winter; E5=Rattray Arnold Research
Station; E6=Chiredzi summer; E7=Chisumbanje.

G29
1. 0
G17

G55
G30 G20
G22
G37
G12 G19
G25 G35
G21
0.5 G13
G33 G63
G23
G18
G24 G36
G50 G41 G15 G59
G56
G14G2G65
G42 G46 G54
scores

G34
IPCAscores

G32G78
G66 G67 G72
G58
0.0 G75 G60
G73 G69 G16 G31
G44 G8
G62
G38
G61
G1 G47G10
G53G9
G3
G80
G6 G70
G64 G11G28
G27
IPCA1

G71 G57
G45
G77 G51
G40 G76 G26
-0.5
G39 G49 G79
G48
G7
Low yielding G5 G52 High yielding
G43 G4
-1.0
G68

G74
-1.5

2.5 3.0 3.5 4.0 4.5 5.0 5.5

Genotypemeans
Genotype means

Figure 4.2 Additive main effect and multiplicative interaction biplot for genotype
grain yield across environments across two seasons.

110
 2. 0
E3

1. 5

1. 0
IPCA1 scores
Ascores

E4
0. 5
IPC

E6
0. 0 E2

E5
-0.5

E1
-1.0

E7

2 3 4 5 6 7

Environment means
Environment means

Figure 4.3 Additive main effects and multiplicative interaction biplot for
environment means across two seasons.
E1=Agricultural Research Trust farm; E2=Harare low nitrogen; E3=Kadoma; E4=Chiredzi winter; E5=Rattray Arnold
Research Station; E6=Chiredzi summer; E7=Chisumbanje.
Unstable

Stable
PC2

ART
Low yielding
Adapted/high yielding
Non-adapted

PC1

Figure 4.4 Genotype and genotype by environment interaction biplot analysis of

yield across seven environments and two seasons.
CH WINTER=Chiredzi winter; CH=Chiredzi summer; CS=Chisumbanje; HRE Low N=Harare low nitrogen; RARS=Rattray
Arnold Research Station; ART=Agricultural Research Trust farm.

111
 PC2

ART
RARS

PC1

Figure 4.5 Yield stability and performance of genotypes for seven environments and
two seasons.
CH WINTER=Chiredzi winter; CH=Chiredzi Summer; CS=Chisumbanje; HRE Low N=Harare low nitrogen; RARS=Rattray
Arnold Research Station; ART=Agricultural Research Trust farm.

Graphical presentation of “which-won-where” was used to assess which genotypes

performed well in which environments. The polygon view of a GGE biplot clearly displayed
the “which-won-where” pattern; henceforth it is a clear summary of the G x E pattern of a
multi-environment yield trial (MEYT) data set. The polygon was formed from genotype
markers G74, G68, G69, G37, G29, G55 and G60 (Figure 4.6). Nine perpendicular lines
were drawn starting from the origin and extended beyond the polygon such that the biplot
was divided into nine sectors and environments fell into two of them. Five environments
namely Rattray Arnold, Agricultural Research Trust farm, Kadoma, Harare low N and
Chisumbanje fell into sector 1 delineated by rays 1 and 2 and the vertex genotypes were G60,
G74 and G68. Two environments Chiredzi summer and Chiredzi winter fell into sector 9.

112
 Ray 1

Sector 1

Mega-environment 1
PC2

ART
RARS

PC1

Figure 4.6 Polygon view of the genotype and genotype by environment interaction
biplot based on symmetrical scaling for the “which-won-where” pattern for
genotypes and environments.
CH WINTER=chiredzi winter; CH=chiredzi summer; CS=Chisumbanje; HRE Low N=Harare low nitrogen; Rattray Arnold
A GGEStation;
Research biplotART=Agricultural
was drawn toResearch
show Trust
which environment discriminated genotypes better than
farm.

A GGE biplot was drawn to show which environment discriminated genotypes better than
the other environments. Environments PC1 had only positive scores, whilst PC2 had both
positive and negative scores (Figure 4.7). Environments Rattray Arnold Research Station,
Agricultural Research Trust farm and Harare Low N had high PC1 scores and near zero PC2
scores. Chiredzi summer, Chiredzi winter and Chisumbanje also had high PC2 scores. PC1
scores correlated with environment yield scores (r=0.569; P≤0.05). Correlation coefficients
among test environments are presented in Table 4.6. All environments were positively
correlated because all angles among them were smaller than 90°. Harare low N and
Agricultural Research Trust farm had the largest highly significant positive correlation
(r=0.929) while the lowest correlation was between Chisumbanje and Chiredzi summer.

113
 interactions
Cross-over
PC2

ART

Non-crossover interactions/More
discriminate or powerful

PC1

Figure 4.7 Genotype and genotype by environment interaction biplot based on

environment-focused scaling for environments.
CH WINTER=Chiredzi winter; CH=Chiredzi summer; CS=Chisumbanje; HRE Low N=Harare low nitrogen; RARS=Rattray
Arnold Research Station; ART=Agricultural Research Trust farm.

Table 4.6 Correlation coefficients among test environments

ART HRE Low N Kadoma CH winter RARS CH

HRE Low N 0.929**

Kadoma 0.463** 0.626**

CH winter 0.291** 0.163** 0.183**

RARS 0.230** 0.460** 0.775** 0.127**

CH 0.455 0.082 0.569** 0.279* 0.137

CS 0.178 0.272 0.380 0.049 0.261 0.89**

**P≤0.001; *P≤0.05; CH WINTER=Chiredzi winter; CH=Chiredzi summer; CS=Chisumbanje; HRE Low N=Harare low nitrogen;
RARS=Rattray Arnold Research Station; ART=Agricultural Research Trust farm.

Rattray Arnold Research Station and Agricultural ResearchTrust farm were also highly and
significantly correlated (r=0.230).

114
A hierarchical cluster analysis was also done for environments (Figure 4.8). Harare low N
was closely related to Chiredzi winter and the two sites comprised of managed stress.
Chiredzi summer and Chisumbanje clustered together and these environments share a similar
geographical location. Chiredzi summer and Chisumbanje also have similar rainfall patterns
and were both used as random drought sites. Rattray Arnold Research Station and Kadoma
clustered together and they have similar rainfall patterns. Agricultural Research Trust farm
did not cluster with any environment and is usually characterised by high rainfall figures that
can sometimes be above 1 000 mm.

4.3.4 Genotype and genotype by environment interaction biplot analysis for 20 best
performing hybrids
An across site analysis was done using Fieldbook software (Banziger and Vivek, 2007)
embedded in an Excel spread sheet and the 20 best performing hybrids were selected. The
yield performance data of these hybrids across seven environments is presented in Table 4.7.
The bold and underlined mean yields are for those hybrids that were the best performers in
each environment. The variability of the best performing hybrid from one environment to the
other shows the existence of possible crossover G x E interaction. G74 (SC727) and G79
(CML395/CML444) were the check varieties and were amongst the best performing hybrids.
G74 was the overall best performing hybrid as well as being the best performer in
environments Agricultural Research Trust farm, Harare low N and Rattray Arnold Research
Station. G61 was the best performing hybrid in Kadoma and G52 was the best performing
genotype in Chiredzi winter. The genotype G57 was the best performer in Chiredzi summer
and G63 in Chisumbanje. A GGE biplot based on genotype-focused scaling was shown in
order to perceive the positioning of the genotypes (Figure 4.9). Genotypes with PC1 scores
>0 were identified as higher yielding and adapted and those that had PC1 scores <0 were
identified as lower yielding. Genotypes G74, G68 and G48 were the highest yielding with
PC1 scores greater than zero, whilst G60, G61 and G63 were lower yielding with PC1 scores
less than zero. G7 and G79 were high yielding and stable with PC2 scores closer to zero. G39
was low yielding but very stable with a PC2 score of zero. Yield stability and performance of
the 20 best performing hybrids is presented in Figure 4.10.

115
 ART farm

HRE low N

CH winter

Kadoma

RARS

CH

CS

0.00 7.00 14.00 21.00 28.00 35.00

Euclidean dissimilarity coefficient

Figure 4.8 Hierarchical cluster analysis of the seven environments.

ART=Agricultural Research Trust farm; HRE low N=Harare low nitrogen; CH winter=Chiredzi winter; RARS=Rattray Arnold Research Station; CH=Chiredzi summer;
CS= Chisumbanje.

116
Table 4.7 Mean grain yield (t ha-1) for 20 genotypes across seven environments in two
seasons
E1 E2 E3 E4 E5 E6 E7 Mean
Genotype t ha-1
G74 10.43 4.08 5.29 2.50 6.16 2.50 4.85 5.12
G61 8.34 2.31 6.68 2.79 3.65 3.65 5.76 4.74
G57 8.20 2.44 6.42 1.78 4.29 4.26 5.19 4.66
G54 7.72 1.54 4.26 2.33 4.90 3.63 6.12 4.36
G48 8.36 1.63 4.60 1.99 5.39 3.23 4.61 4.26
G63 6.88 1.30 5.40 2.62 3.23 4.08 6.51 4.29
G52 8.91 2.35 5.04 3.26 4.04 3.06 3.26 4.27
G79 8.39 1.28 5.51 2.49 4.51 3.82 3.56 4.22
G45 7.52 1.97 6.03 1.92 4.21 3.00 4.59 4.18
G68 9.99 1.70 4.82 1.40 4.22 3.77 3.72 4.23
G10 7.49 1.37 5.00 2.41 4.69 3.31 4.70 4.14
G72 7.46 1.50 5.66 2.31 4.18 3.93 4.16 4.17
G34 7.08 1.68 5.98 2.64 3.60 3.65 4.30 4.13
G38 6.99 1.88 5.82 3.01 3.98 4.16 3.89 4.25
G60 6.69 2.83 6.37 2.51 4.18 3.91 4.31 4.40
G7 7.82 1.66 5.24 1.71 5.23 2.73 3.31 3.96
G58 7.22 1.96 5.51 2.43 3.89 3.93 5.02 4.28
G59 6.78 2.08 5.00 2.84 3.89 3.35 5.15 4.16
G70 8.02 1.03 4.62 2.38 4.05 3.80 4.90 4.11
G8 7.25 2.01 5.60 1.71 4.51 3.15 4.31 4.08
-1
Mean (t ha ) 6.83 1.62 4.50 2.09 3.96 3.17 3.92
E1=Agricultural Research Trust farm; E2=Harare low nitrogen; E3=Kadoma; E4=Chiredzi winter; E5=Rattray Arnold Research
Station; E6=Chiredzi summer; E7=Chisumbanje; Underlined and bold=highest yielder in the given environment.

G74 was again the highest yielding (biggest PC1 score) but unstable in different
environments (large PC2 score). G74 was the highest yielding since it was the best performer
in many testing environments with high mean yields (Table 4.7). G63 was the poorest
performing genotype (low PC1 score) with low yields in different environments but highly
stable (near zero PC2 score). G52 and G57 were also relatively high yielding and very stable
with PC2 scores close to zero. G79 and G7 were also relatively stable, whilst G68, G48, G60
and G61 were unstable in different environments. G68 and G74 were associated with Rattray
Arnold Research Station and Agricultural Research Trust farm, whilst G60 and G61 were
associated with Chiredzi winter. Chiredzi summer was associated with G70 and G54.

Agricultural Research Trust farm had the longest vector followed by Harare low N, Kadoma
and Rattray Arnold Research Station, whilst Chisumbanje had the shortest vector (Figure

117
4.11). Environments Harare low N, Agricultural Research Trust farm, Rattray Arnold
Research Station and Chiredzi summer had positive PC1 scores, whilst Chiredzi winter,
Kadoma and Chisumbanje had negative PC1 scores. G74 had the longest vector, whilst G45
and G22 had the shortest vectors (Figures 4.11 and 4.12). The acute angles between G60,
G61, G57, G52 and G74 indicate that those genotypes performed similarly across environme
nts.

Genotypes G61, G60, G57 and G52 had angles less than 90° between their vectors and
environment vectors for Chiredzi winter, Kadoma and Harare low N. G79, G7, G68 and G48
had acute angles between them and environments Chiredzi summer, Rattray Arnold Research
Station and Agricultural Research Trust farm. G74 and G52 were located nearer to the biplot
origin. The angle between G70 and G74 was more than 90°. The environments Harare low
N, Kadoma and Chiredzi winter had less than 90° angles between them. The angle between
Harare low N and Chiredzi summer was more than 90° as well as between Chiredzi summer
and Chiredzi winter.
Unstable

Stable
ART
PC2

Unstable

Adaptable or higher yielding

Lower yielding

PC1

Figure 4.9 Genotype and genotype by environment interaction biplot based on

genotype-focused scaling for the top 20 yielding genotypes.
PC=principal component; G=genotype; CS=Chisumbanje; CH=Chiredzi summer; CH WINTER=Chiredzi winter; RARS=Rattray
Arnold Research Station; ART=Agricultural Research Trust farm; HRE Low N=Harare low nitrogen.

118
A six-sided polygon was formed from genotype markers G68, G74, G61, G60, G63 and G70
(Figure 4.13). The polygon is produced by linking markers of varieties that are the furthest
away from the biplot origin such that all other genotypes are contained within the polygon.
Six perpendicular lines, starting from the origin were drawn extending beyond the polygon
such that the biplot was divided into six sectors. All six sectors had locations within them.
Three environments Agricultural Research Trust farm, Rattray Arnold Research Station and
Harare low N fell within sector 1 outlined by rays 1 and 2 and the vertex genotypes for this
sector were G74 and G68. Sector 1 therefore formed mega-environment 1. Sector 2 and 3
comprising of environments Kadoma and Chiredzi winter formed mega-environment 2 and
the vertex genotypes were G60 and G61. Chisumbanje and Chiredzi summer formed the third
mega-environment in sectors 4 and 5 and the vertex genotypes were G63 and G70.

ART
PC2

PC1

Figure 4.10 Grain yield stability and performance of the 20 top yielding genotypes in
seven environments across two seasons.
PC=principal component, G=genotype, CS=Chisumbanje, CH=Chiredzi, CH winter=Chiredzi winter, ART=Agricultural Research
Trust farm; RARS=Rattray Arnold Research Station, HRE Low N=Harare low nitrogen.

119
 PC2

ART

PC1

Figure 4.11 Relationship amongst testing environments and genotype by testing

environments for the 20 top yielding genotypes.
PC=principal component; G=genotype; CS=Chisumbanje; CH=Chiredzi summer; ART=Agricultural Research Trust farm; CH
Winter=Chiredzi winter; HRE Low N=Harare low nitrogen.

ART
PC2

PC1

Figure 4.12 Genotype and genotype by environment interaction biplot based on

genotype and environment focused scaling for comparison of genotypes and
environments for top 20 yielding genotypes.
PC=principal component; G=genotype; CS=Chisumbanje; CH=Chiredzi summer; ART=Agricultural Research Trust farm; CH
WINTER=Chiredzi winter; HRE Low N=Harare low nitrogen.

120
 Ray 1

ART
PC2

Ray 3 Sector 1
Sector 2 Ray 2

PC1

Figure 4.13 Polygon views of the genotype and genotype by environment interaction
biplot based on symmetrical scaling for the “which-won-where” pattern for
genotypes and environments for the 20 top yielding genotypes.
PC=principal component; G=genotype; CS=Chisumbanje; CH=Chiredzi summer; ART=Agricultural Research Trust farm; CH
Winter=Chiredzi winter; HRE Low N=Harare low nitrogen.

4.4 Discussion
In this study the environment explained 85% of the total G + E + G x E variation, whilst
genotypes explained 7.0% and G x E interaction 13.1%. These results are consistent with
findings by other researchers (Yan et al., 2000; Kaya et al., 2006; Muungani et al., 2007;
Jalata, 2011). Gauch and Zobel (1997) reported that in normal multi-location yield
experiments location accounted for about 80% of the total variation, whilst genotype and G x
E interaction each accounted for about 10%. As a percentage of the total sum of squares the
environment accounted for 40.7%, genotype 3.6% and G x E interaction for 6.7% of the
variation. These results are also consistent with findings from other investigators (Sabaghnia
et al., 2008; Ramburan et al., 2011). However in this study the contribution by G x E
interaction as a percentage of total sum of squares was lower than what has been reported by
other two research groups. Environment and environment by year accounted for 60% of the
total variability in this study. Muungani et al. (2007) reported environment and environment
x year contributions of 70.44%. This result is an indication that variation amongst
environments within and across years justifies the need for multi-environment yield trials.
The huge yield disparity due to location is pertinent to genotype assessment and mega

121
location analysis (Fox and Rosielle, 1982; Gauch and Zobel, 1996) and again the large G x E
interaction compared to genotype contribution suggests the probable existence of different
mega locations.

In this study G74 (SC727) was the best performing genotype in environments Agricultural
Research Trust farm, Harare low N and Rattray Arnold Research Station and this was an
indication of a non-crossover G x E interaction. Results in this study also indicated the
presence of a crossover G x E interaction as described by Baker (1988), Yan and Hunt
(2001), Kaya et al. (2006) and Jalata (2011). The G x E interaction was further analysed with
the aid of the AMMI model for grain yield stability. The ANOVA indicated highly
significant contribution of environments, genotypes and G x E interaction to variation and
results are in agreement with findings from other studies (Sabaghnia et al., 2008; Ramburan
et al., 2011; Thangavel et al., 2011). There is a declining impact of the G x E interaction sum
of squares with an increasing number of IPCA axes. In this study IPCA1 accounted for
48.27% of the G x E sum of squares interaction and IPCA2 accounted for 27.82% and in
total they accounted for 76.10%. Findings are in line with Yan (2002) and Thangavel et al.
(2011) who reported that most of the interaction occurred in the first few axes. AMMI
analysis appears to be able to extract a large portion of G x E interaction and thus is efficient
in analysing G x E interaction as demonstrated by Zobel et al. (1988).

A GGE biplot based on genotype-focused scaling was illustrated in order to identify the
positioning of genotypes. The GGE biplot analysis of yield for the 80 genotypes across seven
environments explained 61.7% of genotype main effects and G x E interaction with the
primary (PC1) and secondary (PC2) scores contributing 47.5% and 14.2% respectively. On
the other hand the GGE biplot for the 20 genotypes explained 54.6% of the genotype main
effect and G x E interaction. Similar findings have been recorded by other investigators
(Kaya et al., 2006; Muungani et al., 2007; Jalata, 2011). PC1 scores >0 successfully detected
genotypes that are high yielding and PC1 scores <0 discriminated the low yielding ones. On
the other hand the PC2 scores showed the genotypic stability of the genotypes. Genotypes of
interest were divided into two groups, where group 1 consisted of the stable and high

122
yielding genotypes (G52, G57, G79 and G7) and group 2 consisted of unstable but high
yielding genotypes (G74, G68, G48, G61 and G60).

Genotypes with above average means were G48 to G74 and genotypes with below average
means were G45 to G63. The longer the environmental vector, the more significant is the
genotype main effect and the more significant the selection based on mean performance
(Jalata, 2011). Thus genotypes with above average mean performance can be selected for
future breeding. In this study the average environment vector was long enough for the
selection of genotypes to be done based on yield mean performances. An ultimate genotype
must demonstrate both high average yield performance and high stability across locations
(Kaya et al., 2006; Yan and Tinker, 2006; Jalata, 2011). The ideal genotypes in this study
were identified as G52 (CML548/2N3d-B) and G57 (CML444-BB/RS61P-B).

In this study the vertex genotypes for the mega environment 1 were G74 and G68 and these
genotypes were the winning genotypes in the specified environments i.e. Rattray Arnold
Research Station, Agricultural Research Trust farm and Harare low N, whilst G60 and G61
were the winning genotypes in Kadoma and Chiredzi Winter. Since there was a high
correlation between genotype PC1 scores and genotype main effects and as the GGE biplot
sufficiently explained the GGE variation, it can be statistically demonstrated that locations in
the same sector share the same winning genotypes (Yan et al., 2000).

A GGE biplot which hinges on environment-focused scaling was depicted to assess the
pattern of environments. Environment PC1 had only positive scores and similar results were
reported by previous scientists (Yan et al., 2000; Yan and Hunt, 2001; Kaya et al., 2006; Yan
and Tinker, 2006). This consequently proposes that the PC1 represents comparative genotype
yield differences across environments, which leads to a non-crossover G x E interaction.
Genotypes with large PC1 scores can be easily recognised in environments with larger PC1
scores (Yan et al., 2000). The environment PC2 scores had both positive and negative scores
and this was an indication of crossover G x E interaction. This leads to disproportionate
genotype yield differences across environments. In circumstances where resources are
limiting and there is a need to carry out multi-environment yield trials, Kadoma, Harare low

123
N, Agricultural Research Trust farm and Rattray Arnold Research Station may be the better
test environments since they had near zero PC2 scores and large PC1 scores. Agricultural
Research Trust farm was the most discriminating environment (largest PC1 score) and had
the longest environment vector. Chiredzi summer, Chiredzi winter and Chisumbanje had
large PC2 scores which would therefore mean that the cultivar differences observed at these
sites may not exactly reflect cultivar differences in average yield over all sites.

All environments in the study were positively correlated because the angles among them
were less than 90°. Kaya et al. (2006) reported similar results. The angle between Rattray
Arnold Research Station and Agricultural Research Trust farm was very small, hence the
significant correlation (r=0.230). However, there were inconsistencies, since the largest
correlation would have been expected to be between Chiredzi summer and Chiredzi winter
but instead it was between Harare low N and Agricultural Research Trust farm which had a
much larger angle between them than between Chiredzi summer and Chiredzi winter. The
biplot did not explain 100% of the GGE variation so some discrepancies were expected (Yan,
2002).

In a situation where the same trait is measured on the same genotypes in different
environments indirect selection can be applied. The significant correlation coefficients
between test environments suggest that indirect selection for grain yield can be applied
across the test environments. For instance the higher yielding genotypes at Agricultural
Research Trust farm may also show similar responses at Harare low N, Kadoma, Chiredzi
winter and Rattray Arnold Research Station. The existence of significant correlation between
environments showed that the information obtained was similar so that testing environments
can be reduced to minimise the cost without significantly affecting the validity of the data.

An environment that is more representative of other test environments is the one with a
smaller angle with the average environment axis (Yan and Tinker, 2006). Agricultural
Research Trust farm was the most representative, followed by Harare low N, whilst Rattray
Arnold Research Station and Chiredzi summer were the least representative. Environments
that are both discriminating and representative such as Agricultural Research Trust farm are

124
ideal test environments for selecting generally adapted genotypes and the discriminating and
non-representative such as Rattray Arnold Research Station and Chiredzi winter are useful
for selecting specifically adapted genotypes if the target environments can be divided into
mega-environments. Again the discriminating and non-representative environments can be
used for culling unstable genotypes if the target environment is a single mega environment.

In this study hybrids had above average performance in all environments as indicated by
angles that were <90°. The genotype and environment means biplot using AMMI analysis
clustered environments into four groups: Group 1 included Agricultural Research Trust farm
and Rattray Arnold Research Station, group 2 consisted of Chisumbanje, group 3 consisted
of Harare low N, Chiredzi winter and Chiredzi summer and lastly group 4 consisted of
Kadoma. Among all location-year testing environments, environment Agricultural Research
Trust farm interacted with genotypes the same way as environment Rattray Arnold Research
Station. The clustering of environments can be explained by similar weather conditions and
similar growing conditions. Agricultural Research Trust farm and Rattray Arnold Research
Station were optimum environments located in the Highveld and also with annual average
rainfall of ±800 mm. Environments Harare low N, Chiredzi winter and Chiredzi summer
were mainly associated with stress management. Harare was a low N site, Chiredzi winter a
managed drought site and Chiredzi summer was a random drought site, associated with
severe mid-season droughts. Chisumbanje clustered on its own because it has distinct soils
and is located in the south with low altitude. Hierarchical cluster analysis based on hybrid
grain yield clustered environments based on geographical location and stress conditions and
similar results were reported by Gissa (2008). Harare low N and Chiredzi winter are distant
geographical locations but clustered together as a result of prevailing stress conditions in
both environments. When resources are limiting one can select one environment where more
than one environment exist in the cluster.

4.5 Conclusions
The level of G x E interaction of the single cross hybrids in this study was larger than that of
genotype main effects but smaller than that of environment main effects. Genotypes
demonstrated both crossover and non-crossover types of G x E interactions. The former led

125
to substantially different genotype rankings across test environments, therefore making
selection difficult. The stable and high yielding genotypes were identified as well as the
discriminating and representative test environments. Genotypes G52 (CML548/2N3d-B) and
G57 (CML444-BB/RS61P-B) were identified as the most ideal genotypes due to high grain
yield performance and stability. Agricultural Research Trust farm was identified as the most
discriminating and representative test environment. The optimum environments
discriminated the genotypes differently from the stress environments. In some cases
environments sharing the same location but with different stress levels discriminated the
genotypes similarly and this showed the possibility of developing genotypes under both
stress and optimal environments.

4.6 References
Agronomix Software, Inc. 2005. AGROBASE Generation II User’s Manual. Version II,
Revised Edition. www.agronomix.com. Agronomix Software, Winnipeg, M.B,
Canada.
Aina, O.O., A.G.O. Dixon and E.A. Akinrinde. 2007. Additive main effects and
multiplicative interaction (AMMI) analysis for yield of cassava in Nigeria. Journal
of Biological Sciences 7: 796-800.
Annicchiarico, P. and M. Perenzin. 1994. Adaptation patterns and definition of macro
environment for selection and recommendation of common wheat in Italy. Plant
Breeding 113: 197-205.
Baker, R.J. 1988. Tests for cross over genotype-environment interactions. Canadian Journal
of Plant Science 68: 405-410.
Banziger, M. and B.S. Vivek. 2007. Fieldbook: Software for Managing a Maize Breeding
Program. CIMMYT.
Comstock, R.E. and R.H. Moll. 1963. Genotype x environment interactions. Symposium on
Statistical Genetics and Plant Breeding. National Academy Science National
Research Council, Washington, D.C. pp 164-196.
Crossa, J. 1990. Statistical analysis of multilocation trials. Advances in Agronomy 44: 55-85.

126
Ebdon, J.S. and H.G. Gauch. 2002. Additive main effect and multiplicative analysis of
national turfgrass performance trials: Cultivar recommendations. Crop Science 42:
497-506.
Eberhart, S.A. and W.A. Russel. 1966. Stability parameters for comparing varieties. Crop
Science 6: 36-40.
Finlay, K.W. and G.N. Wilkinson. 1963. The analysis of adaptation in a plant breeding
programme. Australian Journal of Agricultural Research 14: 742-754.
Fox, P.N. and A.A. Rosielle. 1982. Reducing the environmental main effects on pattern
analysis of plant breeding environments. Euphytica 31: 645-656.
Gauch, H.G. 1988. Model selection and validation for yield trials with interaction. Biometrics
44: 705-715.
Gauch, H.G. 1992. Statistical analysis of regional yield trials: AMMI analysis of factorial
designs. Elsevier, Amsterdam, Netherlands. p 278.
Gauch, H.G. and R.W. Zobel. 1996. AMMI analysis of yield trials. In: Kang, M.S. and H.G.
Gauch (eds) Genotype by Environment Interaction. CRC Press, Boca Raton, FL. pp
1-40.
Gauch, H.G. and R.W. Zobel. 1997. Identifying mega environments and targeting genotypes.
Crop Science 37: 311-26.
GenStat. 2011. Introduction to GenStat for Windows. 14th edition. VSN International, Hemel
Hemstead, Hertfordshire HPI. IES, UK.
Gissa, D.W. 2008. Genotypic variability and combining ability of quality protein maize
inbred lines under stress and optimal conditions. PhD Thesis. Department of Plant
Sciences, Faculty of Natural and Agricultural Sciences, University of the Free State,
South Africa.
Jalata, Z. 2011. GGE biplot analysis of multi-environment yield trials of barley (Hordeum
vulgare L.) genotypes in south-eastern Ethiopia Highlands. International Journal of
Plant Breeding and Genetics 5: 59-75.
Kang, M.S. and R. Magari. 1996. New developments in selecting for phenotype stability in
crop breeding. In: Kang M.S. and H.G. Gauch Jr (eds) Genotype x Environment
Interaction. CRC Press: Boca Raton, FL. pp 1-14.

127
Kang, M.S., D.P. Gorman and H.N. Pham. 1991. Application of a stability statistics to
international maize yield trials. Theoretical and Applied Genetics 81: 162-165.
Kaya, Y., M. Akcura and S. Taner. 2006. GGE biplot analysis of multi environment yield
trials in bread wheat. Turkish Journal of Agriculture 30: 325-337.
Ma, B.L., W. Yan, L.M. Dwyer, J. Fregeau-Reid, H.D. Voldeng, Y. Dion and H. Nass. 2004.
Graphic analysis of genotype, environment, nitrogen fertiliser and their interaction
on spring wheat yield. Agronomy Journal 96: 169-80.
Matus-Cadiz, M.A., P. Hucl, C.E. Peron and R.T. Tyler. 2003. Genotype x environment
interaction for grain color in hard white spring wheat. Crop Science 43: 219-226.
Muungani, D., P. Setimela and M. Dimairo. 2007. Analysis of multi-environment, mother-
baby trial data using GGE biplots. African Crop Science Conference Proceedings.
African Crop Science & Minia University. El-Minia Egypt. pp 103-112.
Perkins, J.M. and J.L. Jinks. 1968. Environmental and genotype-environmental interactions
and physical measures of the environment. Heredity 25: 29-40.
Pham, H.N. and M.S. Kang. 1988. Interrelationships among respectability of several stability
statistics estimated from international maize trials. Crop Science 28: 925-928.
Ramburan, S., M. Zhou and M. Labuschagne. 2011. Interpretation of genotype x
environment interactions of sugarcane: Identifying significant environmental factors.
Field Crops Research 124: 392-399.
Rohlf, F.J. 1993. NTSYS-pc. Numerical taxonomical and multivariate analysis system.
Exerter software, Setauket, New York, USA.
Sabaghnia, N., S.H. Sabaghpour and H. Dehghani. 2008. The use of an AMMI model and its
parameters to analyse yield stability in multi-environment trials. Journal of
Agricultural Science 146: 571-581.
Samonte, S.O.P.B., L.T. Wilson, A.M. McClung and J.C. Medley. 2005. Targeting cultivars
onto rice growing environments using AMMI and SREG GGE biplot analysis. Crop
Science 45: 2414-2424.
SPSS 15.0 for Windows. 2006. www.spss.com SPSS Inc., Chicago III.
Thangavel, P., A. Anandan and R. Eswaran. 2011. AMMI analysis to comprehend genotype-
by-environment (G x E) interactions in rainfed grown mungbean. Australian Journal
of Crop Science 5: 1767-1775.

128
Yan, W. 2002. Singular value partitioning in biplot analysis of multi-environment trial data.
Agronomy Journal 94: 990-996.
Yan, W. and L.A. Hunt. 2001. Genetic and environmental causes of genotype by
environment interaction for winter yield in Ontario. Crop Science 41: 19-25.
Yan, W. and L.A. Hunt. 2002. Biplot analysis of multi-environment trial data: In: Kang, M.S.
(ed) Quantitative Genetics, Genomics and Plant Breeding. CAB International,
Willingford. pp 289-303.
Yan, W. and M.S. Kang. 2003. GGE Biplot Analysis: A Graphical Tool for Breeders,
Geneticists and Agronomists. CRC Press, Boca Raton, FL. pp 3-93.
Yan, W. and N.A. Tinker. 2005. A biplot approach to the investigation of QTL by
environment patterns. Molecular Breeding 15: 31-43.
Yan, W. and N.A. Tinker. 2006. Biplot analysis of multi-environment trial data: Principles
and applications. Canadian Journal of Plant Sciences 86: 623-645.
Yan, W., L.A. Hunt, Q. Sheng and Z. Szlavnics. 2000. Cultivar evaluation and mega-
environment investigation based on the GGE biplot. Crop Science 40: 597-605.
Yan, W., M.S. Kang, B. Ma, S. Woods and P.L. Cornelius. 2007. GGE Biplot vs AMMI
analysis of genotype by environment data. Crop Science 47: 643-53.
Yan, W., P.L. Cornelius, J. Crossa and H.L. Hunt. 2001. Two types of GGE biplot for
analysing multi-environment trial data. Crop Science 41: 656-63.
Zobel, R.W., M.J. Wright and H.G. Gauch. 1988. Statistical analysis of a yield trial.
Agronomy Journal 80: 388-393.

129
 CHAPTER 5

Genetic variation among CIMMYT and Zimbabwean maize inbred lines

Abstract
Genetic characterisation of breeding lines is of great importance as it enables breeders to
maximise on heterosis in hybrid combinations as well as maintain genetic diversity in the
breeding material. The Zimbabwe National Breeding Programme (DR&SS) acquires
germplasm from CIMMYT, IITA and other breeding programmes to enhance genetic
diversity of its breeding materials. The objective of this study was to analyse the genetic and
morphological diversity and heterotic relationships amongst and between DR&SS and
CIMMYT lines in order to facilitate selection of parents for drought tolerant hybrid crosses.
A set of 23 inbred lines (10 from DR&SS and 13 from CIMMYT) were evaluated for 14
morphological traits across seven sites in Zimbabwe in the 2009/10 and 2010/11 seasons.
The morphological data revealed that there was variability amongst inbred lines that could be
manipulated through selection and hybridisation. The variability was further substantiated
using PCA, where the overall diversity could not be explained by a few eigenvectors. All
traits accounted for the variability; however traits such as grain yield, texture, ear aspect,
common rust, GLS and anthesis days were the major contributors. Lines clustered into two
major clusters and four sub-groups using the Euclidean dissimilarity coefficient and in some
cases lines related by pedigree were tightly clustered. Lines were also fingerprinted using 1
129 SNP markers. Molecular analysis yielded a total of 2 258 alleles, with an average of two
alleles per locus. Cluster analysis based on Rogers’ dissimilarity coefficient revealed two
major clusters and five sub-groups among the lines. Moderate genetic diversity was observed
with an average dissimilarity of 0.32 and average polymorphic information content (PIC)
value of 0.258. The clustering was largely not consistent with the available pedigree
information of the lines. Information generated in this study will, however, aid breeders to
decide on which hybrids to constitute and evaluate.

130
5.1 Introduction
Maize is the staple food crop and main source of carbohydrates in the majority of the
Zimbabwe populace, with a per capita consumption of 150 kg annum-1. There are different
players in the seed industry who are making an effort to provide a wide range of varieties for
selection by farmers and the Zimbabwe National Breeding Programme (DR&SS) is one of
them. The programme acquires germplasm from CIMMYT, IITA and other breeding
programmes to enhance genetic diversity of its breeding materials through initiation of new
breeding projects. Hybrids and populations developed in a maize breeding programme
become important sources for inbred line extraction.

Genetic diversity evaluation is used by maize breeders as an alternative method for

germplasm selection (Devi and Singh, 2011). Genetic progress depends on the existence of
genetic variability. Thus, genetic characterisation of the available germplasm is critical for
efficient development of new varieties (Amorim et al., 2003). Genetic characterisation of
maize germplasm can be used to group different genotypes into heterotic groups whereby
lines in the same group are similar compared to lines in different groups (Lu et al., 2009).
Information on the relationship between maize inbred lines is critical in designing a breeding
programme especially when it comes to selecting parents to be used in hybrid development.
Maize hybrids are generally constituted from inbred lines from complimentary and opposite
heterotic groups in order to exploit hybrid vigour. Hybrid development in maize has
generally become the backbone of the formal seed sector across continents as a result of high
yielding and adapted varieties that have led to infrastructure development and improvement
of the seed value chain as a whole (Devi and Singh, 2011). Hybrid development is therefore
dependant on the availability of diverse homozygous inbred lines and evaluating their
performance in different combinations across different locations (Bauman, 1981; Smith,
1986; Hallauer, 1990; Troyer, 2004; Koutsika-Sotiriou and Karagounis, 2005).

Inbred line development and selection in maize breeding programmes as well as evaluating
hybrid performance in multi-location trials are easy but expensive and require a lot of time. A
small number of inbred lines can be used to produce a large number of hybrids, of which
evaluating all of them usually poses a challenge (Hallauer, 1990). Therefore making use of

131
genetic markers in assessing diversity amongst inbred lines has been suggested as a solution
towards overcoming bottlenecks, thereby allowing the performance of single cross hybrids to
be predicted (Lanza et al., 1997). Genetic diversity and the levels of genetic variation in
maize can be estimated using both molecular and morphological markers. Earlier studies
have shown that molecular markers are not influenced by the environment like
morphological traits (Williams et al., 1990; Smith and Smith, 1992). Recent studies have
further shown remarkable evidence of reasonable environmental influence on plant
development (Molinier et al., 2006; Li et al., 2009). Hence combining morphological and
molecular analysis should form a reliable basis for germplasm improvement. Molecular
markers that have been found in abundance in all genomes that have been studied are SNP
markers.

It is obvious that SNP markers would become a marker of choice considering their
abundance in different genomes as well as their ability to be processed automatically. SNP
markers have been incorporated in different applications such as identifying cultivars, genetic
map construction, positional cloning of targeted loci, genetic diversity assessment,
determining ancestry, association mapping as well as marker-assisted breeding (Gupta et al.,
2001; Rafalski, 2002; Lijavetzky et al., 2007). Maize is considered highly polymorphic
amongst different crop species with an average SNP frequency of 1% (Tenaillon et al.,
2001), followed by rice with a SNP frequency of 0.5-0.78% (IRGSP, 2005). Ravel et al.
(2006) found that wheat had a SNP frequency of 0.5%, whilst Zhu et al. (2003) found that
soybean had a SNP frequency of 0.36%. Rapeseed was found to have the lowest SNP
frequency (0.16%) (Farman et al., 2002). There are a large number of SNP markers
available for use in maize, of which many were developed using DNA sequences of known
genes. According to Tenaillon et al. (2001) the SNP frequency for chromosome 1 of maize
has been estimated to be between 1 in 104 bp for two randomly paired landraces and 1 in 124
bp in 36 inbred lines. The objective of this study was to analyse morphological and genetic
diversity and heterotic relationships among 10 DR&SS elite inbred lines and 13 CIMMYT
inbred lines. The genetic analysis results would then assist in selecting parents for drought
tolerant hybrid crosses. The lines under evaluation in this study are potential candidates for
the national drought breeding programme. Once the heterotic relationships amongst the lines

132
are determined, hybrid development would be done in such a way that inbred lines from
divergent groups are crossed in order to maximise hybrid vigour.

5.2 Materials and methods

5.2.1 Germplasm selection

A total of 23 inbred lines were chosen for characterisation and of these lines, 10 were from
DR&SS and 13 from CIMMYT. Elite lines were chosen from both programmes with
CIMMYT lines being chosen specifically for their biotic stress resistance or abiotic stress
tolerance. All maize lines were the white kernel types. The DR&SS lines included the
parental genotypes of popular and widely grown commercial hybrids but being susceptible to
both biotic and abiotic stresses such as drought and MSV disease. Details of the lines are
given in section 3.2.1. Lines were planted at seven sites in 2009/2010 and again at the same
sites in the 2010/2011 season, to facilitate capturing of morphological data but two random
drought sites were excluded from the analysis.

5.2.2 Site selection

Details of sites used are given in sections 3.2.2 and 4.2.2.

5.2.3 Experimental design and morphological traits

The experimental design is as given in section 3.2.4. Data was recorded as described in Table
3.2. Ear rots were measured as a percentage of rotten ears, whilst disease scores for GLS,
northern leaf blight, common rust (Puccinia sorghi Schw.) and MSV were measured on a
scale of 1-5 where 1 is clean of infection and 5 is severely diseased. Husk cover was
measured as a percentage of plants with ears that were not completely covered by husks.

5.2.4 Deoxyribonucleic acid extraction

Seedlings for all 23 lines were raised in plastic seed trays for about 2-3 weeks (until they
reached the 3-4 leaf stage). Equal amounts of leaf tissue were harvested from 10 plants per
inbred line. The leaf tissue was bulked, cut into pieces with scissors and transferred into 1.2
ml strip tubes that contained two 4 mm stainless steel grinding balls. The tissue was freeze-

133
dried (lyophilised) for three days using a Labconco freeze-dryer (Labconco, Kansas City,
MO, USA) as described in the user’s manual. The lyophilised leaf samples were ground into
fine powder using a GenoGrinder (SPEX Certi Prep, Metuchen, NJ, USA) at 500 strokes per
min for 4 min. Genomic DNA was extracted using a modified version of the high throughput
mini-prep Cetyltrimethylammonium bromide (CTAB) method (Mace et al., 2003). A
modified CTAB extraction buffer was used (200 mM tris (hydroxymethyl) aminomethane
(Tris), pH 7.5; 50 mM ethylenediaminetetra acetate (EDTA), pH 8.0; 2 M NaCl; 2% (w/v)
CTAB; 1% (v/v) beta-mercaptoethanol). A total of 700 µl of the extraction buffer was added
to each sample. Grinding was done for 30 seconds in order to mix the powder with the
extraction buffer. Samples were then incubated at 65°C in a water bath for 30 min with
continuous gentle rocking. Tubes were gently inverted every 10 min to thoroughly mix the
tissue with the extraction buffer. Tubes were removed from the water bath and allowed to
cool for 10 min in a fume hood. Samples were then gently mixed and centrifuged at 3 500
revolutions per min (rpm) for 10 min. A total of 500 µl of the aqueous phase was transferred
into new tubes and 600 µl chloroform:isoamylalcohol (24:1) was added to the sides of the
tubes. This was followed by gentle mixing through rocking for 5 min and centrifuging at
3 500 rpm for 10 min. The upper aqueous layer was transferred into fresh strip tubes and the
chloroform:isoamylalcohol wash repeated. A total of 400 µl of the upper aqueous phase was
transferred into fresh strip tubes and 0.66 volumes of 100% cold isopropanol was added and
mixing was gently done for 5 min in order to precipitate the nucleic acid. Centrifuging at 3
500 rpm was done for 20 min to form a pellet at the bottom of the tube. Pellets were then
washed with 70% (v/v) ethanol (600 µl) followed of centrifugation for 10 min after which the
ethanol was discarded through decantation. The wash step was done twice. Pellets were
allowed to air-dry in the fume hood until the ethanol had completely evaporated. The air-
dried pellets were suspended in 200 µl TE buffer (Tris/EDTA) water bath for 45-90 min at
45°C with gentle mixing every 10 min. The quality of the isolated DNA was checked after
running aliquots of DNA samples on a 0.8% (w/v) agarose gel that contained 15 ng µl-1
GelRed. DNA concentration was measured using a NanoDrop ND-1000 spectrophotometer
(Nanodrop Products, Wilmington, NC, USA).

134
5.2.5 Single nucleotide polymorphism genotyping
DNA samples were sent to KBiosciences laboratory Hoddesdon, Herts, UK, for SNP
genotyping. The assays were validated using KBioscience competitive allele-specific PCR
(KASPar) genotyping chemistry. A total of 1 242 SNPs were used. The SNPs were selected
out of the available 1 250 SNPs, of which eight were not able to return data, which would
pass the in-house quality control checks at KBiosciences. The SNP markers were developed
at Cornell University and were converted in partnership with CIMMYT
(http://www.intergratedbreeding.net/snp-marker-conversion). The analysis was carried out in
a 96-well plate format. The system comprised of two components, the assay mix (three
unlabelled primers, the SNP specific component of the system) and the reaction mix (all
other components required including the universal fluorescent reporting system). Two
forward primers, one for each SNP allele and one common reverse primer were used. There
was a forward primer for each SNP allele and each forward primer had a different SNP
sequence at the 3΄-end and a unique unlabelled tail sequence at the 5΄-end and this would
bind to the fluorescently labelled molecular beacon. The genotyping process basically
involved nine steps. The first step involved the assay design using the Primer Picker software
(http://ww.kbioscience.co.uk/primer-picker/). This was followed by sample arraying in a
microtitre 96-well PCR plate. PCR reactions were set up in a total volume of 8 µl containing
20 ng µl-1 template DNA, 2 µl of reaction mix (mix information proprietary), 0.11 µl of assay
mix (mix information proprietary), 0.026 µl of KTaq polymerase, 0.064 µl MgCl2 and 1.8 µl
of H2O. The concentration of MgCl2 in the reaction mix was 2.2 mM. The combined assay
mix and reaction mix were dispensed over the DNA samples using a liquid dispenser. A
fusion laser welding system was used to seal the plates. The PCR cycling was done in a
KBioscience “Duncan” water bath cycler. A hot start activation that lasted for 15 min was
done at 94°C. Two cycling steps followed where the first consisted of 20 cycles at 94°C for
10 sec, 57°C for 5 sec and 72°C for 10 sec. The second cycle consisted of 18 cycles at 94°C
for 10 sec, 57°C for 20 sec and 72°C for 40 sec. The plates were read using a fluorescence
resonance energy transfer (FRET) plate reader. The fluorophores 6-carboxyfluorescein
(FAM) and 2΄chloro-7΄-phenyl-1,4-dichloro-6-carboxyfluorescein (VIC) were used for
distinguishing between genotypes and the reference dye, 6-Carboxy-X-Rhodamine,

135
succinimidyl ester (ROX) was used as a passive reference. The FAM and VIC data were
plotted on the X- and Y-axes respectively.

5.2.6 Statistical analysis

ANOVA for morphological traits was done using AGROBASE (2005). Pearson correlation
coefficients were computed using SPSS version 15.0 for Windows (2006). The genotypic
(σ2g) and phenotypic (σ2p) variances were calculated from expected mean squares of ANOVA
using the formula suggested by Hallauer and Miranda (1988):

Genotypic variance:
σ2g = (MSg-MSe)/r
Where: MSg = mean square of genotypes
MSe = mean square error
r = number of replications

Phenotypic variance:
σ2p = σ2e+σ2g
Where: σ2e = error variance
σ2g = genotypic variance

Genotypic coefficient of variation:

GCV= (√σ2g / X) 100
Where: σ2g = genotypic variance
X = mean of the trait

Phenotypic coefficient of variation:

PCV = (√ σ2p / X) 100
Where: X = mean of the trait
Broad sense heritability was calculated using the following formulae:
H2B = σ2g/ σ2p
Where h2B = broad sense heritability

136
 σ2g = genotypic variance
σ2p = phenotypic variance
Genetic advance:
GA = kσpH2
Where: k = the standardised selection differential at 5% selection (2.063)
σp = phenotypic standard deviation of the character
H2 = heritability estimate

The importance of different traits in explaining multivariate variation was assessed using
PCA in GenStat 14th Edition (2011). KBiosciences SNP viewer version 1.99 was used to
view the SNP data. Powermaker version 3.25 (Liu and Muse, 2005) was used for computing
summary statistics of the genetic data such as number of alleles, number of genotypes, allele
and genotype frequencies, PIC, heterozygosity and GDs from allele frequencies. GDs based
on morphological data were calculated using the Euclidean dissimilarity coefficient
(Kaufman and Rousseeuw, 1990) using the following formula:

m ni
dE = ∑ ∑ (pij-qij)2
i=1 j=1
√
Where pij and qij = allele frequencies of the jth allele at the ith locus in two individuals under
consideration
ni = number of alleles at the ith locus
m = number of loci

GDs based on SNP data were calculated using the Rogers’ dissimilarity coefficient using the
following formula:

m ni
dR = 1/m ∑ √ ½ ∑ (pij-qij)2
i=1 j=1
√
Where pij and qij = allele frequencies of the jth allele at the ith locus in two individuals under
consideration
ni = number of alleles at the ith locus

137
 m = number of loci

Euclidean (Kaufman and Rousseeuw, 1990) dissimilarity coefficients calculated in SPSS

version 15.0 for Windows (2006) using morphological data were subjected to cluster analysis
using the unweighted pair-group method with arithmetic average algorithm (UPGMA) using
sequential agglomerative hierarchical nested cluster analysis (SAHN) programmes in
NTSYSpc 2.21n software (Rohlf, 1993). On the other hand Rogers’ (Rogers, 1972)
dissimilarity coefficients calculated in Powermaker version 3.25 (Liu and Muse, 2005) based
on SNP data were subjected to cluster analysis using the weighted method with neighbour-
joining (Saitou and Nei, 1987) in NTSYSpc 2.21n. Relationships among inbred lines were
visualised in dendrograms. The goodness of fit of the clustering matrices was calculated
using the COPH and MXCOMP programmes in NTSYSpc. To further visualise the
relationship among the inbred lines PCA was done using DARwin version 5.0.158 software
(Perrier and Jacquemould-Collet, 2006).

5.3 Results

5.3.1 Performance of inbred lines as measured using morphological traits

Combined ANOVAs were done within years and across years (Tables 5.1-5.3). Lines were
significantly different for all measured traits in 2009/10 and 2010/11 and across the two years
except for stem lodging. Environment x line was significant for all other traits in the 2009/10
season except stem lodging, whereas in the 2010/11 season it was only significant for
anthesis days, plant height, ear height, root lodging, ears per plant and ear rot. Environments
were significantly different for all traits within years and across years. Environment x line
was significant for all traits except stem lodging (Table 5.3). There was considerable
variability amongst lines as indicated by differences between the minimum and maximum
values for all traits (Table 5.4). Lines generally showed high prolificacy (ears per plant) of
close to 1 and CML539 had the highest value of 1.36 (Figure 5.1) but this did not translate to
very high yield. The CIMMYT line CML548 had the highest yield of 2.91 t ha-1 and the
lowest yielder was SC5522 with 0.51 t ha-1 (Table 5.4).

138
Table 5.1 Mean squares for grain yield and other morphological traits across five sites in the 2009/10 season

Source DF GYD AD ASI PH EH EPO RL SL EPP ER

Site 4 50.257*** 7759.176*** 570.010*** 72396.516*** 15361.314**** 0.294*** 56289.468*** 4156.685*** 3.538*** 1331.052***
Rep within site 5 0.291 411.872 32.024 1254.612* 141.240 1.06 849.766 42.049 0.106 70.853
Line 24 5.284*** 1640.426*** 179.675*** 3571.819*** 1134.714*** 0.046*** 664.169* 94.286 0.211*** 575.432***
Site x Line 96 1.007*** 1105.274*** 173.252*** 823.974* 255.795** 0.014*** 558.545* 92.486 0.101*** 545.878***
Residual 120 0.389 356.464 15.132 526.962 161.315 0.006 384.174 72.826 0.049 195.883
LSD (0.05) 0.553 16.718 3.444 20.326 11.246 0.068 17.355 7.556 0.196 12.392
SED 0.279 8.444 1.740 10.266 5.680 0.034 8.766 3.816 0.099 6.259
***P≤0.001; **P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem
lodging; EPP=ears per plant; ER=ear rot; LSD=least significant difference; SED=standard error difference; DF=degrees of freedom; Rep=replication.

Table 5.2 Mean squares for grain yield and other morphological traits across five sites in the 2010/11 season

Source DF GYD AD ASI PH EH EPO RL SL EPP ER

Site 4 14.983*** 13306.760*** 219.586*** 49869.996*** 18165.796*** 0.092*** 8235.236*** 6740.890*** 2.958*** 1821.460***
Rep within site 5 3.261 54.064*** 2.988 513.856* 175.668 0.004 807.542*** 33.644 0.156*** 67.915
Line 24 5.310*** 186.000*** 40.119*** 2750.149*** 1219.998*** 0.022*** 806.932*** 93.100 0.299*** 416.727***
Site x Line 96 2.198 13.652*** 17.642 536.634*** 276.671*** 0.006 312.844*** 83.304 0.100*** 172.149*
Residual 120 1.975 5.314 13.780 200.973 114.393 0.004 152.593 86.389 0.034 119.062
LSD (0.05) 1.245 2.041 3.287 12.553 9.470 0.059 10.938 8.230 0.163 9.662
SED 0.629 1.031 1.660 6.340 4.783 0.030 5.524 4.157 0.082 4.880
***P≤0.001; **P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem
lodging; EPP=ears per plant; ER=ear rot; LSD=least significant difference; SED=standard error difference; DF=degrees of freedom; Rep=replication.

139
Table 5.3 Mean squares for grain yield and other traits in the 2009/10 and 2010/11 seasons

Source DF GYD AD ASI PH EH EPO RL SL EPP ER

Site 4 22.562*** 20507.048*** 698.203*** 65054.837*** 21803.078*** 0.340*** 17132.156*** 4920.383*** 4.470*** 2085.520***
Line 24 6.569*** 973.506*** 122.220*** 5141.323*** 1733.709*** 0.037*** 1105.932*** 136.260* 0.446*** 753.513***
Rep within site x
year 4 4.441** 582.420* 43.765* 2210.585*** 396.135* 0.013* 2071.635*** 94.616 0.327*** 173.459
Year 1 1.004 3931.208*** 208.658*** 19158.050*** 13302.482*** 0.173*** 20446.734*** 155.57 0.354** 1699.799**
Site x Line 96 1.899** 528.737*** 92.558*** 684.587*** 301.929*** 0.009*** 439.824*** 88.993 0.106*** 307.768***
Site x Year 4 42.678*** 558.888* 91.393*** 57211.675*** 11724.032*** 0.046*** 47392.547*** 5977.193*** 2.026*** 1066.992***
Line x Year 24 4.025*** 852.921*** 97.575*** 1180.646*** 621.003*** 0.030*** 365.169 51.127 0.064* 238.646*
Site x Line x Year 96 1.306 590.189*** 98.336*** 676.021*** 230.537** 0.010*** 431.566** 86.798 0.095*** 410.259***
LSD (0.05) 0.669 8.274 2.340 11.737 7.223 0.044 10.078 5.489 0.125 7.720
SED 0.340 4.201 1.188 5.958 3.667 0.022 5.117 2.787 0.064 3.919
***P≤0.001; **P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem
lodging; EPP=ears per plant; ER=ear rot; LSD=least significant difference; SED=standard error difference; Rep=replication.

140
Table 5.4 Mean performance of maize inbred lines for 14 traits evaluated in the
2009/10 and 2010/11 seasons

Line GYD AD ASI PH EH EPO RL EPP ER GLS RUST SEN TEX EA

2Kba 1.17 75.7 2.1 151.2 95.7 0.42 31.9 0.75 13.8 2.4 1.9 0.6 5.2 5.0
2N3d 1.98 77.9 3.0 192.3 82.2 0.43 39.3 0.79 27.5 2.6 1.8 3.8 3.9 4.4
CML312 1.68 78.1 2.6 164.6 73.7 0.32 30.0 0.57 13.4 2.9 1.6 3.8 2.8 3.1
CML545 2.58 69.6 -1.2 162.5 83.1 0.47 23.4 0.91 7.3 2.6 1.6 3.9 3.9 3.8
CML395 2.39 78.9 1.7 204.9 98.9 0.49 16.2 0.85 22.3 2.0 1.5 3.2 3.7 4.0
CML544 1.95 75.1 -0.1 141.1 68.6 0.43 17.3 0.98 6.5 2.1 1.3 3.7 3.9 4.1
CML442 2.27 78.8 3.8 164.9 71.3 0.40 19.3 1.02 13.8 2.0 2.0 3.4 4.7 4.9
CML444 2.58 81.6 -0.3 171.2 97.9 0.51 24.4 0.84 2.8 3.3 1.7 3.2 3.0 3.4
CZL03007 1.91 75.5 -0.9 150.1 70.7 0.40 45.7 0.95 5.5 2.0 1.7 3.0 2.2 2.9
CML536 1.81 82.8 0.6 185.3 89.2 0.39 20.6 0.85 4.3 2.2 2.6 2.8 2.2 2.8
CML537 2.32 77.9 1.4 171.7 74.9 0.39 17.7 0.71 8.2 3.0 1.5 3.1 2.8 3.2
CML538 2.37 74.3 -0.6 168.9 79.6 0.42 14.9 0.88 11.7 2.5 1.3 3.5 3.4 4.0
CML539 1.99 71.3 -0.1 134.3 68.7 0.43 29.9 1.36 2.4 2.3 1.5 3.5 3.4 3.8
K64r 1.37 75.7 0.6 129.2 61.9 0.45 35.9 0.82 30.6 2.1 1.9 4.4 4.4 4.7
N3.2.3.3 1.52 79.0 2.3 180.7 90.6 0.51 44.2 0.59 8.5 3.4 1.8 4.0 4.4 4.7
NAW5885 1.30 78.3 6.6 165.9 83.3 0.41 39.3 0.63 21.0 2.6 1.4 3.1 3.2 3.2
RA214P 1.75 80.6 3.9 194.1 92.4 0.45 20.1 0.60 30.8 2.1 1.3 2.9 3.2 3.8
RS61P 2.53 74.8 1.5 146.0 79.0 0.47 27.6 0.86 8.5 2.5 1.3 2.9 3.3 3.5
SC5522 0.51 80.7 3.6 194.7 95.7 0.44 45.2 0.62 22.1 2.3 1.6 4.6 2.6 2.8
SV1P 1.05 62.9 -0.1 133.3 71.9 0.43 32.5 0.91 1.9 2.7 3.1 4.0 3.4 4.1
WCOBY 1.19 75.0 1.5 142.9 71.5 0.48 28.0 0.70 19.8 2.9 1.9 3.3 3.1 3.6
CML548 2.91 76.5 2.0 159.7 74.5 0.42 22.4 0.90 16.1 2.6 1.7 3.4 2.5 3.1
CZL052 1.52 73.0 -0.1 164.1 78.2 0.44 45.5 0.76 3.8 2.4 1.4 3.5 3.1 4.1

Mean 1.88 76.3 1.5 163.6 80.4 0.43 28.8 0.83 12.4 2.5 1.8 3.4 3.4 3.8
Min 0.51 62.9 -1.2 129.2 61.9 0.32 14.9 0.57 1.9 2.0 1.3 0.6 2.2 2.8
Max 2.91 82.8 6.6 204.9 98.9 0.51 45.7 1.36 30.8 3.4 3.1 4.6 5.2 5.0
LSD (0.05) 0.49 1.8 1.7 13.4 14.2 0.05 11.6 0.17 15.9 0.8 0.5 0.8 0.9 0.9
MSE 0.34 4.3 3.5 210.7 94.1 0.00 189.2 0.04 177.0 0.2 0.1 0.3 0.7 0.7
GYD=grain yield (t ha-1); AD=anthesis days; ASI=anthesis silking interval (days); PH=plant height (cm); EH=ear height (cm);
EPO=ear position (0-1); RL=root lodging (%); EPP=ears per plant (#); ER=ear rot (%); GLS=grey leaf spot (1-5); RUST=common
rust (1-5); SEN=senescence (1-10); TEX=texture (1-5); EA=ear aspect (1-5); LSD=least significant difference; MSE=mean square
error; Min=minimum; Max=maximum.

In terms of maturity SV1P was the earliest (62.9 days) and the latest was CML536 (82.8
days). All lines exhibited good anthesis silking interval values except NAW5885, which had
an anthesis silking interval value of 6.6. Lines generally had foliar disease scores of less than
3 except CML444 and N3.2.3.3, which had GLS scores of 3.3 and 3.4 respectively. The rust
scores were lower than the GLS scores (Figure 5.2), however SV1P was the only line with a
high rust score of 3.1. Lines showed high ear rot percentages and this resulted in poor ear
aspect scores (Table 5.4). K64r (30.6) and RA214P (30.8) had the highest ear rot scores. The
texture of the lines ranged from semi-flint to dent.

141
 3.50

3.00

2.50
Grain yield tha-1

2.00

1.50
GYD
1.00 EPP

0.50

0.00
2N3d

N3.2.3.3
2Kba

K64r

RA214P

SV1P
CML312
CML545
CML395
CML544
CML442
CML444
CZL03007
CML536
CML537
CML538
CML539

NAW5885

SC5522

WCOBY
CML548
CZL052
RS61P
Line name

Figure 5.1 Grain yield performance and ears per plant for the lines across seasons.
GYD=grain yield; EPP=ears per plant

35.0

30.0

25.0
Disease score/%

20.0

15.0 ER

10.0 GLS
P SORG
5.0

0.0
2Kba

K64r

RA214P
2N3d
CML312
CML545
CML395
CML544
CML442
CML444
CZL03007
CML536
CML537
CML538
CML539

N3.2.3.3
NAW5885

SC5522

WCOBY
CML548
CZL052
RS61P

SV1P

Line name

Figure 5.2 Response of lines to ear rot and foliar diseases across seasons.
ER=Exserohilum turcicum; GLS=grey leaf spot; P SORG=Puccinia sorghi.

142
The genetic and phenotypic variances as well as heritability estimates for the lines are
presented in Table 5.5. Broad sense heritability estimates calculated showed that half of the
traits were highly heritable with more than 50% heritability. Grain yield had the highest
heritability estimate of 0.81 and stem lodging had the lowest heritability estimate of 0.02.
Other traits with high heritability estimates (>0.50) were plant height, ear height, ear position
and ears per plant (Table 5.5). Plant height had the largest genetic variance (274.8) whilst ear
position had the smallest genetic variance (0.003). Phenotypic variances were higher than
genetic variances for anthesis days, anthesis silking interval, plant and ear height, root and
stem lodging, ears per plant and ear rot. Traits with high genotypic coefficients of variation
included grain yield, anthesis days, anthesis silking interval, plant height, ear height, root
lodging and ear rot (Table 5.6). Anthesis silking interval had the highest phenotypic
coefficient of variation (34.6) whilst ear position had the lowest value (10.3). Plant (16.30)
and ear (9.14) height had high genetic advance values compared to other traits. Grain yield
had the highest genetic advance (25.0%) as a percentage of the mean, whilst ear position had
the lowest value (1.1%).

PCA computed using the correlation matrix grouped the 14 traits into 14 components, which
accounted for 100% of the variability present among the lines evaluated. The first nine PCs
explained 94% of the total variation, whilst the five eigenvectors with eigenvalues greater
than one accounted for 74.1% of the entire variability available among the inbred lines
(Table 5.7). The first PC that explained 26.2% of the variation among genotypes was mainly
attributed to variation in anthesis days, anthesis silking interval, plant and ear height, stem
lodging, ears per plant and ear rot. The second PC that accounted for 17.7% of the variation
was dominated by grain yield, anthesis silking interval and root and stem lodging.

In the third PC (12.1%) GLS, texture, ear position, ear height, anthesis silking interval, root
lodging, ear rot and common rust were the most important traits. Ear position, ear aspect, ear
rot and GLS were important delineating traits associated with the fourth PC, which accounted
for 9.7% of the total variation. The fifth PC which explained 8.4% of the total variation was
associated with variation due to senescence, common rust, ear position, ear height, ear rot
and grain yield. Each trait was found to be an important source of variation in the PCA.

143
Table 5.5 Genetic and phenotypic variances and heritability estimates

Trait Genetic variance Phenotypic variance Heritability

GYD 0.428 0.528 0.81

AD 53.515 164.043 0.33
ASI 0.642 17.968 0.04
PH 274.785 357.189 0.77
EH 87.892 113.471 0.78
EPO 0.003 0.005 0.71
RL 10.562 66.417 0.16
SL 0.180 9.429 0.02
EPP 0.011 0.021 0.52
ER 2.955 57.543 0.05
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root
lodging; SL=stem lodging; EPP=ears per plant; ER=ear rot.

Table 5.6 Estimates of genotypic and phenotypic coefficients of variation and genetic
advance of the maize inbred lines across all environments in the 2009/10 and
2010/11 seasons

Trait GCV (%) PCV(%) GA GA (% of mean)

GYD 47.7 53.0 0.47 25.0
AD 83.8 14.7 5.75 7.5
ASI 65.4 34.6 0.14 9.3
PH 12.9 14.8 16.30 10.0
EH 10.5 11.9 9.14 11.4
EPO 8.8 10.3 0.05 1.1
RL 60.6 15.2 2.89 10.0
EPP 11.5 15.9 0.11 13.3
SL 12.7 91.8 0.16 1.4
ER 48.9 21.5 0.65 5.2
GCV=genotypic coefficient of variation; PCV=phenotypic coefficient of variation; GA=genetic advance; GYD=grain yield;
AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem
lodging; EPP=ears per plant; ER=ear rot.

5.3.2 Correlation coefficients among morphological traits

Pearson coefficient correlations for the traits are presented in Table 5.8. Significant
correlation coefficients are highlighted in bold. Grain yield was negatively and significantly
(P≤0.01) correlated with root lodging (r=-0.613) and positively correlated (P≤0.05) with ears
per plant (r=0.415). Correlation of grain yield with anthesis silking interval, ear height, stem
lodging, ear rot, rust, senescence, texture and ear aspect was negative and non-significant.

144
Table 5.7 Eigenvectors, eigenvalues and individual and cumulative percentage of
variation explained by first nine principal components for 14 morphological
traits of maize inbred lines
Eigenvectors
Variables PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9
Grain yield 0.02 0.54 -0.12 -0.04 0.21 -0.03 0.01 0.41 0.11
Anthesis days -0.41 0.02 -0.18 -0.10 0.00 0.23 -0.02 0.49 -0.05
Anthesis silking interval -0.33 -0.29 -0.23 0.03 -0.05 -0.23 0.02 0.33 -0.01
Plant height -0.44 0.07 0.01 -0.03 0.08 0.30 0.25 -0.08 0.45
Ear height -0.40 0.10 0.25 0.17 -0.22 0.37 -0.03 -0.13 0.08
Ear position -0.06 0.10 0.46 0.45 0.33 0.26 -0.10 0.07 -0.47
Root lodging 0.03 -0.49 0.23 -0.08 0.05 0.22 -0.42 0.04 0.30
Stem lodging 0.30 -0.40 -0.14 0.02 0.03 0.32 -0.10 0.40 -0.04
Ears per plant 0.38 0.24 -0.19 0.16 0.05 0.34 0.01 0.23 0.10
Ear rot -0.27 -0.26 -0.31 0.29 0.29 -0.17 0.17 -0.04 -0.36
Grey leaf spot -0.09 0.04 0.59 -0.26 0.02 -0.42 -0.05 0.42 -0.01
Common rust 0.17 -0.18 0.23 0.03 -0.42 0.13 0.74 0.18 -0.12
Senescence 0.10 -0.17 0.12 -0.23 0.72 0.05 0.41 -0.09 0.21
Ear aspect 0.09 -0.06 0.07 0.72 -0.01 -0.33 0.04 0.14 0.52
Grain texture 0.06 0.21 -0.61 0.04 -0.01 0.12 -0.01 0.14 -0.18
Eigenvalue 3.67 2.47 1.69 1.36 1.18 0.20 0.82 0.55 0.44
Individual percentage variation
explained 26.2 17.7 12.1 9.7 8.4 6.8 5.9 4.0 3.2
Cumulative percentage variation
explained 26.2 43.9 56.0 65.7 74.1 80.9 86.8 90.8 94.0
PC= principal component.

Anthesis days were positively and significantly correlated with anthesis silking interval
(r=0.489) and plant (r=0.662) and ear (r=0.499) height, whilst on the other hand it was
significantly and negatively correlated with ears per plant (r=-0.430). The largest significant
and positive correlation was between grain texture and ear aspect (r=0.938). Anthesis silking
interval was significantly and negatively correlated with ears per plant (r=-0.527) as well as
positively and significantly correlated with ear rot (0.617). Correlation between plant height
and ear height was significant and positive (r=0.753). Stem lodging and ears per plant were
significantly and negatively correlated with plant and ear height. Root and stem lodging were
significantly and positively correlated with each other. Ears per plant were significantly and
negatively correlated with ear rot. GLS, rust, ear position and senescence were not
significantly correlated with any of the morphological traits.

145
Table 5.8 Pearson coefficient correlations for grain yield and other morphological traits measured from the inbred lines
in the 2009/10 and 2010/11 seasons

Trait GYD AD ASI PH EH EPO RL SL EPP ER GLS RUST SEN TEX

AD 0.073

ASI -0.307 0.489*

PH 0.089 0.662** 0.410

EH -0.046 0.499* 0.266 0.753**

EPO 0.086 -0.024 -0.173 0.099 0.390

RL -0.613** -0.087 0.161 -0.083 -0.017 0.047

SL -0.386 -0.259 -0.046 -0.493* -0.479* -0.168 0.546**

EPP 0.415* -0.430* -0.527** -0.468* -0.421* 0.007 -0.269 0.354

ER -0.234 0.0377 0.617** 0.344 0.129 0.060 0.050 -0.004 -0.450*

GLS 0.039 0.003 0.003 0.024 0.164 0.231 0.127 -0.274 -0.388 -0.263

RUST -0.321 -0.283 -0.139 -0.191 -0.088 -0.104 0.090 0.300 0.093 -0.213 0.072

SEN -0.090 -0.150 -0.086 0.020 -0.330 0.108 0.206 0.218 0.006 0.105 0.104 0.028

TEX -0.085 -0.170 0.110 -0.170 0.023 0.276 -0.037 0.074 0.086 0.203 -0.064 0.025 -0.185

EA -0.082 -0.245 -0.008 -0.182 -0.070 0.291 0.001 0.070 0.133 0.165 -0.087 0.104 -0.117 0.938**
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; Ear height; EPO=ear position; RL=root lodging; SL=stem lodging; EPP=ears
per plant; ER=ear rot; GLS=grey leaf spot; RUST=common rust; SEN=senescence; TEX=texture; EA=ear aspect.

146
5.3.3 Genetic distances and heterotic grouping among lines based on morphological
data
Estimates of genetic distances based on morphological data expressed as Euclidean distances
for all 253 pairwise comparisons averaged 0.37 and ranged from 0.109 between CML538
and CML537 to 0.911 between K64r and CML395 (Table 5.9). Low genetic distances were
observed amongst the inbred lines CML538 and CML537 (0.109), CML442 and CML312
(0.126), RS61P and WCOBY (0.146), CML537 and CML442 (0.160), CML312 and
CML545 (0.169), CML442 and CML548 (0.175), CML545 and CML537 (0.179), RS61P
and CML548 (0.180), CML538 and CML545 (0.182), SV1P and CML539 (0.183), CML312
and CML537 (0.188) and WCOBY and CML548 (0.185). Most of these lines are related by
pedigree and they belong to the same heterotic groups. High genetic distances were observed
between K64r and CML395 (0.911), CML395 and SV1P (0.846), CML395 and CML539
(0.852), RA214P and K64r (0.808), RA214P and CML539 (0.790), RA214P and SV1P
(0.762), SC5522 and K64r (0.756), SC5522 and SV1P (0.726), CML395 and CML544
(0.725), CML395 and CZL03007 (0.724), CML539 and SC5522 (0.721), SC5522 and
CML544 (0.715) and K64r and CML536 (0.719).

The dendrogram obtained from UPGMA cluster analysis using the Euclidean dissimilarity
coefficient classified the inbred lines into five main clusters (Figure 5.3). The dendrogram
exhibited a poor goodness of fit with a matrix correlation of r=0.70. The total detected
morphological variation between the 23 inbred lines was a dissimilarity of 53%. Generally
grouping of lines was not in line with the known CIMMYT A and B and DR&SS SC and N3
heterotic grouping. In main groups I and II lines basically did not group according to already
known heterotic groupings. Group I clustered separately from the other groups, indicating
that the four lines in this group were the most distantly related from the other lines, based on
morphological data. Group II also clustered separately indicating that the three lines differed
significantly from the other lines.

147
 Table 5.9 Estimates of genetic distances based on Euclidean distances and morphological data for all pair-wise
comparisons of 23 inbred lines

LINE 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

2Kba 1 0.462 0.264 0.226 0.578 0.349 0.317 0.300 0.319 0.396 0.359 0.356 0.397 0.486 0.361 0.276 0.535 0.256 0.507 0.397 0.335 0.328 0.356
2N3d 2 0.336 0.409 0.331 0.619 0.387 0.423 0.506 0.329 0.384 0.421 0.686 0.695 0.277 0.311 0.316 0.541 0.250 0.697 0.558 0.441 0.434
CML312 3 0.169 0.510 0.294 0.126 0.315 0.255 0.303 0.188 0.246 0.390 0.464 0.303 0.209 0.455 0.256 0.453 0.408 0.293 0.213 0.281
CML545 4 0.493 0.273 0.200 0.235 0.321 0.287 0.179 0.182 0.404 0.537 0.325 0.272 0.457 0.246 0.496 0.382 0.325 0.238 0.305
CML395 5 0.725 0.510 0.404 0.724 0.308 0.438 0.436 0.852 0.911 0.422 0.497 0.213 0.671 0.384 0.846 0.714 0.555 0.611
CML544 6 0.279 0.438 0.336 0.503 0.319 0.320 0.295 0.420 0.540 0.412 0.649 0.223 0.715 0.279 0.243 0.286 0.421
CML442 7 0.349 0.323 0.297 0.160 0.228 0.384 0.466 0.370 0.273 0.456 0.257 0.491 0.419 0.287 0.175 0.340
CML444 8 0.462 0.223 0.257 0.241 0.575 0.705 0.279 0.319 0.383 0.383 0.449 0.543 0.471 0.356 0.371
CZL03007 9 0.482 0.392 0.441 0.252 0.369 0.387 0.305 0.661 0.231 0.552 0.279 0.278 0.317 0.227
CML536 10 0.226 0.277 0.594 0.719 0.255 0.339 0.336 0.431 0.350 0.605 0.509 0.347 0.382
CML537 11 0.109 0.479 0.587 0.331 0.284 0.379 0.309 0.478 0.467 0.355 0.203 0.332
CML538 12 0.515 0.610 0.368 0.303 0.357 0.326 0.514 0.475 0.361 0.228 0.371
CML539 13 0.304 0.573 0.466 0.790 0.227 0.721 0.183 0.272 0.365 0.401
K64r 14 0.663 0.488 0.808 0.367 0.756 0.367 0.280 0.430 0.516
N3.2.3.3 15 0.219 0.385 0.408 0.234 0.556 0.474 0.360 0.238
NAW5885 16 0.388 0.282 0.348 0.442 0.296 0.224 0.224
RA214P 17 0.588 0.368 0.762 0.597 0.450 0.534
RS61P 18 0.568 0.208 0.146 0.180 0.267
SC5522 19 0.726 0.607 0.485 0.418
SV1P 20 0.243 0.349 0.358
WCOBY 21 0.185 0.327
CML548 22 0.272

148
 2Kba
CML312

V
CML442
CML545
CML537
CML538
CML544
RS61P
WCOBY
CML548
CZL03007
CZL052

IV
N3.2.3.3
NAW5885
CML444

III
CML536
CML539
SV1P
II K64r
2N3d
SC5522
I

CML395
RA214P
0.60 0.50 0.40 0.30 0.20 0.10 0.00
Euclidean dissimilarity coefficient

Figure 5.3 Unweighted pair-group method with arithmetic average algorithm cluster analysis of 23 maize inbred lines
based on morphological data combined over two seasons and seven locations.
Dark blue=A group; Light blue=N3 group; Red=B group; Orange=SC group

149
Most of the lines (16) clustered together in groups III, IV and V. Lines clustered mainly
according to heterotic groups in the subgroups of groups IV and V. Although group IV
consisted of four lines belonging to two different heterotic groups, lines paired according to
B and N3 groups in the two subgroups. Group V was the biggest group with 10 lines and
contained subgroups in which the lines grouped according to the known heterotic groupings.
However CML548 belonging to heterotic group B, clustered with three lines belonging to
heterotic groups B and SC. Five CIMMYT lines belonging to heterotic group A (CML312,
CML442, CML545, CML537 and CML538) clustered together in the same subgroup and
were also closely related based on pedigree data. Morphologically the most similar lines were
CML537 and CML538 [MAS(CML206/CML312)-23-2-1-1-B and ZM621A-10-1-1-1-2-B]
with 89% similarity followed by CML312 and CML442 [S89500-F2-2-2-1-1-B and
(M37W/ZM607-#-B-F37SR-2-3SR-6-2-X)-8-2-X-1-B] with a similarity of 87% and RS61P
and WCOBY (share a similar parent SC5522) with a similarity of 85%.

On the other hand grouping in terms of morphological traits followed a certain trend. The
four lines in group I (RA214P, CML395, SC5522 and 2N3d) all had mean plant heights of
above 190 cm, with high ear placement. These lines were also late maturing with a mean
number of days to anthesis of above 75 days and high ear rot percentages. Group II (K64r,
SV1P and CML539) was generally characterised by early to medium maturing and shorter
plants of semi flint to semi dent texture. Lines in group III were tall, late maturing and had a
similar number of ears per plant. The sub-groups within group IV consisted of lines with
similar anthesis silking interval values, where NAW5885 and N3.2.3.3 had positive anthesis
silking interval values whilst CZL052 and CZL03007 had negative anthesis silking interval
values. Finally in group V CML537 and CML538 had similar grain yield means, whilst
CML442 and CML312 had similar days to anthesis, plant heights and ear rot scores. RS61P
and WCOBY had similar GLS scores and their anthesis silking interval values were the same
(1.5). The outlier lines did not pair with other lines because of the unique morphological
characteristics that they exhibited. CML548 was the highest yielder (2.91 t ha-1) amongst the
lines, whilst K64r had the second highest ear rot score (30.6). CML545 had the best anthesis
silking interval value (-1.2) and 2Kba had a poor ear aspect score (5.0).

150
5.3.4 Single nucleotide polymorphism performance and quality
The data for all individuals tested was presented as a graph using the SNP viewer and the
graph indicated the colour of fluorescence and the identified genotype as shown in Figure
5.4, where three clearly defined groups can be seen namely homozygotes G:G (red),
homozygotes C:C (blue), heterozygotes G:C (green) while the genotypes indicated in pink
were outliers that did not cluster clearly with one of the identified groups. All SNP data (23
lines by 1 242 SNPs) were initially scored. SNP markers that were monomorphic (26
markers) or had more than 20% missing data points (87 markers) were removed from further
analyses. As a result, a total of 1 129 SNPs (91%) were included in the analysis of which 297
SNPs had 10-20% missing data whilst 832 had less than 10% missing data.

Figure 5.4 Example of information extracted from each single nucleotide polymorphism
marker using the single nucleotide polymorphism viewer. Data presented is
for single nucleotide polymorphism marker PHM12749_13, which detects a
C/G single nucleotide polymorphism in the maize genome.

151
In the 23 inbred lines, 2 258 alleles were detected at 1 129 loci with two alleles per locus. An
even distribution of minor allelic frequency was observed with continued classes from 0.01
to 0.50 (Figure 5.5 and Appendix 9). Only 9.5% (107/1 129) of the SNPs had a minor allele
frequency of less than 0.05, whilst approximately 53.3% (602/1 129) of the markers had a
minor allele frequency of more than 0.20. In addition 123, (10.9%) of the markers had a
minor allele frequency close to 0.50 thereby showing almost equal allele frequencies for the
two alternative alleles. The average PIC was 0.252 and ranged from 0.024 to 0.375 with a
peak distribution between 0.311 and 0.375 (Figure 5.6 and Appendix 10). About 41.7% of
the markers had PIC values greater than 0.30. The estimated gene diversity ranged from
0.025 to 0.500 with an average of 0.312 (data not shown). SNPs were well dispersed across
the 10 chromosomes and the coverage ranged from 65 on chromosome 7 to 206 on
chromosome 1 (Table 5.10). The inbred lines exhibited heterozygosity at 0.09% of the
genetic loci. The number of heterozygous loci per inbred line was determined. There were
four inbred lines with a large number of heterozygous loci and these were 2N3d (427),
CML537 (403), WCOBY (358) and CZL052 (312) (Table 5.11). There was no line with
100% homozygous loci, however CML442 had 99.29% homozygous loci.

300

250
Number of SNPs

200

150

100

50

0
0.0-0.10 0.11-0.20 0.21-0.30 0.31-0.40 0.41-0.50
Frequency of minor allele

Figure 5.5 Frequency distribution of minor alleles among 23 inbred lines based
on 1 129 single nucleotide polymorphism (SNP) markers.

152
5.3.5 Genetic distances and heterotic grouping of lines based on single nucleotide
polymorphism markers
The Rogers dissimilarity coefficient based on SNP data revealed that low GDs were recorded
amongst lines. Distances ranged from 0.11 to 0.38 with a mean of 0.32 (Table 5.12). Lines
with the lowest genetic distances were SC5522 and CML548 (0.108), SV1P and WCOBY
(0.143), 2N3d and WCOBY (0.172), 2Kba and N3.2.3.3 (0.176), CML537 and WCOBY
(0.192) and RA214P and CML544 (0.195).

300
250
Number of SNPs

200
150
100
50
0

PIC value

Figure 5.6 Polymorphic information content (PIC) among 23 inbred lines based
on 1 129 single nucleotide polymorphism (SNP) markers.

Table 5.10 Distribution of single nucleotide polymorphism markers over the 10

maize chromosomes
Chromosome number Number of SNPs
1 206
2 121
3 139
4 120
5 130
6 89
7 65
8 102
9 83
10 74
Total 1 129
SNP=single nucleotide polymorphism

153
Table 5.11 Number of heterozygous loci and percentage homozygosity of maize inbred
lines

Name of inbred line Origin Number of heterozygous loci % homozygosity

2Kba DR&SS 29 97.43

2N3d DR&SS 427 62.18
CML312 CIMMYT 28 97.52
CML545 CIMMYT 72 93.62
CML395 CIMMYT 32 97.17
CML544 CIMMYT 31 97.25
CML442 CIMMYT 8 99.29
CML444 CIMMYT 91 91.94
CZL03007 CIMMYT 17 98.49
CML536 CIMMYT 15 98.67
CML537 CIMMYT 403 64.30
CML538 CIMMYT 27 97.61
CML539 CIMMYT 14 98.76
K64r DR&SS 13 98.85
N3.2.3.3 DR&SS 65 94.24
NAW5885 DR&SS 27 97.61
RA214P DR&SS 20 98.23
RS61P DR&SS 19 98.32
SC5522 DR&SS 12 98.94
SV1P DR&SS 55 95.13
WCOBY DR&SS 358 68.29
CML548 CIMMYT 14 98.76
CZL052 CIMMYT 312 72.36

The highest recorded distances were between CML536 and 2N3d (0.384), CZL03007 and
CML536 (0.381), CML395 and CML536 (0.378), CML544 and CML536 (0.376), CML539
and CML536 (0.376), CML536 and CML442 (0.375) and CML538 and CML536 (0.374).

Neighbour-joining cluster analysis in NTSYS grouped the 23 lines into two major groups
(Figure 5.7). The dendrogram showed a good goodness of fit (r=0.87). There was no clear
separation of lines according to their origin (CIMMYT and DR&SS). The major groups did
not show grouping according to the known heterotic groups. However, within some of the
subgroups of the main groups there was consistency with the known heterotic groups. Group
I consisted of two lines that both belong to similar heterotic groups (SC and B). Group II was
divided into two subgroups A and B.

154
Table 5.12 Estimates of genetic distances based on single nucleotide polymorphism data and Rogers’ distances for all pairwise
comparisons

Line 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
2N3d 2 0.298
CML312 3 0.302 0.289
CML545 4 0.360 0.328 0.286
CML395 5 0.339 0.341 0.319 0.352
CML544 6 0.320 0.335 0.288 0.338 0.340
CML442 7 0.351 0.357 0.302 0.348 0.272 0.300
CML444 8 0.329 0.295 0.265 0.216 0.321 0.297 0.296
CZL03007 9 0.340 0.333 0.289 0.331 0.306 0.308 0.310 0.266
CML536 10 0.358 0.384 0.326 0.237 0.378 0.376 0.375 0.332 0.381
CML537 11 0.349 0.204 0.266 0.326 0.330 0.351 0.345 0.320 0.338 0.364
CML538 12 0.341 0.332 0.296 0.334 0.313 0.253 0.222 0.288 0.306 0.374 0.337
CML539 13 0.350 0.341 0.317 0.339 0.314 0.337 0.331 0.292 0.287 0.376 0.348 0.318
K64r 14 0.275 0.357 0.316 0.361 0.232 0.340 0.270 0.331 0.325 0.366 0.315 0.308 0.332
N3.2.3.3 15 0.176 0.259 0.245 0.327 0.346 0.323 0.364 0.328 0.316 0.350 0.349 0.343 0.361 0.333
NAW5885 16 0.334 0.343 0.299 0.318 0.314 0.301 0.319 0.292 0.290 0.344 0.340 0.308 0.289 0.330 0.339
RA214P 17 0.332 0.350 0.297 0.331 0.326 0.195 0.285 0.283 0.295 0.368 0.337 0.262 0.320 0.335 0.339 0.290
RS61P 18 0.335 0.335 0.283 0.322 0.332 0.309 0.299 0.273 0.275 0.363 0.345 0.293 0.299 0.335 0.339 0.295 0.266
SC5522 19 0.344 0.353 0.312 0.240 0.341 0.326 0.332 0.281 0.306 0.363 0.341 0.306 0.316 0.342 0.338 0.332 0.328 0.304
SV1P 20 0.277 0.280 0.243 0.259 0.290 0.285 0.304 0.276 0.270 0.256 0.272 0.292 0.299 0.279 0.295 0.260 0.285 0.277 0.272
WCOBY 21 0.294 0.172 0.275 0.279 0.342 0.333 0.348 0.246 0.320 0.285 0.192 0.326 0.349 0.327 0.288 0.321 0.328 0.321 0.334 0.143
CML548 22 0.352 0.351 0.303 0.276 0.334 0.321 0.329 0.251 0.306 0.369 0.354 0.310 0.298 0.332 0.357 0.316 0.318 0.291 0.108 0.266 0.338
CZL052 23 0.334 0.239 0.294 0.329 0.325 0.216 0.243 0.269 0.282 0.360 0.276 0.223 0.319 0.332 0.330 0.307 0.203 0.297 0.329 0.292 0.244 0.324

155
 2Kba
N3.2.3.3
CML312
2N3d
WCOBY
b CML537
SV1P
CML395
K64r
CML544
RA214P
B

CZL052
CML442
CML538
CML545
II

CML536
a

CML444
SC5522
CML548
A

CML539
NAW5885
CZL03007
I

RS61P
0.00 0.05 0.10 0.15 0.20 0.25
Rogers' dissimilarity coefficient

Figure 5.7 Neighbour-joining cluster analysis for the 23 maize inbred lines based on Rogers’ dissimilarity coefficient using single
nucleotide polymorphism data.
Dark blue=A group; Light blue=N3 group; Red=B group; Orange=SC group.

156
Group A comprised of two lines belonging to similar groups namely NAW5885 (N3) and
CML539 (A). Group B was further divided into subgroups and later into sub-subgroups. In
subgroup a CML536 and CML545 clustered together and these two lines are related by
pedigree since they share a common parent. Furthermore SC5522 and CML548 in subgroup
a clustered together but these lines belong to opposite groups according to the known
classification. CML442 and CML538, clustering together in subgroup b, were derived from
similar populations. CML395 and K64r clustered together and these lines belong to related
groups B and SC respectively. 2N3d has N3.2.3.3 as one of its parents and the two lines
clustered in the same subgroup and a similar scenario was noted with CML312 and CML537,
where CML537 has CML312 as one of its parents and the two again clustered in the same
subgroup. The rest of the lines did not group according to known heterotic grouping and/or
pedigree. It was noted that there were four lines with a high number of missing data namely
CML545 (33.92%), SV1P (29.05%), WCOBY (28.79%) and 2N3d (27.90%) and since
missing data can skew the data these lines were later excluded from the analysis and a
dendogram consisting of 19 lines was constructed (Figure 5.8). This was done to ascertain if
the grouping was going to be improved without the four lines. The dendrogram also showed
a good goodness of fit (r=0.81). Lines were still divided into two major groups and similar
groupings were maintained. One of the exceptions was for RS61P and CZL03007. In the
initial grouping these lines grouped in the same subgroup but after removal of the four lines,
these lines were still in the same group but did not cluster together in the same subgroup.
Furthermore CML536 did not show any relationship with any of the lines where it previously
grouped together with CML545.

A PCA biplot was further drawn for the 23 inbred lines using the SNP data in order to assess
the grouping pattern of lines (Figure 5.9). Lines were classified into five main groups. There
was still a mixture of different heterotic groupings within each group. However, lines known
to be in the same group appeared closer together within the biplot. Group I consisted of
WCOBY, RS61P, SV1P (SC group) and CML544 (B group). Group II consisted of five lines
namely 2Kba, CML312, CML548, CML442 and CML545.

157
 2Kba

I
N3.2.3.3
CML312

C
CML537
CML395

a
K64r
CML544
II

RA214P
CZL052
D

CML442
CML538
CML444
b

SC5522
CML548
CZL03007
CML539
NAW5885
RS61P
CML536
0.00 0.05 0.10 0.15 0.20
Roger's dissimilarity coefficient

Figure 5.8 Neighbour-joining cluster analysis for the 19 inbred maize inbred lines based on Rogers’ dissimilarity coefficient using
single nucleotide polymorphism data. (Lines with a high percentage missing data excluded from the analysis.)
Dark blue=A group; Light blue=B group; Red=B group; Orange=SC.

158
 IV V

CZL052 CZL03007

III CML548

CML545 I

II CML544

Figure 5.9 Principal component analysis for the 23 maize inbred lines based on single nucleotide polymorphism data.
Dark blue=A group; Light blue=N3 group; Red=B group; Orange=SC group.

159
This group was mainly dominated by lines from group A (CML312, CML548, CML442 and
CML545) and these lines are related by pedigree. The lines that constituted group III were
CML537, CML538, CML536 (group A), CML444 (group B), RA214P (group N3) and
CML395 (group B).

Lines CML537 and CML538 closely grouped together in the biplot and are also in the same
heterotic grouping (A), whilst CML536, also in group A, appeared a little further from the
other two lines. Group IV comprised of N3.2.3.3, 2N3d and NAW5885. Lines 2N3d and
N3.2.3.3 appeared closer to one another on the biplot and these lines are closely related as
2N3d is a direct derivative from N3.2.3.3. These three lines belong to the N3 heterotic group.
Group V comprised of CZL052, CZL03007 (both from group B) and K64r (group SC).

5.3.6 Comparison of dendrograms based on morphological and single nucleotide

polymorphism data
The two dendrograms showed different groupings of lines and no clear grouping according to
known heterotic groups was observed for both dendrograms. However, the morphological
dendrogram showed more consistency based on known heterotic groups compared to the
SNP dendrogram. This may be mainly because lines were initially grouped using testers and
the grouping was mainly based on field data. The grouping in the morphological dendrogram
corresponded with the morphological data of the lines and in some cases lines related by
pedigree grouped together. In the morphological dendrogram lines CML312, CML442,
CML545, CML537 and CML538 clustered in the same group and these lines belong to group
A, whereas in the SNP dendrogram these lines were split into different groups although some
of them still appeared together within those groups. The grouping in the SNP dendrogram
showed that it was mostly consistent with the pedigree relationships amongst the lines as
revealed by close clustering of CML312 and CML537. CML537 has CML312 as one of its
parents hence its close relationship with the line. Lines CML442 and CML538 clustered
together in the SNP dendrogram and these two lines were derived from similar populations.
The grouping in the SNP dendrogram did not seem to have any relationship with the
morphological traits of the lines.

160
5.4 Discussion
Existence of genetic variability in any given germplasm is important as it facilitates the
process of selection and superior germplasm can be easily identified. The significant
morphological differences between lines for the tested traits were an indication that there
existed variability amongst the 23 tested lines. However the most ideal sample for the best
estimate of genetic variability is at least more than 30 lines. Morphological trait variations
amongst maize inbred lines were also reported by Bar-Hen et al. (1995) and Gissa (2008).
Prasanna et al. (2001) reported that genetic variability for most traits in maize is high and
amenable to enhancements. However, Karanja et al. (2009) reported similarities in most of
the morphological traits in maize inbred lines and they concluded that this could have been
due to lines being able to balance environmental response differences. Again Karanja et al.
(2009) only used 10 lines which were perhaps from a similar background The variation
detected in the current study can therefore be exploited by breeders through selection,
hybridisation and recombination of desirable genotypes.

Environments were significantly different in 2009/10 and 2010/11 as shown in the combined
analysis and this was due to differences in the growing conditions under which lines were
evaluated. Lines were evaluated under optimum, drought and low N conditions. Significant
environment x line interaction for most traits was undesirable as it indicated different ranks
of lines in performance over the five locations. These results are contrary to the findings by
Gissa (2008) where non-significant environment x line interaction was reported for most
traits. Gissa (2008) evaluated maize inbred lines over two locations, where one site was in
Harare, Zimbabwe and the other site in Bako, Ethiopia. The significant environment x line
interaction detected in the current study makes selection decisions difficult. The DR&SS
lines were lower yielding than CIMMYT lines for all environments, indicating that more
breeding efforts should be devoted at improving the yield performance of these lines,
especially under stress environments. Nevertheless the most critical issue is the performance
of the lines in hybrid combinations.

Knowledge of the relationship amongst different morphological traits in maize is critical in

designing effective selection programmes for crop improvement (Gissa, 2008). Positive and

161
significant correlation of grain yield with ears per plant shows that the higher the number of
ears per plant the higher the grain yield. On the other hand, negative and significant
correlation of grain yield with root lodging is an indication that an increase in root lodging
results in reduced grain yield. Positive and non-significant correlation of grain yield with
plant height might be an indication that taller plants tended to have higher yield compared to
shorter ones and this might have been so in a few cases. Bolanos and Edmeades (1996)
reported that correlations between grain yield and anthesis silking interval were weak under
optimal conditions and strong under stress conditions. However in this study grain yield was
negatively but non-significantly correlated with anthesis-silking interval across all
environments. The results show that under non-stress environments anthesis-silking interval
does not have much influence on grain yield compared to stress environments.

According to Johnson et al. (1955) heritability by itself does not give an insight as to the
amount of genetic progress that would be expected through selection of superior genotypes,
but should always be considered simultaneously with GA. In this study five traits namely
grain yield, plant and ear height, ear position and ears per plant had broad sense heritability
estimates >0.50. Results therefore suggest that further selection for these traits would be
effective for the set of lines used in this study. Singh (2005) reported that selection for a
character is fairly easy if its heritability is high, but selection may be difficult or impractical
for a character with low heritability. Lines exhibited both high GCVs and heritability
estimates for grain yield, plant height and ear height. Genetic coefficients in combination
with heritability estimates are expected to give a better picture of the GA that a breeder can
expect from selection (Assefa et al., 1999). Therefore the traits that showed a high GCV,
heritability and GA as percentage of the mean in the current study, namely grain yield, plant
and ear height and ears per plant would be useful as a base for selection. High genetic gain
indicates that there is a better scope for selection of traits for genetic improvement of the
crop.

PCA further substantiated the existence of broad morphological variation among genotypes.
Results indicated that the overall diversity could not be explained by a few eigenvectors and
these findings are in agreement with findings reported by Gissa (2008). In that study the first

162
four PCs explained 65.7% of the variation contrary to the findings by Yoseph et al. (2005)
who reported that 71.8% of the total variation in 62 traditional Ethiopian highland maize
accessions was explained by the first four PCs. In the current study variations in all traits
were dissected into 14 PCs which accounted for 100% of the variability present among lines.
However, traits such as grain yield, texture, ear aspect, common rust, GLS and anthesis days
were the major contributors.

A total of 14 morphological traits were used in the study and these traits successfully
revealed the existence of considerable variability among the inbred lines. These findings are
in agreement with findings by other researchers. Karanja et al. (2009) used 28 morphological
traits to group 10 maize inbred lines, whilst Lucchini et al. (2003) used 32 morpho-
phenological and agronomic traits to group 20 Italian flint maize landraces. There were cases
in the current study where lines closely related by pedigree tightly clustered together.
Therefore the dendrogram showed the resolution power of morphological traits for grouping
maize inbred lines. However the dendrogram had a poor goodness of fit (r=0.70) and this
could be explained by the small number of morphological traits that were used in this study.
Groupings were also sometimes in agreement with known heterotic groupings. Gerdes and
Tracy (1994) successfully grouped closely related inbred lines using morphological data in
agreement with pedigree data. However, reports consistently indicate that morphological
markers have shortcomings in that they are subject to prevailing environmental conditions
(Bernardo, 1992; Yoseph et al., 2005). Lines related by pedigree also grouped together in
some cases. CML548 and CML538 clustered together with CML312 and the two lines have
CML312 as one of their parents. CML442 and CM537 clustered together and the two lines
share a common parent. Gissa (2008) also reported maize inbred lines closely related by
pedigree clustering together using morphological traits. The grouping was also in some cases
in agreement with existing heterotic grouping, for example RS61P and WCOBY both belong
to SC group and N3.2.3.3 and NAW5885 to the N3 group and these lines clustered together
in the dendrogram. Five CIMMYT lines, CML312, CML442, CML545, CML537 and
CML538 belonging to group A, also clustered in the same group.

163
Clustering of lines using morphological traits in this study also followed a certain pattern.
Lines clustered according to grain yield, anthesis days, anthesis silking interval, plant height,
ear rot, GLS and ears per plant. Other authors have also reported lines clustering according to
morphological traits for example Yoseph et al. (2005) reported lines clustering together
according to grain yield and plant and ear height, whilst Gissa (2008) reported lines
clustering according to plant and ear height, ear diameter, number of rows per ear and leaf
area.

The 23 inbred lines were furthermore characterised using SNP markers. Markers with ≤20%
missing values were used in this study and this constituted 91% of the tested SNP markers.
Lu et al. (2009) eliminated SNPs with more than 20% missing data from the analysis, which
therefore suggests that SNPs with up to 20% missing data can be safely used to produce good
results. In the 23 inbred lines 2 258 alleles were detected at 1 129 loci with two alleles as
expected due to the bi-allelic nature of SNP markers. In this study an average PIC value of
0.258 was reported and Lu et al. (2009) reported similar results. Kota et al. (2008) reported
that generally the SNPs are biallelic and possess the maximum PIC value of 0.50, findings
from this study are in agreement with these findings. SNP markers generally show low PIC,
hence the need to increase marker density (Lu et al., 2011). Gene diversity shows the
likelihood that two randomly selected alleles from the test sample are different. Mean gene
diversity reported in the current study was 0.31 and similar results on SNP markers in maize
were reported by Lu et al. (2009) and Kassa et al. (2012). Lu et al. (2009) reported mean
gene diversity of 0.32, whilst Kassa et al. (2012) reported mean gene diversity of 0.25.
Literature has revealed that SNP markers generally show low gene diversity (Liu et al., 2009;
Kassa et al., 2012). Nonetheless, SNP markers are gaining popularity over SSR markers in
genetic diversity studies especially in maize because of their abundance and they have been
automated leading to significantly reduced costs.

Approximately 9.5% of markers had a minor allele frequency of less than 0.05 in the current
study, whilst Lu et al. (2009) reported 8.8% of SNP markers with less than 0.05 minor allele
frequency and Yan et al. (2009) reported 16.3% of markers with below 0.010 minor allele
frequency. Kassa et al. (2012) reported 37.7% of SNP markers with minor allele frequencies

164
between 0.051 and 0.200 and in the current study 53.3% of the markers had a minor allele
frequency of more than 0.200 with 10.9% having minor allele frequency close to 0.500. On
the other hand Lu et al. (2009) reported 18.7% of the SNP markers with minor allele
frequency close to 0.500. Markers exhibiting higher minor allele frequency values are
considered important for screening maize germplasm from diverse sources. Results from the
current study therefore suggest that SNP markers used were appropriate since more than half
of the markers had high minor allele frequency and the sources of the germplasm used were
diverse. The tested lines generally exhibited more than 90% homozygosity except for the
four lines 2N3d, WCOBY, CML537 and ZM523B. Results suggest that these lines have not
reached the required level of homozygosity therefore there exists a need to continue selfing
these lines until they reach the desired homozygosity before they can be used in the breeding
programme. The other reason for the high levels of heterozygosity could be the purity of the
seed that was used. Low genetic purity of some of the lines might have resulted in reduced
effectiveness of SNPs in discriminating lines according to genetic distances as a result
clustering was generally not in agreement with known heterotic groupings.

The dendrogram obtained from the neighbour-joining clustering algorithm based on SNP
markers showed that all inbreds could be distinguished from each other. Two dendrograms
were presented, one with all 23 lines and another with 19 lines after removal of the four lines
that had high percentages of missing data. The dendrograms showed a good fit (r=0.87 and
r=0.80 respectively). It was noted that removal of the four lines did not improve clustering of
the lines, since a similar pattern was observed. The average GD amongst the inbred lines
based on Rogers’ dissimilarity coefficient was 0.32, indicating low levels of polymorphism
in the inbreds. The GDs in the current study ranged from 0.108 to 0.384 and results are
similar to findings reported by other researchers (Lu et al., 2009; Kassa et al., 2012). Lu et
al. (2009) working on CIMMYT, Chinese and Brazilian maize germplasm reported genetic
distances ranging from 0.140 to 0.349 using SNP markers, whist Kassa et al. (2012) working
on CIMMYT germplasm from eastern and southern Africa reported genetic distances ranging
from 0.003 to 0.450, again using SNP markers. This can be explained by the fact that the
National Breeding Programme has over the years used CIMMYT germplasm in the
development of its new maize inbred lines. On the other hand the National Breeding

165
Programme’s heterotic grouping is based on N3 and SC and the two lines N3.2.3.3 and
SC5522 are part of CIMMYT’s heterotic groups A and B, respectively (Mickelson et al.,
2001). Low genetic distances could also be explained by the fact that CIMMYT has been
mixing germplasm during the time they were emphasising on population breeding at the
expense of hybrid breeding.

The SNP data was also subjected to PCA in order to ascertain if a better grouping of lines be
obtained. Both neighbour-joining cluster analysis and PCA managed to cluster lines related
by pedigree in some cases. In the current study the pattern of grouping from PCA was more
reliable than neighbour-joining clustering and similar results were reported by Kassa et al.
(2012). Although PCA grouped lines better than neighbour-joining cluster analysis, clear
differentiation according to known heterotic groups was not evident. The CIMMYT lines
used in the study could not be divided into known heterotic groups based on SNP data. All
previous SNP and SSR data available for CIMMYT germplasm also did not show clear
separation/grouping based on heterotic groups (Kassa, personal communication, 2011).
Kassa et al. (2012) reported that SNP data did not show clear genetic differentiation of inbred
lines in group A and B as defined by maize breeders at CIMMYT. This is not entirely
unexpected because CIMMYT inbred lines are generally drawn from a pool, population, or
mixture of pools and populations (Warburton et al., 2001). Likewise the DR&SS lines did
not cluster into known heterotic groups. The 1 250 SNP markers used in the study were the
SNPs available at K-Biosciences and they were not necessarily the required number for
studying genetic distances and this could also explain the failure of the markers to clearly
cluster the lines. In other studies more SNP markers have been used for example Lu et al.
(2009) used 1 536 SNPs, Lu et al. (2011) used 1 936 SNPs whilst Kassa et al. (2012) used
1 536 SNPs.

The information generated in this study can be used for selecting single cross hybrids to be
used as testers in the National Breeding Programme. According to Warburton et al. (2005)
clusters with more than five lines may be considered to form a good potential heterotic
group. Lines used in the current study namely RS61P, NAW5885, CML444, ZM523A,
CML442, CML537 and CML539 are potential parents for the drought breeding programme

166
at DR&SS. Therefore the information that has been generated will assist breeders in selecting
parents for producing hybrids with good hybrid vigour. This will be done by crossing lines
from divergent groups. On the other hand good single cross testers will be identified from
lines within the same group and these will then be used as female parents and they will be
crossed with male parents from another group. This will help in speeding up the process of
producing and releasing drought tolerant hybrids for commercial production. Lines in similar
groups can also be used to develop F2 populations that can be used in recombinant inbred line
production.

As with all marker classifications, the classification obtained in this study is subject to
change with addition of new lines and use of different markers. Random sampling of lines
should be done so that there is no biasness towards lines with similar pedigrees or ancestry as
this is bound to affect the effectiveness of clustering. Again genetic contamination of the
lines used should be minimised as seen in the current study that low genetic purity might
have contributed towards reduced effectiveness of SNP markers in clustering lines. SSR
markers have been compared with SNP markers in genetic diversity studies in maize and
Jones et al. (2007) found that SNPs had a clear advantage over SSRs in repeatability of
genotyping results and proportion of missing data. However Hamblin et al. (2007) suggested
that more SNP markers were required in genetic diversity studies compared to SSR markers.
Therefore the clustering observed in the study cannot be considered to be static. However,
the data generated should assist breeders in decision making when it comes to which hybrids
to constitute and evaluate, also considering currently available information and experience.
Information from DNA-based markers can help in improving the efficiency of field trials
conducted for the purpose of identifying promising heterotic patterns thereby resulting in
efficient maize breeding programmes.

5.5 Conclusions
Characterisation of maize inbred lines with the aim of exploring genetic diversity is of great
importance and helps breeders in selecting superior germplasm to be used as parents in a
breeding programme. However, DR&SS lines performed poorer than the CIMMYT lines,
especially under stress environments. Hence the National Breeding Programme’s breeding

167
efforts should be devoted towards increasing grain yield and other traits such as anthesis
silking interval, ears per plant, tolerance to foliar diseases, ear rot and standability traits
(plant height, root and stem lodging). Lines displayed a substantial amount of variability for
the studied morphological traits. The existing genetic variability can therefore be exploited in
developing single cross hybrids to be used as testers, three-way cross hybrids and synthetics.
The selected lines can also be recombined to develop recycled inbred lines. Based on the
morphological traits, some inbred lines related by pedigree grouped together, indicating that
the agronomic traits can be used for primary characterisation of the maize inbred lines. In the
23 inbred lines 2 258 alleles were detected at 1 129 loci each with two alleles per locus as
expected. The SNP based genetic distances were low with an average of 0.32 suggesting
minimal genetic variation among the CIMMYT and DR&SS inbred lines and the low GDs
might have been aggravated by genetic contamination. The UPGMA clustering algorithm
grouped the inbred lines into two major groups, which generally did not agree with their
pedigree records but within subgroups some lines grouped according to pedigree and in some
cases lines from the same heterotic group clustered together. Results of this study revealed
that SNP markers did not consistently cluster lines according to known heterotic groups.
However, there were cases where lines related by pedigree or in the same heterotic group
clustered together. Both neighbour-joining cluster analysis and PCA managed to cluster
together lines related by pedigree or lines within the same heterotic group, however PCA
proved to be more efficient than neighbour-joining. Information generated using both
morphological data and SNP markers can be used to better understand the genetic
relationships amongst CIMMYT and DR&SS lines which should lead to more effective
utilisation of the inbred lines in the breeding programme.

5.6 References
Agronomix Software, Inc. 2005. AGROBASE Generation II User’s Manual. Version II,
Revised Edition. www.agronomix.com. Agronomix Software, Winnipeg, M.B.,
Canada.
Amorim, E.P., C.C. deSouza Almeida, M.J.C. Melo Sereno, F. Bered and J.F. Barbosa Neto.
2003. Genetic variability in sweet corn using molecular markers. Maydica 48: 177-
181.

168
Assefa, K., S. Ketema, H. Tefera, H.T. Nguyen, A. Blum, M. Ayele, G. Bai, B. Simane and
T. Kefyalew. 1999. Diversity among germplasm lines of the Ethiopian cereal tef
(Eragrostis tef (Zucc.) Trotter). Euphytica 106: 87-97.
Bar-Hen, A., A. Charcosset, M. Bourgoin and J. Guiard. 1995. Relationship between genetic
markers and morphological traits in maize inbred line collection. Euphytica 84: 145-
154.
Bauman, L.F. 1981. Review of methods used by breeders to develop superior corn inbreds.
Proceedings of Annual Corn and Sorghum Industry Research Conference 36: 199-
208.
Bernardo, R. 1992. Relationships between single-cross performance and molecular markers
heterozygosity. Theoretical and Applied Genetics 83: 628-634.
Bolanos, J. and G.O. Edmeades. 1996. The importance of the anthesis-silking interval in
breeding for drought tolerance in tropical maize. Field Crops Research 48: 65-80.
Devi, P. and N.K. Singh. 2011. Heterosis, molecular diversity, combining ability and their
inter-relationships in short duration maize (Zea mays L.) across the environments.
Euphytica 178: 71-81.
Farman, M.L., Y. Eto, T. Nakao, Y. Osa, H. Nakayashiki, S. Mayama and S.A. Long. 2002.
Analysis of the structure of the AVRI-C039 avirulence locus in virulent rice-infecting
isolates of Magnaporthe grisea. Molecular Plant Microbe Interactions 15: 6-16.
GenStat. 2011. Introduction to GenStat for Windows. 14th edition. VSN International, Hemel
Hemstead, Hertfordshire HPI. IES, UK.
Gerdes, J.T. and W.F. Tracy. 1994. Diversity of historically important sweet corn inbreds as
estimated by RFLPs, morphology, isozymes and pedigree. Crop Science 34: 26-33.
Gissa, D.W. 2008. Genotypic variability and combining ability of quality protein maize
inbred lines under stress and optimal conditions. PhD thesis. Department of Plant
Sciences, Faculty of Natural and Agricultural Sciences, University of the Free State,
South Africa.
Gupta, P.K., J.K. Roy and M. Prasad. 2001. Single nucleotide polymorphisms: A new
paradigm for molecular marker technology and DNA polymorphism detection with
emphasis on their use in plants. Current Science 80: 524-535.
Hallauer, A.R. 1990. Methods used in developing maize inbreds. Maydica 35: 1-16.

169
Hallauer, A.R. and J.B.F. Miranda. 1988. Quantitative Genetics in Maize Breeding. 2nd
Edition. IOWA State University Press, Ames.
Intergrated Breeding Platform https://www.intergratedbreeding.net/snp-marker-conversion
International Rice Genome Sequencing Project (IRGSP). 2005. The map- based sequence of
the rice genome. Nature 436: 793-800.
Johnson, H.W., H.F. Robinson and R.E. Comstock. 1955. Estimates of genetic and
environmental variability in soybeans. Agronomy Journal 47: 314-318.
Karanja, J., N.O. Amugune, J. Ininda, J.N. Kimatu and J.W. Dansin. 2009. Microsatellite
analysis of the correlation between molecular and morphological traits in assorted
maize inbred lines. African Crop Science. 17: 133-144.
Kassa, S., C. Magorokosho, B.S. Vivek, D. Makumbi, Y. Beyene, S. Mugo, B.M. Prasanna
and M.L. Warbuton. 2012. Molecular characterization of diverse CIMMYT maize
inbred lines from eastern and southern Africa using single nucleotide polymorphic
markers. BioMed Central Genomics 3: doi: 10.1186/1471-2164-13-113.
Kaufman, L. and P.J. Rousseeuw. 1990. Finding Groups in Data: An Introduction to Cluster
Analysis. Wiley-Interscience Publication, New York. pp 3-44.
Kota, R., R.K. Varshney, M. Prasad, H. Zhang, N. Stein and A. Graner. 2008. EST-derived
single nucleotide polymorphism (SNP) markers for assembling genetic and physical
maps of the barley genome. Functional Intergrative Genome 8: 223-233.
Koutsika-Sotiriou, M.S. and A. Karagounis. 2005. Assessment of maize hybrids. Maydica
50: 63-70.
Lanza, L.L.B., C.L. Souza Jr, L.M.M. Ottobani, M.L.C. Vierira and A.P. Souza. 1997.
Genetic distance of inbred lines and prediction of maize single-cross performance
using RAPD markers. Theoretical and Applied Genetics 94: 1023-1030.
Li, Y., C. Liu, Y. Shi, Y. Song, T. Wang and Y. Li. 2009. Genetic location and evaluation of
chromosomal segments for drought tolerance at flowering stage in maize using
seleced backcross populations. Plant Breeding 128: 342-347.
Lijavetzky, D., J.A. Cabezas, A. Ibanez, V. Rodriguez and J.M. Martinez-Zapater. 2007.
High throughput SNP discovery and genotyping in grape vine (Vitis vinifera L.) by
combining a re-sequencing approach and SNPlex technology. BioMed Central
Genomics 8: 424.

170
Liu, K. and S.V. Muse. 2005. Powermaker: an intergrated analysis environment for genetic
marker analysis. Bioinformatics 21: 2128-2129.
Lu, Y., J. Yan, C.T. Guimarães, S. Taba, Z. Hao, S. Gao, S. Chen, J. Li, S. Zhang, B.S.
Vivek, C. Magorokosho, S. Mugo, D. Makumbi, S.N. Parentoni, T. Shah, T. Rong,
J.H. Crouch and Y. Xu. 2009. Molecular characterisation of global maize breeding
germplasm based on genome-wide single nucleotide polymorphisms. Theoretical
and Applied Genetics 120: 93-115.
Lu, Y., T. Shah, Z. Hao, S. Taba, S. Zhang, S. Gao, J. Liu, M. Cao, J. Wang, A.B. Prakash,
T. Rong and Y. Xu. 2011. Comparative SNP and haplotype analysis reveals a higher
genetic diversity and rapider linkage disequilibrium decay in tropical than temperate
germplasm in maize. PLoS ONE 6: e24861. doi: 10.1371/journal,pone. 0024861.
Lucchini, M., G. Barcaccia and P. Parrini. 2003. Characterisation of a flint maize (Zea mays
L. convar. mays) Italian landrace: I. Morpho-phenological and agronomic traits.
Genetic Resources and Crop Evolution 50: 315-327.
Mace, E.M., H.K. Buhariwalla and J.H. Crouch. 2003. A high-throughput DNA extraction
protocol for tropical molecular breeding programs. Plant Molecular Biology
Reporter 21: 459a-459h.
Mickelson, H.R., H. Cordova, K.V. Pixley and M.S. Bjarnason. 2001. Heterotic relationships
among nine temperate and subtropical maize populations. Crop Science 41: 1012-
1020.
Molinier, J., R. Gerhard, Z. Cyril and B. Hohn. 2006. Transgeneration memory of stress in
plants. Nature 442: 1046-1049.
Perrier, X. and J.P. Jacquemould-Collet. 2006. DARwin software http://darwin.cirad.fr/
Prasanna, B.M., S.K. Vasal, B. Kassahun and N.N. Singh. 2001. Quality Protein Maize.
Current Science 81: 1308-1319.
Primer Picker Software http://ww.kbioscience.co.uk/primer-picker/
Rafalski, A. 2002. Applications of single nucleotide polymorphisms in crop genetics.
Current Opinion in Plant Biology 5: 94-100.
Ravel, C., S. Praud, A. Murigneux, A. Canaguier, F. Sapet, D. Samson, F. Balfouer, P.
Dufour, B. Chalhoub, D. Brunel, M. Beckert and G. Charmet. 2006. Single

171
 nucleotide polymorphism frequency in a set of selected lines of bread wheat
(Triticum aestivum L.). Genome 49: 1131-1139.
Rogers. J.S. 1972. Measures of genetic similarity and genetic distance. In: Studies in
Genetics VII. University of Texas Publication 7213: 145-153.
Rohlf F.J. 1993. NTSYS-pc. Numerical taxonomical and multivariate analysis system.
Exerter software, Setauket, New York, USA.
Saitou, N. and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing
phylogenetic trees. Molecular Biology and Evolution 4: 406-425.
Singh, R.K. 2005. Plant Breeding: Principles and Methods. 7th edition Kalyani Publishers,
New Delhi, India. pp 698-731.
Smith. J.S.C. and O.S. Smith. 1992. Fingerprinting crop varieties. Advances in Agronomy 47:
85-140.
Smith, O.S. 1986. Covariance between line per se and testcross performance. Crop Science
26: 540-543.
SPSS 15.0 for Windows. 2006. www.spss.com SPSS Inc., Chicago III.
Tenaillon, M.I., M.C. Sawkins, A.D. Long, R.L. Gaut, J.F. Doebley and B.S. Gaut. 2001.
Patterns of DNA sequence polymorphism along chromosome 1 of maize (Zea mays
L.). Proceedings of the National Academy of Sciences of the United States of
America 98: 9161-9166.
Troyer, A.F. 2004. Background of U.S. hybrid corn II: Breeding climate and food. Crop
Science 44: 370-380.
Warburton, M.L., J.M. Ribaut, J. Franco, J. Crossa, P. Dubreuil and F.J. Betran. 2005.
Genetic characterisation of 218 elite maize inbred lines using RFLP markers.
Euphytica 142: 97-106.
Warburton, M., X. Xianchun, S. Ambriz, L. Diaz, E. Villordo and D. Hoisington. 2001. Use
of molecular markers in maize diversity studies at CIMMYT. Seventh Eastern and
Southern Africa Regional Maize Conference. 11-15 February 2001. pp 130-133.
Williams, J.G.K., A.R. Kubelik, K.J. Livak, J.A. Rafalski and S.V. Tingey. 1990. DNA
polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic
Acids Research 18: 6531-6535.

172
Yan, J., T. Shah, M.L. Warburton, E.S. Buckler, M.D. McMullen and J. Crouch. 2009.
Genetic characterization and linkage disequilibrium estimation of a global maize
collection using SNP markers. PLoS ONE 4: e8451.
doi:10.1371/journal.pone.0008451.
Yoseph, B., A.M. Botha and A.A. Myburg. 2005. A comparative study of molecular and
morphological methods of describing genetic relationships in traditional Ethiopian
highland maize. African Journal of Biotechnology 4: 586-595.
Zhu, Y.L., Q.J. Song, D.L. Hyten, C.P. Van Tassell, L.K. Matukumalli, D.R. Grimm, S.M.
Hyatt, E.W. Fickus, N.D. Young and P.B. Cregan. 2003. Single nucleotide
polymorphisms in soyabean. Genetics 163: 1123-1134.

173
 CHAPTER 6

Relationships between heterosis, genetic distances and combining ability

data in maize hybrids
Abstract
Hybrid vigour (heterosis) is critical for grain yield and is one of the crucial factors
determining growth of the hybrid seed sector in maize. Success of a maize hybrid breeding
programme that handles many crosses and field evaluations can be improved greatly with the
use of superior hybrid value predictors. The current study was done to estimate heterosis
effects among CIMMYT and DR&SS elite maize inbred lines and to determine correlations
between GD, heterosis and SCA under optimum, drought and low N conditions. Seventy-two
hybrids generated using a NCDII from 10 DR&SS and 13 CIMMYT elite inbred lines were
evaluated across seven environments in the 2009/10 and 2010/11 seasons under optimum,
low N and drought conditions. The 23 lines used in the crosses were genotyped using SNP
markers. An average of 112.29% MPH and 76.40% HPH were realised across environments
and this was an indication of the potential of the inbred lines for hybrid development. The
negative heterosis (-5.1% MPH and -7.4% HPH) detected for days to anthesis indicated
earlier anthesis of hybrids compared to their parental inbred lines. Correlations of GDs
determined using either morphological or SNP data with heterosis under drought conditions
were too low to be of predictive value. Significant positive correlations and regressions were
recorded for SCA with MPH, HPH and per se performance of hybrids. The HPH and MPH
also showed significant positive associations and linear regressions along with high
coefficients of determination with per se performance of hybrids especially under drought
conditions. Results indicated that CIMMYT lines CML444, CZL03007, CML536, CML539
and CML548 can contribute to the DR&SS maize stress breeding programme. Therefore the
genetic and morphological distances were of little importance in predicting performance,
heterosis and SCA effects of hybrids.

174
6.1 Introduction
Knowledge of heterotic groups and heterotic patterns is helpful in plant breeding as it helps
breeders to utilise their germplasm in a more efficient and consistent manner through
utilisation of complementary lines for maximising the outcomes of a hybrid breeding
programme. The choice of parents is extremely important in the production of hybrids and
hybridisation itself does not result in hybrid vigour, hence heterosis depends on the genetic
background of parents. Heterosis is defined as the improved performance of hybrids and
shows the advantage of hybrid performance compared with the parental lines from which
they were constituted (Hallauer et al., 2010). Precise information on the relationship between
maize inbred lines is critical for their identification and manipulation of heterotic patterns in
germplasm pools (Livini et al., 1992). The idea of heterotic grouping was initially established
by maize researchers who observed that inbred lines from different populations had a
tendency to produce greater performing hybrids when crossed with inbred lines from
different groups (Hallauer et al., 2010).

Hallauer et al. (2010) defined a heterotic group as a group of interrelated or unrelated inbred
lines from a similar or diverse population, which shows a comparable combining ability
when hybridised with inbred lines from another population. The Zimbabwe national maize
breeding programme uses two heterotic groups namely SC and N3, whilst CIMMYT uses
heterotic groups A and B. The SC grouping is equivalent to the B grouping, whilst the N3
grouping is equivalent to the A grouping. Genotypes within a group will usually show no or
little heterosis when crossed to each other because they are generally genetically closely
related but there are exceptions to this rule. High heterosis from germplasm derived from
within a heterotic group has been observed in some experiments (Vasal et al., 1999).
Heterotic groups thus generally represent broad sources of germplasm, which exhibit
optimum heterosis when crosses are made between groups. Heterosis estimation is thus
important in maize breeding in order to identify the best combinations of progenitors to form
potential hybrids and to study gene action.

Breeding approaches in maize have been established to benefit from expression of hybrid
vigour in inbred line crosses. Hybrid production exploits the phenomenon of heterosis. The

175
search for high yielding varieties in Zimbabwe maize breeding programmes saw a major shift
from OPV to hybrid development in the early 1930s. Zimbabwe maize breeding has been a
success story over the years with hybrids being developed for both small scale and
commercial production. According to Mashingaidze (2006) commercial adoption of hybrids
in terms of area planted increased from 22% (1949/50) to 88% (1960/61) and 93% (1966/67).
Adoption was, however, slower in communal lands before independence. The Mangwende
communal area for instance had a hybrid adoption of 42% in 1975 but by 1985 it stood at
99% (Mashingaidze, 2006). Currently 90% of the total production area (in both the
commercial and communal sector) is planted to hybrids with the remainder being planted to
other products such as OPVs, synthetics and recycled seed. The National Breeding
Programme uses introductions from CIMMYT and other breeding programmes to take
advantage of genetic diversity. The level of genetic diversity between genotypes is usually
unknown and the only alternative is to investigate it through development of crosses
(Hallauer et al., 2010). If a cross between two parents exhibits high heterosis it is usually
determined that the two genotypes are genetically more diverse than two genotypes that show
slight or no heterosis in their crosses.

Since heterosis in F1 progeny is determined by genetic diversity of the inbred parents, the
usual method of selecting parents only on the basis of their performance and adaptation under
different environments does not lead to significant selection results (Allard, 1960). Therefore
in selecting ideal parental lines for a hybridisation programme, information on GD,
combining ability and heterotic patterns is vital (Beck et al., 1990). Hybrid performance has
been predicted using GD and the effectiveness of the prediction was better with crosses
amongst inbred lines from the same heterotic group than in crosses amongst inbred lines
from diverse heterotic groups (Melchinger, 1999). Several investigators have used GD to
predict hybrid performance (Cheres et al., 2000; Betran et al., 2003; Dhliwayo et al., 2009;
Devi and Singh, 2011; George et al., 2011). Some investigators have reported significant
association between marker based GD and heterosis (Lee et al., 1989; Smith et al., 1990;
Melchinger et al., 1992; Betran et al., 2003; Amorim et al., 2006; Srdic et al., 2007; George
et al., 2011), whilst others have reported non-significant or no association (Balestre et al.,
2008; Legesse et al., 2008; Dhliwayo et al., 2009; Devi and Singh, 2011). Therefore the

176
application of molecular markers in maize is not conclusive. The study was done to i) assess
the validity of known heterotic groups, ii) estimate heterosis effects among CIMMYT and
DR&SS maize inbred lines and iii) estimate correlations between GD, heterosis and SCA
under optimum, drought and low N environments.

6.2 Materials and methods

6.2.1 Germplasm
Seventy-two single cross hybrids and 23 parental inbred lines were evaluated for agronomic
performance in a total of 14 sites in 2009/10 and 2010/11 seasons. Details of parental lines
are given in section 3.2.1.

6.2.2 Sites
The site details are given in sections 3.2.2 and 4.2.2. The details of the measured and derived
traits are as described in Table 3.3.

6.2.3 DNA extraction and SNP genotyping

Details of DNA extraction and SNP genotyping are given in sections 5.2.4 and 5.2.5.

6.2.4 Statistical analysis

ANOVA was performed for individual and across sites using the PROC MIXED procedure
from SAS (SAS Institute, 2002). Varieties were taken as fixed effects. The means were used
for SCA estimation. Line x tester analysis was used to estimate SCA effects for hybrids in all
environments and across environments.
Mid-parent heterosis (MPH) was calculated as:
MPH = [(F1 – MP)/MP] x 100
Where: F1 = Mean of the F1 hybrid performance
MP = (P1 + P2)/2 in which P1 and P2 are the means of the inbred parents
respectively

High-parent heterosis (HPH) was calculated as:

177
HPH = [(F1-HP)/HP] x 100
Where: HP = Mean of the better parent
GCA and SCA effects were estimated according to Singh and Chaudhary (1977) using the
following formulae:
GCA effects for lines:
gi = xi…/tr – x…./1tr
GCA effects for testers:
gt = x.j./lr – x…./ltr
SCA effects:
sij = xij./r – xi.../tr – x.j./lr – x…/lrt
Where: l = number of lines
t = number of testers
r = number of replications
S.E. (gi – gj) line = (2Me/r x t)½
S.E. (gi - gj) tester = (2Mr/l x r)½
S.E. (sij- skl) = (2Me/r)½

GDs for SNP data were calculated from allele frequencies using Powermaker version 3.25
(Liu and Muse, 2005) as described in section 5.2.6. Means per environment and across
environments were used to calculate Pearson correlation coefficients (r) between GD, F1
grain yield, mid-parent (MP), high-parent (HP), MPH, HPH and SCA using SPSS version
15.0 for Windows (2006). Morphological traits were used for computing Euclidean distances
in SPSS version 15.0 for Windows (2006) as described in section 5.2.6.

6.3 Results
The means presented in this chapter are for the 10 best and 10 poorest hybrids in terms of
SCA effects and data for all crosses is presented in the appendices section (Appendices 11
and 12) for better presentation. Also to note is that the mean values, minimum and maximum
values presented in the different tables are for the entire data set and not only for the crosses
presented.

178
6.3.1 Grain yield, specific combining ability and mid- and high-parent heterosis across
all environments
The average MPH and HPH across all environments were 112.29% and 76.40% respectively.
Hybrid L5 (2N3d) x T1 (CML395) had the best SCA effects of 0.86 but it was not
necessarily the best in terms of mean grain yield, MPH and HPH (Table 6.1). L3 (RS61P) x
T9 (CML444) had the highest mean yield of 4.78 t ha-1 but did not have the highest heterosis
values. The GD between these two lines was 0.27. L4 (NAW5885) produced crosses with
high mean yields as well as high MPH. SC5522 (L7) produced two crosses with high positive
SCA effects of 0.77 and 0.75 respectively but these crosses had mean yields of below 3.0 t
ha-1 and MPH of 140.92% and 140.17% respectively. This line also produced a cross with
CZL03007 which had the highest MPH (237.58%) (data not shown) and the GDs between
the respective parental lines were above average. L6 (2Kba) x T11 (CML548), L1 (K64r) x
T3 (CML539) and L4 (NAW5885) x T6 (CZL03007) had MPH and HPH values above
100%. L6 (2Kba) x L9 (CML444) had the lowest MPH of 22.20% and the lowest negative
HPH of -11.19%. MPH and HPH for the 10 hybrids with best SCA effects across all
environments are presented graphically in Figure 6.1. There was no significant difference
between MPH and HPH values for L3 (RS61P) x T9 (CML444). L7 (SC5522) x T1
(CML395) and L7 (SC5522) x T10 (CML536) had high MPH values compared to the HPH
values, whilst L6 (2Kba) x T11 (CML548), L1 (K64r) x T3 (CML539) and L4 (NAW5885)
x T6 (CZL03007) had high MPH and HPH values. L5 (2N3d) x T1 (CML395) despite having
the highest SCA value (Table 6.1) had the second lowest MPH amongst the 10 hybrids
(Figure 6.1).

179
Table 6.1 Hybrid mean grain yield, specific combining ability, mid- and high-parent
heterosis and genetic distance across all environments
Hybrid GYD( t ha-1) SCA MPH (%) HPH (%) GD

L5/T1 2.43 0.86 88.41 74.60 0.34

L3/T9 4.78 0.79 93.18 91.31 0.27
L7/T1 2.29 0.77 140.92 27.03 0.34
L7/T10 2.34 0.75 140.17 53.92 0.36
L6/T11 4.13 0.68 166.80 136.08 0.35
L4/T4 4.22 0.51 159.25 32.09 0.32
L1/T3 3.93 0.50 175.72 132.76 0.33
L1/T7 4.06 0.50 88.07 56.05 0.36
L4/T6 4.06 0.47 195.02 147.91 0.29
L1/T5 3.86 0.36 124.83 78.79 0.31
L2/T11 3.45 -0.44 80.19 37.15 0.36
L5/T5 2.28 -0.45 88.41 74.60 0.20
L5/T7 1.99 -0.50 73.36 53.20 0.33
L4/T11 3.35 -0.50 77.91 28.69 0.32
L4/T3 3.16 -0.52 111.97 75.22 0.29
L8/T1 2.11 -0.54 87.20 62.14 0.33
L3/T4 3.64 -0.65 69.65 60.94 0.30
L8/T3 2.97 -0.76 127.62 74.99 0.32
L1/T6 2.52 -0.81 89.94 60.51 0.33
L6/T9 2.10 -0.92 22.20 -11.19 0.33
Mean 4.00 0.008 112.29 76.40 0.31
Min 2.29 -0.92 22.20 -11.19 0.20
Max 5.31 0.86 237.58 181.71 0.37
GYD=Grain yield; SCA=specific combining ability; MPH=mid parent heterosis; HPH=high parent heterosis; GD = genetic
distance; L1= K64r; L2=N3.2.3.3; L3=RS61P; L4=NAW5885; L5=2N3d; L6=2Kba; L7=SC5522; L8=RA214P; T1=CML395;
T3=CML539; T4= CML442; T5=CML537; T6=CZL03007; T7=CML545; T9=CML444; T10=CML536; T11=CML548 ;
Min=minimum; Max=maximum.

The F1 means were higher than the parental means for plant height, ear height and ears per
plant across all environments (Table 6.2). Anthesis days had negative MPH and HPH values
as shown by the minimum and the maximum values. The F1 hybrids showed considerable
heterosis for both plant and ear height. The minimum and maximum hybrid values for plant
height and ear height were higher than the parental values. Ears per plant had mean heterosis
values of less than 10% for both HPH and MPH respectively. The minimum MPH and HPH
for ears per plant were both negative but the maximum values were 86.72 and 71.45
respectively.

180
 250.00

% Heterosis 200.00

150.00

100.00 MPH
HPH
50.00

0.00

Hybrids

Figure 6.1 The high- and mid-parent heterosis for 10 selected hybrids across all
environments.
L1=K64r; L3=RS61P; L4=NAW5885; L5=2N3d; L6=2Kba; L7=SC5522; T1=CML395; T3=CML539; T4=CML442;
T5=CML537; T6=CZL03007; T7=CML548; T9=CML444; T10=CML536; MPH=mid-parent heterosis; HPH=high-
parent heterosis.

Table 6.2 F1, parental, mid- and high-parent heterosis means for anthesis days and
other agronomic traits across all environments

AD PH EH EPP
F1 mean 67.5 259.9 125.4 0.53
Min 63.8 209.7 96.2 0.46
Max 72.7 291.1 156.3 0.94
Parental mean 71.2 163.3 82.1 0.48
Min 68.2 129.9 65.3 0.43
Max 75.9 195.6 97.3 0.55
MPH mean -5.1 59.9 53.3 9.80
Min -9.7 28.9 15.6 -4.12
Max -1.2 80.0 58.5 86.72
HPH mean -7.4 49.9 42.7 4.80
Min -14.5 10.3 4.1 -12.84
Max -2.7 76.1 71.4 71.45
MPH=mid-parent heterosis (%); HPH=high-parent heterosis (%); AD=anthesis days; PH=plant height (cm); EH=ear height (cm);
EPP=ears per plant (#); Min=minimum; Max=maximum.

Under optimum conditions the parental mean was higher than the F1 mean for anthesis days
indicating that hybrids flowered earlier than the parental lines (Table 6.3). Both MPH (-5.1)
and HPH (-3.5) for anthesis days were negative. The F1 mean for plant height was higher

181
than the parental mean and positive MPH and HPH were observed. Ear height means for both
F1 and parental lines were similar, but positive heterosis was observed for the trait. Negative
MPH and HPH means were observed for ears per plant and the minimum and maximum F1
values were lower than the parental minimum and maximum values. However the maximum
values for both MPH and HPH were positive.

The F1 means for anthesis days were lower than the parental means under both low N and
drought conditions (Tables 6.4 and 6.5). Negative MPH and HPH means were observed
under both conditions, however maximum values for MPH and HPH were positive under low
N and under drought conditions only HPH was positive. Positive MPH and HPH for plant
and ear height were realised under both low N and drought conditions. The F1 mean for ears
per plant was higher than the parental mean under low N conditions and positive MPH and
HPH means were realised. The minimum values for both MPH and HPH for ears per plant
under low N conditions were negative, however high maximum values of 349.3 and 311.2
respectively were observed. Under drought conditions F1 mean for ears per plant was lower
than the parental mean, but the maximum F1 value was higher than the parental maximum
value. A positive MPH mean (1.80) was realised and the HPH mean (-7.90) was negative.
The minimum values for both MPH and HPH were negative but high maximum values were
realised.

6.3.2 Mean grain yield, specific combining ability, mid- and high-parent heterosis
under optimum conditions
The minimum grain yield was 2.91 t ha-1 and the maximum was 6.79 t ha-1. L5 (2N3d) x T1
(CML395) and L3 (RS61P) x T9 (CML444) were the best specific combinations under
optimum conditions (0.99) in terms of the SCA effects (Table 6.6). However, they were not
the best in terms of grain yield as well as MPH and HPH. MPH ranged from 16.91% to
253.53% whilst HPH ranged from -9.21% to 162.7% (Appendix 11).

182
Table 6.3 F1, parental, mid- and high-parent heterosis means for anthesis days and
other agronomic traits under optimum conditions

AD PH EH EPP
F1 mean 67.7 161.5 80.4 0.94
Min 63.8 124.9 61.9 0.66
Max 73.1 200.9 98.9 1.18
Parental mean 70.3 160.9 80.9 1.00
Min 66.7 129.9 65.3 0.84
Max 75.9 195.6 80.9 1.48
MPH mean -5.1 59.1 57.8 -6.00
Min -11.0 18.9 9.6 -43.10
Max 1.2 82.4 99.1 20.41
HPH mean -3.5 57.1 56.9 -13.70
Min -8.1 23.8 9.5 -56.60
Max 1.8 109.3 102.4 10.70
MPH=mid-parent heterosis (%); HPH=high-parent heterosis (%); AD=anthesis days; PH=plant height (cm); EH=ear height (cm);
EPP=ears per plant (#); Min=minimum; Max=maximum.

Table 6.4 F1, parental, mid- and high-parent heterosis means for anthesis days and
other agronomic traits under low nitrogen conditions
AD PH EH EPP
F1 mean 72.9 217.0 77.9 0.67
Min 69.2 177.7 58.2 0.36
Max 78.5 268.8 98.3 0.94
Parental mean 77.1 106.8 40.4 0.41
Min 70.8 93.0 32.2 0.15
Max 85.6 135.1 53.6 0.80
MPH mean -6.6 99.3 90.0 79.60
Min -16.6 39.5 19.7 -37.70
Max 5.6 141.1 183.8 349.30
HPH mean -6.0 93.9 79.3 49.30
Min -18.7 26.5 -1.5 -45.80
Max 9.7 139.8 188.8 311.20
MPH=mid-parent heterosis (%); HPH=high-parent heterosis (%); AD=anthesis days; PH=plant height (cm); EH=ear height (cm);
EPP=ears per plant (#); Min=minimum; Max=maximum.

183
Table 6.5 F1, parental, mid- and high-parent heterosis means for anthesis days and
other agronomic traits under managed drought conditions
AD PH EH EPP
F1 mean 96.4 223.7 109.3 0.77
Min 88.4 166.9 66.4 0.10
Max 104.8 260.3 148.3 1.49
Parental mean 105.5 174.9 85.6 0.81
Min 97.9 138.9 46.9 0.41
Max 117.7 227.5 129.6 1.26
MPH mean -9.9 26.9 27.5 1.80
Min -18.6 -10.6 -21.8 -87.20
Max -2.1 58.9 105.9 137.40
HPH mean -9.7 35.2 44.9 -7.90
Min -21.7 -8.7 -19.5 -88.40
Max 9.4 81.8 186.7 163.90
MPH=mid-parent heterosis (%); HPH=high-parent heterosis (%); AD=anthesis days; PH=plant height (cm); EH=ear height (cm);
EPP=ears per plant (#); Min=minimum; Max=maximum.

L7 (SC5522) x T6 (CZL03007) had the highest MPH of 253.53% and L1 (K64r) x T2

(CML312) had the highest HPH of 162.77%. L3 (RS61P x T9 (CML444), L1 (K64r) x T3
(CML539), L8 (RA214P) x T8 (CML538), L4 (NAW5885) x T6 (CZL03007), L2 (N3.2.3.3)
x T9 (CML444), L8 (RA214P) x T6 (CZL03007) and L2 (N3.2.3.3) x T6 CZL03007 had
high values above 100% for both MPH and HPH.

6.3.3 Mean grain yield, specific combining ability, mid- and high-parent heterosis
under low nitrogen conditions
The mean grain yield, SCA, MPH, HPH and GD for the hybrids under low N conditions are
presented in Table 6.7. L5 (2N3d) x T1 (CML395) was again the best specific combination
under low N conditions with a HPH value of 126.62%. A number of hybrids had negative
MPH and HPH values, however there are some that had high heterosis values for example L1
(K64r) x T2 (CML312) had the highest MPH of 273.74% and L6 (2Kba) x T5 (CML537)
had the highest HPH value of 155.50% (Appendix 12).

6.3.4 Mean grain yield, mid- and high-parent heterosis and specific combining ability
under drought conditions
The MPH and HPH minimum values were negative but maximum values of 180.61% and
162.07% respectively were recorded for the hybrid L5 (2N3d) x T11 CML548 (Table 6.8).

184
The hybrid also had the highest grain yield of 3.26 t ha-1. L5 (2N3d) x T9 (CML444) was the
best specific combination but had low MPH and a negative HPH. L8 (RA214P) x T3
(CML539) was the poorest yielder with 0.59 t ha-1 accompanied by lowest MPH and HPH
values of -65.23% and -74.83% respectively. L4 (NAW5885) x T6 (CZL03007) was the
poorest specific combiner with an SCA value of -1.07.

Table 6.6 F1 mean grain yield (t ha-1), specific combining ability, mid- and high-parent
heterosis and genetic distance under optimum conditions

Hybrid F1 GYD SCA MPH HPH GD

L5/T1 3.66 0.99 76.68 61.49 0.34
L3/T9 6.42 0.99 124.14 155.18 0.27
L1/T3 5.56 0.94 173.19 131.78 0.33
L7/T10 2.95 0.84 136.19 49.48 0.36
L7/T1 3.10 0.83 134.68 40.44 0.34
L8/T8 6.16 0.76 156.75 127.89 0.26
L6/T12 3.72 0.74 133.05 89.06 0.33
L6/T6 5.28 0.72 123.93 81.66 0.34
L4/T6 5.85 0.70 175.41 130.06 0.29
L2/T9 6.05 0.67 187.98 151.43 0.33
L8/T6 6.09 -0.54 153.83 144.65 0.29
L5/T11 3.79 -0.55 97.91 66.71 0.35
L2/T6 4.84 -0.55 143.84 116.91 0.32
L4/T8 4.81 -0.57 105.21 60.11 0.31
L4/T9 4.47 -0.68 125.11 84.85 0.29
L5/T7 3.24 -0.71 85.74 73.83 0.33
L6/T9 3.23 -0.86 49.51 19.32 0.33
L1/T6 3.45 -0.97 72.36 47.44 0.33
L8/T3 4.32 -1.05 94.20 85.41 0.32
L3/T4 4.29 -1.10 48.63 59.10 0.30

Mean 5.19 0.03 117.39 85.29 0.31

Min 2.91 -1.10 16.91 -9.21 0.20
Max 6.79 0.99 253.53 162.77 0.37
GYD=grain yield (t ha-1); SCA=specific combining ability; MPH=mid-parent heterosis (%); HPH=high-parent heterosis;
GD=genetic distance; Min=minimum; Max=maximum; L1=K64r; L2=N3.2.3.3; L3=RS61P; L4=NAW5885; L5=2N3d; L6=2Kba;
L7=SC5522; L8= RA214P; T1=CML395; T3=CML539; T4=CML442; T6=CZL03007; T7=CML545; T8=CML538; T9=CML444;
T10=CML536; T11=CML548.

6.3.5 Heterotic grouping in relation to field heterosis

Mean MPH and HPH for across all environments, known heterotic groupings and grouping
according to SNP markers are presented in Table 6.9. The DR&SS N3 grouping is equivalent
to CIMMYT A grouping, whilst SC is equivalent to CIMMYT B grouping. The grouping
according to SNP markers was in some cases consistent with the known heterotic grouping,

185
e.g. L5 (2N3d) and T5 (CML537) are in a similar known heterotic grouping (N3 and A) and
the SNP markers grouped them into the same group (Figure 5.7). The hybrid had moderate
MPH (88.41) and HPH (74.60) and its mean yield across environments was 4.05 t ha-1. L6
(2Kba) and T11 (CML548) are in different known heterotic groups (SC and A respectively)
and the grouping was consistent even based on SNP markers. The cross between these two
lines exhibited high MPH and HPH of 166.8% and 136.08% respectively (Table 6.9).

Table 6.7 F1 mean grain yield (t ha-1), specific combining ability, mid- and high-parent
heterosis and genetic distance under low nitrogen conditions

Hybrid F1GYD SCA MPH HPH GD

L5/T1 4.53 0.74 66.63 126.62 0.36
L1/T4 1.60 0.73 142.85 55.71 0.27
L7/T1 2.97 0.67 95.24 48.38 0.34
L3/T3 5.35 0.66 90.32 92.37 0.30
L3/T9 4.54 0.49 52.87 46.45 0.27
L7/T10 2.69 0.47 132.01 110.26 0.36
L4/T6 3.10 0.45 170.81 126.34 0.29
L2/T5 3.76 0.43 46.20 106.45 0.35
L8/T8 3.16 0.43 -18.45 2.07 0.26
L6/T5 3.83 0.41 148.86 155.50 0.35
L1/T6 2.02 -0.39 15.00 -12.28 0.33
L5/T3 2.57 -0.41 -17.48 -7.68 0.34
L3/T5 3.68 -0.41 69.78 145.61 0.34
L3/T6 2.39 -0.45 13.51 68.26 0.27
L2/T1 2.68 -0.50 -4.93 34.04 0.35
L6/T9 0.94 -0.53 -59.76 -69.63 0.33
L3/T4 3.27 -0.55 21.25 28.15 0.30
L4/T8 1.89 -0.56 -6.15 -39.15 0.31
L4/T5 1.95 -0.57 61.15 30.07 0.34
L8/T3 1.76 -0.64 -52.81 -36.86 0.32
Mean 2.90 0.01 33.09 25.73 0.31
Min 0.94 -0.64 -66.25 -71.92 0.20
Max 5.35 0.74 273.74 155.50 0.37
GYD=grain yield (t ha-1); SCA=specific combining ability; MPH=mid-parent heterosis (%); HPH=high-parent heterosis (%);
GD=genetic distance; Min=minimum; Max=maximum; L1=K64r; L2=N3.2.3.3; L3=RS61P; L4=NAW5885; L5=2N3d; L6=2Kba;
L7= SC5522; L8=RA214P; T1=CML395; T3=CML539; T4=CML442; T5=CML537; T6=CZL03007 T8=CML538; T9=CML444;
T10=CML536.

L1 (K64r) and T3 (CML539) clustered into separate groups using SNP markers and their
cross exhibited high heterosis even though they were in similar (N3 and A respectively)
heterotic groups. L4 (NAW5885) x T6 (CZL03007) and L6 (2Kba) x T11 (CML548)
grouped both into separate heterotic groups and SNP clusters had crosses with high heterosis.

186
Table 6.8 Hybrid F1 grain yield (t ha-1), mid- and high-parent heterosis, specific
combining ability and genetic distance under drought conditions

Hybrid GYD MPH HPH SCA GD

L5/T9 1.99 7.33 -19.19 1.16 0.29

L4/T5 2.28 66.61 22.08 0.79 0.34

L5/T11 3.26 180.61 162.07 0.79 0.35

L3/T3 2.51 13.43 7.54 0.78 0.30

L4/T4 3.01 64.31 60.83 0.66 0.32

L6/T7 2.15 7.02 6.58 0.64 0.36

L2/T7 2.41 28.52 20.33 0.62 0.33

L8/T5 2.03 111.66 94.05 0.58 0.34

L4/T8 2.04 6.15 3.43 0.53 0.31

L1/T7 2.79 35.48 31.75 0.50 0.36

L1/T12 2.28 18.71 7.48 -0.56 0.33

L4/T7 1.91 -1.10 -4.33 -0.59 0.32

L2/T8 1.71 -7.88 -13.10 -0.60 0.34

L6/T9 1.40 -37.63 -43.27 -0.69 0.33

L3/T4 2.65 36.75 27.00 -0.74 0.30

L8/T1 1.36 1.14 -17.45 -0.76 0.33

L8/T3 0.59 -65.23 -74.83 -0.77 0.32

L5/T4 1.95 28.30 8.75 -0.80 0.34

L2/T5 1.98 51.50 13.25 -0.87 0.35

L4/T6 1.92 6.62 2.65 -1.07 0.29

Mean 2.09 26.08 10.34 0.00 0.31

Min 0.59 -65.23 -74.83 -1.07 0.20

Max 3.26 180.61 162.07 1.16 0.37

GYD=grain yield (t ha-1); MPH=mid-parent heterosis (%); HPH=high-parent heterosis (%); SCA=specific combining ability;
GD=genetic distance; Min=minimum; Max=maximum; L1=K64r; L2=N3.2.3.3; L3=RS61P; L4=NAW5885; L5=2N3d; L6=2Kba;
L8=RA214P;T1=CML395;T3=CML539;T4=CML442;T5=CML537;T6=CZL03007;
T7=CML545; T8=CML538; T9 =CML444; T11=CML548; T12=CZL052.

Two crosses L1 (K64r) x T5 (CML537) and L4 (NAW5885) x T3 (CML539) exhibited high

MPH and yet the parental lines clustered into similar groups using SNP markers and they had
similar (N3 and A) heterotic groupings. A number of crosses between lines from similar
known heterotic groups exhibited high MPH and these are L7 (SC5522) x T1 (CML395), L4
(NAW5885) x T4 (CML442), L1 (K64r) x T3 (CML539), L1 (K64r) x T5 (CML537), L4

187
(NAW5885) x T5 (CML537), L4 (NAW5885) x T3 (CML539) and L8 (RA214P) x T3
(CML539). Some crosses such as L7 (SC5522) x T10 (CML536), L6 (2Kba) x T11
(CML548) and L4 (NAW5885) x T6 CZL03007 constituting of lines from known different
heterotic groups exhibited high MPH. L6 (2Kba) and T9 (CML444) are in similar known
heterotic groups (SC and B) and although SNP markers grouped them into separate groups
they exhibited low MPH and negative HPH.

Table 6.9 Mid- and high-parent heterosis for hybrids as well as known heterotic
groupings and grouping according to single nucleotide polymorphism
markers

Hybrid Known heterotic group Grouping according to SNP markers Field heterosis

P1 P2 P1 P2 MPH HPH
L5/T1 N3 B II b II b 88.41 74.60
L3/T9 SC B I II a 93.18 91.31
L7/T1 SC B II a II b 140.92 27.03
L7/T10 SC A II a II a 140.17 53.92
L6/T11 SC A II b II a 166.80 136.08
L4/T4 N3 A IIC II b 159.25 32.09
L1/T3 N3 A II b IIC 175.72 132.76
L1/T7 N3 A II b II a 88.07 56.05
L4/T6 N3 B IIC I 195.02 147.91
L1/T5 N3 A II b II b 124.83 78.71
L4/T5 N3 A IIC II b 112.72 65.96
L2/T11 N3 A II b II a 80.19 37.15
L5/T5 N3 A II b II b 88.41 74.60
L5/T7 N3 A II b II a 73.36 53.20
L4/T11 N3 A IIC II a 77.91 28.69
L4/T3 N3 A IIC IIC 111.97 75.22
L8/T1 N3 B II b II b 87.20 62.14
L3/T4 SC A I II b 69.65 60.94
L8/T3 N3 A II b IIC 127.62 74.99
L1/T6 N3 B II b I 89.94 60.51
L6/T9 SC B II b II a 22.20 -11.19
P1=parent 1; P2=parent 2; MPH=mid-parent heterosis; HPH=high-parent heterosis; L1=K64r; L2=N3.2.3.3.; L3=RS61P;
L4=NAW5885; L5=2N3d; L6=2Kba; L7=SC5522; L8=RA214P; T1=CML395; T3=CML539; T4=CML442; T5=CML537;
T6=CZL03007; T7=CML545; T9=CML444; T10=CML536; T11=CML548.

6.3.6 Correlation between genetic distance, specific combining ability, high- and mid-
parent heterosis and F1 grain yield
There was a non-significant correlation of GD with F1 grain yield, SCA, MPH and HPH
across all environments, under optimum and low N conditions (Table 6.10). Under drought

188
conditions GD was positively and significantly correlated with per se performance of hybrids
(0.38) and HPH (0.37). Pearson’s rank correlations of SCA effects with F1 grain yield across
all environments (0.31) and under drought (0.42) were significantly positive at P≤0.01 and
under optimum conditions (0.24) was significantly positive at P≤0.05. SCA correlation was
non-significant with F1 grain yield under low N conditions. MPH was positive and
significantly correlated with F1 grain yield across all environments (0.36), optimum (0.46)
and drought (0.72) conditions and the correlation was not significant again under low N
conditions. Highly significant and positive correlations were realised between HPH and F1
grain yield across all environments (0.57), under optimum (0.76), drought (0.77) and low N
conditions (0.64). MPH and HPH were significantly correlated with SCA in all environments
except under low N.

All the 72 single cross hybrids were used in construction of Figures 6.2 to 6.6. The linear
regression of per se performance of F1s on HPH and MPH was significant with R2 values of
0.59 and 0.52 respectively (Figure 6.2). SCA also established significant positive
associations as well as linear regressions with HPH and MPH (Figure 6.3) and with per se
performance of hybrids (Figure 6.4). The linear regressions of GD on HPH, MPH and SCA
were not significant with R2 values of 0.02 for both HPH and MPH and 0.008 for SCA
(Figures 6.5 and 6.6).

189
Table 6.10 Average mid- and high-parent heterosis, and correlation among F1 grain yield, mid- and high-parent
heterosis and specific combining ability for all hybrids across all environments, optimum, drought and low nitrogen
environments

Environment Average MPH Average HPH r(F1, SCA) r(F1, MPH) r(F1, HPH) r(SCA, HPH) r(SCA, MPH) r(SCA, GD) r(F1, GD) r(HPH, GD) r(MPH, GD)

All environments 112.29 76.73 0.31** 0.36** 0.57** 0.31** 0.48** 0.09 -0.02 0.13 0.16

Optimum 117.39 85.29 0.24* 0.46** 0.76** 0.25* 0.37** 0.08 -0.06 -0.07 0.02

Drought 26.08 10.34 0.42** 0.72** 0.77** 0.38** 0.37** 0.06 0.38** 0.37** 0.08

Low N 72.41 25.73 0.08 0.04 0.64** -0.07 -0.07 -0.03 -0.01 -0.01 0.14
**P≤0.01; *P≤0.05; MPH=mid-parent heterosis; HPH=high-parent heterosis; SCA=specific combining ability; F1=grain yield (t ha-1) for hybrids; GD =genetic distance; r=Pearson’s
coefficient of correlation.

190
 200.00

150.00 % MPH
y = 56.913x - 94.38
R² = 0.5186

100.00
% Heterosis

% HPH
y = 56.374x - 109.2
50.00 R² = 0.5868

0.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
MPH
-50.00 HPH
Linear (MPH)

-100.00 Linear (HPH)

F1 grain yield (t ha-1)

Figure 6.2 Relation of per se performance of hybrids with high- and mid-parent
heterosis under drought conditions.
MPH=mid-parent heterosis; HPH=high-parent heterosis.

250.00 % HPH
y = 28.084x + 77.598
R2 = 0.0939

200.00 % MPH
y = 47.831x + 113.5
R2 = 0.2256

150.00
% Heterosis

HPH

MPH
100.00

50.00

0.00
-1.50 -1.00 -0.50 0.00 0.50 1.00
SCA

Figure 6.3 Relation of specific combining ability with high- and mid-parent heterosis
across all environments.
MPH=mid-parent heterosis; HPH=high-parent heterosis; SCA=specific combining ability.

191
 1.50

y = 0.3843x - 0.8182
1.00 R² = 0.1744

0.50
SCA

0.00 SCA
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 Linear (SCA)

-0.50

-1.00

-1.50
F1 grain yield (t ha-1)

Figure 6.4 Relation of specific combining ability with per se performance of hybrids.
SCA=specific combining ability.

250.00
% HPH
y = 127.56x + 37.173
R2 = 0.0159
200.00

% MPH
% Heterosis

150.00 y = 169.92x + 59.741

R2 = 0.0234

100.00
HPH
MPH
50.00
Linear (HPH)
Linear (MPH)

0.00
0 0.1 0.2 0.3 0.4

Genetic distance

Figure 6.5 Relation of genetic distance with high- and mid-parent heterosis across all
environments.
MPH=mid-parent heterosis; HPH=high-parent heterosis.

192
 1.00

0.80

0.60
y = 0.995x - 0.3089
0.40 R² = 0.0081

0.20

0.00 SCA
SCA

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

-0.20 Linear (SCA)

-0.40

-0.60

-0.80

-1.00

-1.20
Genetic distance

Figure 6.6 Relation of genetic distance with specific combining ability across all
environments.
SCA=specific combining ability.

6.4 Discussion
Presence of a reasonable magnitude of heterosis for grain yield and related traits is critical in
any hybrid breeding programme. The degree of heterosis is therefore determined by genetic
diversity that exists within the germplasm being used. The average degree of MPH and HPH
ranged from 26.08% and 10.34% in drought environments to 117.39% and 85.29% in
optimum environments. The expression of heterosis was greater under optimum than stress
conditions. This could be explained by the fact that CIMMYT inbred lines used as testers
were bred for tolerance to stress conditions so they tended to perform well, thereby resulting
in an increase in parental and better parent means. Results are contrary with results reported
by Betran et al. (2003) and George et al. (2011) who reported higher MPH and HPH values
under severe stress than under non-stress environments using tropical germplasm and
CIMMYT South America maize programme germplasm, respectively. In the current study
intermediate drought stress was applied whereas Betran et al. (2003) applied both

193
intermediate and severe drought stress and in addition the materials evaluated and the sample
size was different. Most parents showed positive heterosis for grain yield across all
environments, indicating the existence of substantial heterosis in hybrids.

When all crosses were considered across all environments for grain yield MPH averaged
112.29% and HPH 76.40%. These values are higher than those reported by Legesse et al.
(2008) (MPH mean ranging between 0.9% and 77.2%) but lower than those reported by
Betran et al. (2003) (MPH 171% and HPH 132%). Grain yield is expected to show higher
levels of hybrid vigour compared to other traits as it has been suggested that it is a
multiplicative trait that draws variation from other traits (Williams, 1959; Lippman and
Zamir, 2007). Therefore it is assumed that lower values of heterosis detected for other traits
may interrelate in a non-linear manner to create better heterosis levels for yield (Flint-Garcia
et al., 2009). The levels of heterosis observed in this study indicate that there is an
opportunity of using the germplasm for developing hybrid varieties suitable for stress and
non-stress environments. Lines L1 (K64r), L2 (N3.2.3.3), L3 (RS61P), L4 (NAW5885) and
L7 (SC5522) and testers T3 (CML539), T9 (CML444) and T6 (CZL03007) were identified
as good parental lines for producing hybrids for non-stress environments. On the other hand
L3 (RS61P), L4 (NAW5885), L5 (2N3d), T11 (CML548), T6 (CZL03007) and T10
(CML536) were identified as potential parental lines for producing stress tolerant hybrids.

Hybrids showed negative MPH and HPH for anthesis days with averages of -5.1% and -7.4%
respectively. These results were desirable as it indicated that hybrids were earlier in anthesis
than their inbred parents and similar results were reported by Gissa (2008). Due to increasing
unpredictability of global weather patterns, especially amount of rainfall and its distribution,
use of early maturing varieties has become a better strategy for farmers to reduce risks
associated with drought. Early maturing maize provides options concerning inter-crops, relay
crops, late planted crops, drought avoidance and earlier harvesting (CIMMYT-Zimbabwe,
2000). Moderately high positive MPH and HPH for plant and ear height indicate the
superiority of dominance effects among the parental inbred lines for taller plant height
(Gissa, 2008). In the current study mean MPH and HPH for plant height (59.9% and 49.9%)
and ear height (53.3% and 42.7%) were reported and similar results were reported by Gissa

194
(2008) and Legesse et al. (2008). Gissa (2008) reported mean MPH and HPH for plant
height (57.1% and 46.8%) and ear height (62.6% and 49.8%), whilst Legesse et al. (2008)
reported MPH values for plant height ranging between 10.6% and 37.8%. Positive MPH and
HPH values for ears per plant reported in this study were also ideal as it is an indication that
hybrids are more prolific than their inbred parents.

Heterotic groups have been extensively used to streamline maize breeding (Tracy and
Chandler, 2006). The predefined heterotic groupings of lines did not consistently predict the
performance of hybrids e.g. L5 (2N3d) in the N3 group and T1 (CML395) in the B group
displayed low MPH (88.41%) and HPH (74.60%) means and this was also true for L8
(RA214P) in N3 group and T1 (CML395) in B group (MPH 87.2% and HPH (62.14%).
Dhliwayo et al. (2009) reported even lower MPH means with CIMMYT B x IITA A (4.97%)
and CIMMYT B x CIMMYT A (5.84%). However, there were cases when predefined
groupings resulted in corresponding high heterosis values e.g. L6 (2Kba) in the SC group and
T11 in the A group showed high MPH (166.80%) and HPH (136.08%) and L4 (NAW5885)
in the N3 group crossed with T6 ([CML445/ZM621B]-2-1-2-3-1-B*5) in the B group also
showed high MPH (195.02%) and HPH (147.91%). In this study heterotic groupings
determined by SNP markers also resulted in corresponding high MPH and HPH values but in
some cases the opposite was true e.g. L4 (NAW5885) and T4 (CML442) clustered in
different groups using SNP markers and showed high MPH (159.25%), whilst L7 (SC5522)
and T10 (CML536) clustered in the same group but displayed high MPH (140.17%). Such
phenomenon has been reported frequently in maize when genetically diverse parents produce
crosses with high heterosis (Hallauer and Miranda, 1988; Hallauer et al., 2010). In the
current study it was also noted that lines within the same heterotic group exhibited offspring
with lower HPH values and similar results were reported by Flint-Garcia et al. (2009).
Literature indicates that the use of heterotic groups and genetic distance to predict levels of
heterosis have been of limited success (Moll et al., 1965; Melchinger, 1999). Although
crosses between individuals in the same group have revealed lesser heterosis than crosses
between individuals in different groups there are many exceptions, making group identity a
poor predictor of heterosis (Flint-Garcia et al., 2009).

195
Research done earlier has shown that low levels of GD significantly reduced the levels of
heterosis (Moll et al., 1965) thereby prompting other researchers to think that GD can be
used to predict heterosis. Results from this study have revealed that there was significant
correlation between GD and heterosis under drought conditions and the following researchers
reported similar results: Betran et al. (2003), Amorim et al. (2006), Srdic et al. (2007) and
George et al. (2011). However, Betran et al. (2003) reported increased correlations under
non-stress compared to stress environments. The inbred lines used as testers in this study
were specifically screened for drought tolerance and as a result their performance compared
to lines used as females was different and this might explain the significant correlation of GD
and heterosis. In addition GDs reported in the current study had a narrow range and this
might also explain it being able to predict hybrid performance since in other studies
(Melchinger, 1999; George et al., 2011) it has been reported that a better correlation of GD
with MPH and HPH was observed among lines that are more closely related than among
lines that are distantly related. Different environments furthermore influence association of
GD and heterosis to different degrees, mainly because performance of parental lines and
hybrids varies across environments. Previous studies have seldom detected strong association
amongst heterosis and parental GD (Melchinger, 1999; Singh and Singh, 2004). Moll et al.
(1965), Melchinger (1999) and George et al. (2011) have suggested that the relationship
between GD and heterosis is complicated given that as the GD between the parental lines
increases the level of heterosis also increases up to a certain point after which heterosis starts
declining.

On the other hand there was non-significant correlation between GD and heterosis across all
environments as well as under optimum and low N conditions and similar results were
reported by Balestre et al. (2008), Dhliwayo et al. (2009) and Devi and Singh (2011). In the
current study there was a positive and significant correlation between GD and per se
performance of hybrids under drought conditions but generally the per se performance of F1s
was little influenced by GD as evidenced from non-significant negative correlations. These
results are an indication that meagre variation attributable to SCA, HPH, MPH and per se
performance of hybrids could not be explained due to SNP based GD of the parents. There
were significant positive correlations between SCA, per se performance of hybrids, MPH and

196
HPH across all environments and under optimum and drought conditions but not under low
N conditions. These results indicate that an improvement in selection for SCA will result in
an indirect improvement of MPH and HPH for hybrids under optimum, drought and across
environments. Under low N conditions the opposite is true. The non-significant correlation of
SCA, HPH and MPH with per se performance of hybrids under low N conditions might be
due to the low inbred line genetic variability caused by low yields and high error variability
associated with the stress environment. Contrary to results reported by Betran et al. (2003)
where r(F1, SCA) across environments was double that of r(F1, MPH) and r(F1, HPH), in this
study almost double r(F1, HPH) was reported. Similar to the results reported by Devi and
Singh (2011), MPH and HPH were also found to be key determinants of per se performance
of hybrids in this study. HPH correlation with per se performance was consistent in all
environments and according to Flint-Garcia et al. (2009) better parent heterosis is an
economically relevant trait. The HPH and MPH also established positive linear regressions
with per se performance of hybrids. As a consequence of the differential response of the
inbred lines under stress conditions the correlation that involved MPH was more erratic and
inconsistent than the HPH correlation.

6.5 Conclusions
The most important factor in developing adapted and high yielding hybrids is identifying
parental lines that combine well and result in superior varieties. The germplasm used
generally exhibited good heterosis for grain yield and other traits. Lines L1 (K64r), L2
(N3.2.3.3), L3 (RS61P), L4 (NAW5885) and L7 (SC5522) and testers T3 (CML539), T9
(CML444) and T6 (CZL03007) were identified as potential parental lines for producing
hybrids for optimum environments, whilst L3 (RS61P), L4 (NAW5885), L5 (2N3d), T11
(CML548), T6 (CZL03007), T9 (CML444) and T10 (CML536) were identified as potential
parents for producing hybrids for stress environments. The positive and significant
association of SCA and grain yield under optimum and drought conditions confirmed SCA as
a good predictor for grain yield of F1 hybrids. MPH had poor predictive value for grain yield
performance of hybrids under study, whilst HPH had a better predictive value. The GDs
found in this study were small and some of the lines had low homozygosity which might
have influenced the levels of heterosis realised in the study. A significant and positive SCA,

197
MPH and HPH association is an indicator that SCA can be used to predict MPH and HPH
during selections under both optimum and drought conditions which is a desirable selection
outcome. The negative MPH and HPH for days to anthesis showed that hybrids were earlier
than their parental inbred lines and this is a desirable outcome as the National Breeding
Programme is inclined towards breeding for early maturing hybrids. The majority of farmers
targeted by the breeding programme are situated in low rainfall areas characterised by erratic
mid-season droughts, so early maturing hybrids would be a better option for them. Even
though the correlation of SNP distances with HPH and MPH were significant under drought
conditions, the magnitudes were too low to be of high predictive value. SNP and
morphological distances were found to be useful in identifying closely related and distantly
related maize inbred lines but they were found to be of limited importance in predicting
HPH, MHP, SCA and per se performance of lines. The known heterotic groups are valid
since high heterosis could be observed from lines in the opposite groups, however confirming
their validity with the SNP data was not conclusive maybe due to genetic contamination of
some of the lines.

6.6 References
Allard, R.W. 1960. Principles of Plant Breeding. Wiley, NY.
Amorim, E.P., V.B.O. Amorim, J.B. dos Santos, A.P. de Souza and J.C. de Souza. 2006.
Genetic distance based on SSR and grain yield of inter and intrapopulation maize
single cross hybrids. Maydica 51: 507-513.
Balestre, M., R.G. Von Pinho, J. C. Souza and J.L. Lima. 2008. Comparison of maize
similarity and dissimilarity genetic coefficients based on microsatellite markers.
Genetics and Molecular Research 7: 695-705.
Beck, D.L., S.K. Vasal and J. Crossa. 1990. Heterosis and combining ability of CIMMYT’s
tropical early and intermediate maturity maize (Zea mays L.) germplasm. Maydica
35: 279-285.
Betran, F.J., D. Beck, M. Banziger and G.O. Edmeades. 2003. Genetic analysis of inbred and
hybrid grain yield under stress and non-stress environments in tropical maize. Crop
Science 43: 807-817.

198
Cheres, M.T., J.F. Miller, J.M. Crane and S.J. Knapp. 2000. Genetic distance as a predictor
of heterosis and hybrid performance within and between heterotic groups in
sunflower. Theoretical and Applied Genetics 100: 889-894.
CIMMYT-Zimbabwe. 2000. CIMMYT-Zimbabwe: 2000 Research Highlights. Harare.
Zimbabwe. pp 3-7.
Devi, P. and N.K. Singh. 2011. Heterosis, molecular diversity, combining ability and their
interrelationships in short duration maize (Zea mays L.) across the environments.
Euphytica 178: 71-81.
Dhliwayo, T., K. Pixley, A. Menkir and M. Warbuton. 2009. Combining ability, genetic
distances and heterosis among elite CIMMYT and IITA tropical maize inbred lines.
Crop Science 49: 1201-1210.
Flint-Garcia, S.A., E.S. Buckler, P. Tiffin, E. Ersoz and N.M. Springer. 2009. Heterosis is
prevalent for multiple traits in diverse maize germplasm. PLoS ONE 4: e7433.
Gissa, W.G. 2008. Genotypic variability and combining ability of quality protein maize
inbred lines under stress and optimal conditions. PhD thesis, Department of Plant
Sciences, Faculty of Natural and Agricultural Sciences, University of Free State,
South Africa.
George, M.L.C., F. Salazar, M. Warbuton, L. Narro and F.A. Vallejo. 2011. Genetic distance
and hybrid value in tropical maize under P stress and non-stress conditions in acid
soils. Euphytica 178: 99-109.
Hallauer, A.R. and F.J.B. Miranda. 1988. Quantitative Genetics in Maize Breeding. Iowa
State University Press, Ames, USA. pp 116-123.
Hallauer, A.R., M.J. Carena and J.B. Miranda Filho. 2010. Handbook of Plant Breeding:
Quantitative Genetics in Maize Breeding. Springer Science and Business Media,
LLC. New York. pp 477-519.
Lee, M., E.B. Godshalk, K.R. Lamkey and W.W. Woodman. 1989. Association of restriction
fragment length polymorphisms among maize inbreds with agronomic performance
of their crosses. Crop Science 29: 1067-1071.
Legesse, B.W., A.A. Meyburg, K.V. Pixley and A.M. Botha. 2008. Relationship between
hybrid performance and AFLP based genetic distance in highland maize inbred lines.
Euphytica 162: 313-323

199
Lippman, Z.B. and D. Zamir. 2007. Heterosis: revisiting the magic. Trends in Genetics 23:
90-66.
Liu, K. and S.V. Muse. 2005. PowerMarker: an intergrated analysis environment for genetic
marker analysis. Bioinformatics 21: 2128-2129.
Livini, C., P. Ajmone-Marsan, A.E. Melchinger, M.M. Messmer and M. Motto. 1992.
Genetic diversity of maize inbred lines within and among heterotic groups revealed
by RFLPs. Theoretical and Applied Genetics 84: 17-25.
Mashingaidze, K. 2006. Maize research and development. In: Rukuni, M., P. Tawonezvi and
C. Eicher (eds) Zimbabwe’s Agricultural Revolution Revisited. University Of
Zimbabwe Publication. pp 363-377.
Melchinger, A.E. 1999. Genetic diversity and heterosis. In: Coors, J.G. and S. Pandey (eds)
The Genetics and Exploitation of Heterosis in Crops. ASA, CSSA, and SSSA,
Madison. pp 99-118.
Melchinger, A.E., J. Boppenmaier, B.S. Dhillon, W.G. Pollmer and R.G. Herrman. 1992.
Genetic diversity for RFLPs in European maize inbreds. II Relation to performance of
hybrids within versus between heterotic group for forage traits. Theoretical and
Applied Genetics 84: 672-681.
Moll, R.H., J.H. Lonnquist, J.V. Fortuno and E.C. Johnson. 1965. The relationship of
heterosis and genetic divergence in maize. Genetics 52: 139-144.
SAS Institute. 2002. 2002-2008 by SAS Institute Inc., Cary, NC. USA.
Singh, R.K. and B.D. Chaudhary. 1977. Biometrical Methods in Quantitative Genetic
Analysis. Kalyani Publishers, New Delhi.
Singh, S.P. and M. Singh. 2004. Multivariate analysis in relation to genetic improvement in
Cuphea procumbens. Journal of Genetics and Breeding 58: 105-112.
Smith, O.S., J.S.C. Smith, S.L. Bowen, R.A. Tenborg and S.J. Wall. 1990. Similarities
among a group of elite maize inbreds as measured by pedigree, F1 grain yield,
heterosis and RFLPs. Theoretical and Applied Genetics 80: 833-840.
SPSS 15.0 for Windows. 2006. www.spss.com SPSS Inc., Chicago III.
Srdic, J., S. Miladenovic-Drinic, Z. Pajic and M. Filipovic. 2007. Characterisation of maize
inbred lines based on molecular markers, heterosis and pedigree data. Genetika 39:
355-363.

200
Tracy, W.F. and M.A. Chandler. 2006. The historical and biological basis of the concept of
heterotic patterns in corn belt dent maize. In: Lamkey, K.R. and M. Lee. (eds) Plant
Breeding: The Arnel R Hallauer International Symposium. Blackwell Publishing,
Ames, IA. pp 219-233.
Vasal, S.K., H. Cordova, S. Pandey and G. Srinivasan. 1999. Tropical maize and heterosis.
In: Coors, J.G. and S. Pandey (eds) The genetics and exploitation of heterosis in
crops. CSSA, Madison, WI. pp 363-373.
Williams, W. 1959. Heterosis and the genetics of complex characters. Nature 184: 527-530.

201
 CHAPTER 7

Performance of F3 testcrosses developed from CIMMYT drought tolerant

donors and Zimbabwean elite inbred lines

Abstract
Drought is one of the most devastating factors for maize production especially in sub-
Saharan Africa. Therefore the urgent need to breed for drought tolerant varieties cannot be
over-emphasised. In this study six DR&SS elite white maize inbred lines were crossed to two
CIMMYT Zimbabwe and seven CIMMYT Mexico drought tolerant donors to initiate a
segregating population after which selfing was done until the F3 generation. A total of 196
segregating lines belonging to heterotic group A were testcrossed to group B tester
CML444/CML395 and 209 segregating lines belonging to heterotic group B were testcrossed
to group A tester CML539/CML442 under isolation. All testcrosses were divided into early
and late maturing trials and evaluated under drought and optimum conditions using a 0.1
alpha lattice design with two replications across three environments in the 2011 winter
season. The objectives of this study were to determine agronomic performance of the
testcrosses under drought and well-watered conditions and to identify superior testcrosses for
grain yield and yield related traits. Significant (P≤0.001) mean squares for the majority of
traits were an indication of the presence of genetic variability amongst the testcrosses. The G
x E interaction mean squares were highly significant (P≤0.001) for the majority of traits.
Testcrosses containing the lines derived from DR&SS lines K64r, RS61P, N3.2.3.3 and
CIMMYT drought tolerant donors based on DTPWC9 were generally amongst the best
performing testcrosses in both trial sets under drought and well-watered conditions and
across environments. There was significant correlation of grain yield with secondary traits
such as anthesis days, anthesis silking interval and ears per plant under drought conditions.
Genetic variances and genetic gains for plant and ear height were generally higher than those
for other traits. Results therefore provide the basis for selection of segregating lines for grain
yield and related traits for further generation advancement.

202
7.1 Introduction
Maize is the world’s leading cereal crop with production of 695 million ton and per unit area
yield of 4 815 kg ha-1 (FAOSTAT, 2008). It is grown as a staple food crop in different agro-
ecological zones of Zimbabwe. Currently 80-90% of the total area under maize is attributable
to communal farmers and they encounter variations in rainfall distribution from year to year,
exposing their maize crop to drought. The 1981/82 drought reduced maize production by
about 70% (Rukuni et al., 2006) and hence emphasised the need to develop drought tolerant
varieties. The most elite inbred lines used by the Zimbabwe National Maize Breeding
Programme under DR&SS lack tolerance to drought and there exists a need to develop
drought tolerant varieties in the shortest possible time. Hence there is a need to use quick and
efficient methods that will enable identification and isolation of superior drought tolerant
genotypes.

The maize improvement programme involves development of lines, assessment of their

performance and selection of superior ones that are later used as parental lines (Pixley et al.,
2006). The presence of satisfactory genetic diversity and good quality genetic factors derived
from parents define progress from selection for improved adaptation and desirable agronomic
traits (Dreisigacker et al., 2004). Introgression of exotic maize into adapted breeding
germplasm can widen and diversify the genetic base of adapted germplasm (Menkir et al.,
2007). Crosses constituted from diverse genotypes maximise the number of segregating
alleles, causing large genetic variance in the offspring (Cox et al., 1985; Messmer et al.,
1993) thus increasing the chance for selecting genotypes that may be superior (Becelaere et
al., 2005). Exotic maize inbred lines with higher levels of diversity can lead to concentration
of alleles for increased yield potential, early maturity, short plant height, resistance to root
and stem lodging, tolerance to abiotic stresses (e.g. drought) and resistance to diseases and
insects (Menkir et al., 2007).

Estimates of genetic components of variance are useful for two aspects that are important to
applied breeding programmes and these are estimates of heritability and predicted response
to selection. Progress in plant breeding depends on the availability and maintenance of
genetic variability. Heritability expresses the proportion of the total variance that is

203
attributable to the average effects of genes and determines the resemblance between relatives
(Falconer, 1960). Repeatability can be defined as the proportion of the genotypic variance to
the total phenotypic variance. There are certain assumptions that are made in computing
repeatability and the first assumption is that the variances of different measurements are
equal and have their components in the same proportions and the second assumption is that
the different measurements reveal what the same character is genetically (Falconer, 1960).

As the numbers of lines to be tested at various stages of inbreeding increase over time, their
evaluation in all possible hybrid combinations is not feasible. Test cross performance of
experimental lines is the major selection criterion in hybrid maize breeding (Mihaljevic et al.,
2005). Two basic systems are used namely late and early testing. Late testing involves
evaluation of hybrid performance at later stages of inbreeding (S5-S6) and more testers are
used compared to early testing. Early testing of inbreds first proposed by Jenkins (1935) has
become a matter of great interest to maize breeders. In early testing, evaluation of hybrid
performance is conducted in early generations of inbreeding (S1-S3) whereby plants are
outcrossed to a tester and the resulting progeny evaluated for grain yield and general
performance. Early testing procedure is of significance where yield is an important
consideration or where other important factors can be evaluated easily and efficiently by a
suitable tester (Allard, 1960). Therefore test crossing has been adopted extensively to
evaluate the relative per se performance of inbred lines to aid in line advancement in
pedigree breeding. Hence the objectives of this study were to determine agronomic
performance of testcrosses derived from CIMMYT and DR&SS lines under drought and
well-watered conditions through pedigree breeding and to identify superior lines for
generation advancement.

7.2 Materials and methods

7.2.1 Germplasm
Plant selections were made at each generation and these were based on synchronisation
between pollen shedding and silking, low ear placement, well filled ears and resistance to
lodging and MSV. Six DR&SS elite inbred lines were crossed with nine CIMMYT drought

204
tolerant donors to develop a segregating population (Table 7.1). CIMMYT drought tolerant
donor lines were crossed to all the DR&SS lines and 158 F1s were produced. Selfing was
done and the ear to row technique was applied for generation advancement until the F3. The
F3 segregating population was grouped according to the CIMMYT heterotic grouping, after
which 196 lines belonging to group A were testcrossed to the group B tester
CML444/CML395 and 205 lines belonging to group B were testcrossed to group A tester
CML539/CML442. Testcrossing was done under isolation at Gwebi in Zimbabwe (17.130S,
310E, 1406 masl). Three hundred and twenty-four testcrosses were divided into maturity
groups (early and late) to constitute two trial sets. The early testcross trial constituted 223
entries and seven check varieties (013WH29, 023WH31, SC513, SC403, SC411, CZH0524
and CZH0946), whilst the late testcross trial constituted of 101 entries and seven check
varieties (013WH03, 013WH01, SC635, SC627, SC727, CZH0837 and CZH0616).

Table 7.1 Pedigree, source and heterotic grouping of the inbred lines used to develop
the F3 population

Inbred Pedigree Source Heterotic group

1 SC5522 DR&SS SC
2 N3.2.3.3 DR&SS N3
3 RS61P DR&SS SC
4 NAW5885 DR&SS N3
5 K64r DR&SS N3
6 RA214P DR&SS N3
7 CML442 CIMMYT-Harare A
8 CML444 CIMMYT-Harare B
9 DTPWC9-F16-1-1-1-1-BBB CIMMYT-Mexico B
10 DTPWC9-F104-5-4-1-1-BBB CIMMYT-Mexico B
11 DTPWC9-F92-2-1-1-1-BB CIMMYT-Mexico B
12 DTPWC9-F115-1-2-1-2-BBB CIMMYT-Mexico B
13 G16BNSeqC4-(F25x2)-4-1-4-B CIMMYT-Mexico A
14 G16BNSeqC4-(F20x17)-3-1-5-BB CIMMYT-Mexico A
15 LaPostaSeqC7-F103-1-2-1-1-BBB CIMMYT-Mexico B

7.2.2 Evaluation sites

Trials were evaluated under drought and optimal conditions during the winter of 2011. Three
sites were used: Chiredzi Research Station (21.02°S, 31.58°E, 433 masl), Chisumbanje
Research Station (20°S, 33°E, 455 masl) and Save Valley Research Station (200S, 320E, 556

205
masl). At each site two early and two late testcross trials were planted (one under managed
drought and another under optimal conditions).

7.2.3 Management
Trials under optimal conditions were grown under irrigation and water was applied as and
when it was necessary to ensure that the crop does not suffer from moisture stress. At all
three sites drought was managed through irrigation at critical times only. A total of 280 mm
irrigation was applied in the first eight weeks of the crop’s growth for the trials under
managed drought. This resulted in drought coinciding with flowering and grain filling. The
stress level projected to be achieved in this trial was a yield of about 15- 20% of yields
achieved under optimal conditions. This stress level delays silking and causes ear abortion in
non-stress tolerant genotypes. The testcross evaluation was done using the 0.1 alpha lattice
design. There were two replications in each trial, with each entry being planted in a one row
plot 4 m long, with a 90 cm between row and 30 cm in-row spacing. Two seeds were planted
per station and later thinned to give a plant population of 48 000 plants ha-1. A total of 400 kg
ha-1 maize fertiliser was applied as basal dressing and 350 kg ha-1 ammonium nitrate as
topdressing split applied at four and eight weeks after crop emergence. Two applications of
Dipterex 2.5% granules into the funnel of each plant were done at three and six weeks after
crop emergence and the rate of application used was 4 kg ha-1. Termite control was done as
and when necessary using Carbaryl 85% wettable powder.

7.2.4 Data collection and analysis

Data for flowering dates (at 50% anthesis and 50% silking), plant and ear height, plant stand
ability (root and stem lodging), leaf senescence, ear aspect, texture, ear rot and grain weight
were recorded as described in section 3.2.4 and Table 3.2. Some derived traits such as
anthesis silking interval, lodging percentage, ears per plant and yield per hectare (at 12.5%
moisture adjustment) were also calculated. The individual and combined site ANOVA was
done using AGROBASE (2005). In the combined ANOVA, testcrosses and locations were
considered as fixed effects while replications and years were considered as random effects.
The variance components were computed following Wricke and Weber (1986).

206
Genetic variance:
σ2g = (M5 + M2- M3- M4)/rly
Where: σ2g = genetic variance
M5 = σ2e + rσ2gly + ry σ2gl + rl σ2gy + rly σ2g
M4 = σ2e + rσ2gly + rl σ2gy
M3 = σ2e + rσ2gly + ry σ2gl
M2 = σ2e + rσ2gly
rly = replication x location x year
gly = genotype x location x year
gl = genotype x location
gy = genotype x year

Genotype x year variance:

σ2gy = ( M4 – M2)/rl
Where: rl = replication x location

Genotype x location variance:

σ2gl = (M3 – M2)/ry
Where: ry = replication x year

Genotype x location x year variance:

σ2gly = (M2 – M1)/r
Where: r = replication

Error variance:
σ2e = M1

Phenotypic variance:
σ2P = σ2g + σ2gl + σ2gy + σ2gly + σ2e
L Y LY RLY

207
Repeatability:
b = σ2g/σ2 P
Where: σ2g = genotypic variance
σ2 P = phenotypic variance
Genetic gain:
∆g = k σ2p b
Where: k = selection intensity
σ2p = phenotypic variance
b= repeatability

In this case repeatability is taken to be the same as broad sense heritability. Pearson
correlation coefficients (r) between grain yield and secondary traits were calculated from
means across environments and per environment. Statistical computations were performed
with SPSS 15.0 version for Windows (2006).

7.3 Results

7.3.1 Performance of early maturing testcrosses under managed drought conditions

The means presented are for the 10 best and 10 poorest genotypes in terms of grain yield and
overall grand means for each data set are presented. The early maturing testcrosses planted at
Chiredzi Research Station under managed drought stress conditions were severely affected
by heat scorch soon after water stress was induced at two weeks bracketing flowering,
resulting in most of the entries completely failing to flower. As a result grain yield was
recorded for only a few entries and the site was therefore excluded from the analysis. The
ANOVA for grain yield under managed drought stress conditions for Chisumbanje and Save
Valley Research Stations is presented in Table 7.2.

The two sites, genotypes and G x E interaction were significant (P≤0.001) for grain yield.
Results for anthesis days and other agronomic traits across the three sites under managed
drought stress are presented in Tables 7.3 and 7.4. Sites were significantly different for all
traits (P≤0.001). Genotypes were significantly different at P≤0.001 for plant height, ear
height, root lodging, stem lodging, ears per plant and ear aspect, significantly different at

208
P≤0.01 for ear rot and significantly different at P≤0.05 for senescence and texture. However
genotypes were not significantly different for anthesis days and anthesis silking interval. The
G x E interaction was significant for all traits except anthesis silking interval, senescence and
texture.

Table 7.2 Analysis of variance for grain yield under managed drought conditions at
Chisumbanje and Save Valley for early maturing testcrosses in the 2011
winter season

Source DF MS

Site 1 109.68***

Genotype 229 0.55***

GxE 229 0.66***

Residual 278 0.24

***P≤0.001; DF=degrees of freedom; MS=mean square; Env=environment; G x E=genotype by environment interaction.

Table 7.3 Analysis of variance for anthesis days and other agronomic traits under
managed drought conditions across three sites for early maturing testcrosses
in the 2011 winter season

Source DF AD ASI PH EH RL SL SEN

Site 2 163938.91*** 1830.03*** 22237.52*** 5451.52*** 46510.69*** 83817.67*** 14.948***

Genotype 229 12.86 13.51 250.72*** 293.06*** 171.42*** 304.34*** 0.027*

GxE 458 12.12** 13.30 237.76*** 255.76*** 175.83*** 231.60*** 0.025

Residual 417 9.45 12.25 3.52 105.10 57.82 87.40 0.022

***P≤0.001; **P≤0.01; *P≤0.05; DF=degrees of freedom; AD=anthesis days; ASI=anthesis silking interval; PH=plant height;
EH=ear height; RL=root lodging; SL=stem lodging; SEN=senescence; Env=environment; GxE=genotype by environment
interaction.

209
Table 7.4 Analysis of variance for ears per plant, ear aspect, texture and ear rot under
managed drought conditions across three sites for early maturing test crosses
in the 2011 winter season

Source DF EPP EA TEX ER

Site 2 15.750*** 51.95*** 37.80*** 99206.70***
Genotype 229 0.038*** 1.07*** 1.03* 632.77**
GxE 458 0.034*** 1.01*** 0.93 602.51**
Residual 417 0.005 0.71 0.82 456.72
***P≤0.001; **P≤0.01; *P≤0.05; DF=degrees of freedom; EPP=ears per plant; EA=ear aspect; TEX=texture; ER=ear rot;
Env=environment; G x E=genotype by environment interaction.

The means for the best and poorest performing early maturing testcrosses under drought are
presented in Table 7.5. The genotype G122 (DTPWC9-F104-5-6-1-1-B*4/K64r)-B-
3//CML444/CML395 was the best performer in terms of grain yield under managed drought
conditions with a mean yield of 2.68 t ha-1 followed by G25 (DTPWC9-F16-1-1-1-1-
BBB/RS61P)-B-23//CML539/CML442 with a mean yield of 2.51 t ha-1. The 10 best
genotypes performed well above the trial grand mean of 1.53 t ha-1. The poorest performer
was G215 (DTPWC9-F104-5-6-1-1-B*4/NAW5885)-B-4//CML444/CML395 with an
average mean yield of 0.70 t ha-1 and the genotype had a poor ear aspect score of 4.3 and ear
rot percentage of 35.3%. Six of the ten best performing testcrosses constituted of K64r sister
lines and the rest were two RS61P sister lines, one N3.2.3.3 line and one NAW5885 line,
whilst the majority of testcrosses within the ten poorest performing testcrosses were
NAW5885 sister lines. Most lines within the ten poorest performing lines had ear rot scores
of above 20% and ear aspect scores of above 3.5. Heritability scores generally showed that
genotypes played a minor role in expression of traits leaving the environment to play a major
role and this was further confirmed by the lower genotypic variances compared to phenotypic
variances. The R2 values were above 0.50 for all traits measured under managed drought
conditions.

210
Table 7.5 Performance of early maturing maize testcrosses for grain yield and other agronomic traits under managed
drought

Genotype GYD PH EH RL SL SEN EPP EA ER

Ten best testcrosses G122 2.68 170.8 95.0 13.7 23.1 0.58 0.41 3.4 26.0
G25 2.51 17.8 81.7 8.3 27.0 0.55 0.62 3.8 39.9
G131 2.44 161.7 82.5 10.7 26.2 0.55 0.53 3.6 40.8
G14 2.44 180.8 92.5 6.3 25.1 0.55 0.53 3.5 25.4
G51 2.36 165.0 89.2 1.5 24.8 0.53 0.46 3.1 26.8
G46 2.36 170.8 92.5 15.0 22.1 0.55 0.55 3.7 18.6
G128 2.28 179.2 95.0 15.2 20.7 0.55 0.57 3.3 10.5
G209 2.28 180.0 104.2 14.7 15.9 0.58 0.51 3.3 13.7
G190 2.28 171.7 90.8 31.9 9.0 0.52 0.60 3.3 16.1
G114 2.27 178.3 101.7 16.3 16.4 0.53 0.48 3.9 23.6

Ten poorest testcrosses G129 0.94 165.8 80.8 7.1 22.2 0.55 0.34 3.9 32.7
G196 0.94 160.8 93.3 15.8 12.5 0.57 0.33 3.5 26.7
G43 0.94 165.0 90.8 10.7 17.7 0.60 0.44 4.0 48.5
G162 0.94 170.8 100.8 10.6 16.3 0.53 0.40 3.4 24.9
G40 0.94 160.8 84.2 2.6 18.8 0.58 0.40 4.3 44.5
G154 0.86 170.8 82.5 2.0 17.7 0.53 0.36 4.4 34.9
G126 0.86 175.0 104.2 6.4 18.4 0.58 0.25 3.7 24.2
G100 0.86 166.7 92.5 9.9 16.4 0.55 0.42 3.8 27.9
G65 0.71 178.3 107.5 16.9 7.6 0.55 0.42 3.9 28.4
G215 0.70 167.5 90.0 17.7 8.2 0.48 0.35 4.3 35.3
Grand mean 1.53 170.7 95.3 11.9 17.8 0.56 0.45 3.7 28.1
LSD 0.57 5.34 9.79 7.2 8.9 0.14 0.07 0.8 20.3
SED 0.35 3.24 5.94 4.39 5.40 0.090 0.040 0.49 12.34
CV% 31.84 3.29 10.80 33.5 32.6 26.36 16.56 23.0 46.0
2
R 0.90 0.96 0.86 0.92 0.92 0.86 0.97 0.79 0.81
GYD=grain yield (t ha-1); PH=plant height (cm); EH=ear height (cm); RL=root lodging (%); SL=stem lodging (%); SEN=senescence (%); EPP=ears per plant (#); EA=ear aspect (1-5);
ER=ear rot (%); LSD=least significant difference;SED=standard error difference; CV=coefficient of variation; R2=coefficient of determination.

211
7.3.2 Performance of early maturing testcrosses under well-watered conditions
The ANOVA for grain yield and other agronomic traits are presented in Tables 7.6 and 7.7.
Sites were significantly different (P≤0.001) for all traits except anthesis silking interval.
Genotypes were significantly different (P≤0.001) for grain yield, plant height, ear height, ear
position, ear aspect, ear rot and texture, whilst they were not significantly different for
anthesis silking interval and ears per plant. The G x E was not significant for anthesis days
and anthesis silking interval. Genotype means for grain yield and other agronomic traits are
presented in Table 7.8. The grain yield trial mean was 3.85 t ha-1 and the two best performing
genotypes G82 ((NAW5885/CML442)-B-9//CML444/CML395) and G206 ((DTPWC9-F66-
2-1-1-2-BBB/N3233)-B-2//CML444/CML395) performed 41.8% above the trial mean with a
mean yield of 5.46 t ha-1, whilst the poorest performing genotype G58 ((DTPWC9-F92-2-1-
1-1-BB/K64r)-B-11//CML539/CML442) performed 57.1% below the trial mean with a mean
yield of 2.20 t ha-1. Testcrosses that featured within the ten best performers in terms of grain
yield were mainly constituted from K64r, RS61P, NAW5885 and N3.2.3.3 with DTPWC9
donor lines. Testcrosses within the ten best performers all showed ear aspect scores of below
3.5 and the texture scores showed that genotypes ranged from semi-flint to semi-dent. The
ear rot scores for all genotypes were generally below 30% with a trial mean of 17%. The R2
values were again high and were all above 0.50. The coefficient of variation values were
below 20% for all traits except ear rot.

7.3.3 Performance of early maturing testcrosses across environments

In the combined ANOVA environments were significantly different (P≤0.001) for grain
yield, anthesis days, plant height, ear height, ear position, ears per plant, root and stem
lodging, senescence and ear rot except anthesis silking interval (Tables 7.9 and 7.10).
Genotype as well as G x E mean squares were significant for all traits. All other interactions
were also highly significant. The genotype G111 ((DTPWC9-F104-5-4-1-1-BBB/K64r)-
B33//CML444/CML395) was the overall best performer in terms of grain yield with a mean
yield of 4.17 t ha-1 and this was 39.9% higher than the trial mean (Table 7.11).

212
Table 7.6 Analysis of variance for grain yield for early maturing testcrosses under
well-watered conditions in the 2010/11 season

Source DF MS P
Site 1 109.68 ***
Genotype 229 0.55 ***
GxE 229 0.66 ***
Residual 278 0.24
***P≤0.001; Env=environment; G x E=genotype by environment interaction;; DF=degrees of freedom; MS=mean square;
P=probability.

Table 7.7 Analysis of variance for anthesis days and other agronomic traits under well-
watered conditions across three sites for early maturing testcrosses in the
2011 winter season

Source DF AD ASI PH EH EPO EPP EA ER TEX

Site 2 41325.50*** 9227.75 561563.97*** 118377.88*** 0.19*** 0.95*** 3.37*** 6553.21*** 114.46***

Genotype 229 10.30* 9501.18 374.81*** 362.70*** 0.07*** 0.011 0.41*** 114.08*** 0.28***

GxE 458 9.02 9511.65 265.50*** 222.37*** 0.006*** 0.01* 0.35*** 104.53*** 0.30***

Residual 417 7.93 9476.79 26.43 58.7 0.004 0.01 0.16 68.75 0.12
***P≤0.001; *P≤0.05; DF=degrees of freedom; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height;
EPO=ear position; EPP=ears per plant; EA=ear aspect; ER=ear rot; TEX=texture; Env=environment; G x E=genotype by
environment interaction.

The genotype G111 also ranked fifth under well-watered conditions. It also exhibited good
stand ability qualities as indicated by root and stem lodging percentages of less than 20% as
well as good prolificacy (ears per plant mean) above the trial mean. Genotypes G131 ((DTP
WC9-F104-5-6-1-1-B*4/K64r)-B13//CML444/CML395), G25 ((DTPWC9-F16-1-1-1-1-
BBB/RS61P)-B- 23//CML539/CML442) and G209 ((DTPWC9-F104-5-4-1-1-BBB/N3233)-
B-3//CML444/CML395) were within the ten best performing testcrosses both under
combined analysis as well as under managed drought conditions. The poorest performing
genotype under drought and well-watered conditions was G69 ((N3233/CML442)-B-
12//CML444/CML395) with a mean yield of 1.84 t ha-1 and it was also amongst the poorest
performing testcrosses under drought conditions.

213
Table 7.8 Performance of early maturing testcrosses for grain yield and other agronomic traits under optimum conditions

Genotype GYD AD PH EH EPO EA ER TEX

Ten best testcrosses G82 5.46 73.5 173.3 95.8 0.55 3.2 19.8 2.9
G206 5.46 72.3 168.3 93.3 0.57 3.3 7.1 3.0
G24 5.30 72.7 169.2 90.0 0.55 3.0 15.0 3.1
G211 5.22 71.5 163.3 89.2 0.55 3.3 17.9 3.0
G111 5.19 74.2 175.8 100.8 0.58 3.2 15.7 2.9
G79 5.06 73 175.0 92.5 0.52 3.1 11.6 3.1
G96 5.06 70.3 180.0 94.2 0.53 3.3 14.9 2.9
G143 4.98 73.3 170.8 96.7 0.55 3.4 12.8 3.0
G95 4.98 73 174.2 98.3 0.58 3.0 14.6 2.8
G29 4.95 72.3 168.3 90.8 0.57 3.0 11.2 3.2

Ten poorest testcrosses G179 2.67 74.5 170.0 90.0 0.53 3.9 13.9 3.3
G50 2.67 72.7 152.5 73.3 0.48 3.7 19.4 3.6
G64 2.60 75.2 170.0 97.5 0.57 3.8 21.2 3.3
G176 2.59 72.3 175.0 101.7 0.58 4.2 20.3 3.5
G38 2.59 71.3 156.7 80.0 0.53 3.7 16.5 3.3
G154 2.59 75.3 163.3 90.8 0.57 3.9 24.1 3.6
G167 2.44 73 161.7 93.3 0.58 3.2 11.2 3.0
G56 2.35 73 158.3 85.8 0.57 3.8 19.1 3.5
G62 2.28 72.3 169.2 93.3 0.57 3.6 17.6 3.5
G58 2.20 72.7 175.0 95.0 0.55 4.0 24.6 3.4
Grand Mean 3.85 72.8 169.3 93.3 0.55 3.4 17.0 3.2
LSD 0.60 2.68 4.89 7.29 0.06 0.39 7.89 0.34
CV% 13.42 3.87 3.04 8.21 11.42 11.82 48.8 11.14
2
R 0.96 0.96 0.99 0.96 0.81 0.87 0.83 0.91
LSD 0.57 2.68 4.89 7.29 0.06 0.39 7.89 0.34
SED 0.35 1.63 2.97 4.42 0.037 0.23 4.79 0.20
GYD=grain yield (t ha-1); AD=anthesis days; PH=plant height (cm); EH=ear height (cm); EPO=ear position (0-1); EA=ear aspect (1-5); ER=ear rot (%); TEX=texture (1-5); LSD =least
significant difference; CV=coefficient of variation; R2=coefficient of determination; LSD=least significant difference; SED=standard error deviation.

214
Table 7.9 Analysis of variance for grain yield across environments for early maturing
testcrosses
Source DF MS
Site 1 48.17***
Genotype 229 1.15***
Treatment 1 3843.01***
GxE 229 1.30***
ExT 1 473.13***
G xT 229 1.16***
G x E xT 229 1.06***
Residual 0.06
***P≤0.001; DF=degrees of freedom; MS=mean square; Env=environment; G x E=genotype by environment interaction; G x T=
genotype by treatment interaction; G x E x T=genotype by environment by treatment interaction.

Genotype G69 had ears per plant mean (0.50) that was below the trial mean (0.62). The
majority of genotypes within the ten best performing genotypes had DTPWC9 donor as one
of the parents.

7.3.3.1 Variance components for early maturing testcrosses

The mean grain yield under drought conditions for the early maturing testcrosses was 40% of
the mean grain yield under optimum conditions (Figure 7.1). The general overview of the
performance of the testcrosses under drought conditions is presented in Figure 7.2. The
majority of testcrosses performed below the trial mean (1.53 t ha-1). Plant height, ear height
and ears per plant had repeatability values higher than 0.20. Genetic gain expected for grain
yield was 10 g in grain yield per plant.

7.3.3.2 Correlation between grain yield and secondary traits for early maturing
testcrosses under managed drought conditions
Grain yield was highly but negatively correlated (P≤0.01) with anthesis silking interval
(Table 7.13). Correlation with anthesis days was negative but non-significant. Ears per plant
were highly significant and positively correlated with grain yield. Anthesis days were
significantly and negatively correlated (-0.159) with anthesis silking interval. Senescence
was highly and negatively correlated with ears per plant.

215
Table 7.10 Analysis of variance for anthesis days and other agronomic traits under combined environments for early
maturing testcrosses

Source DF AD ASI PH EH EPO EPP RL SL SEN ER

Site 2 157649.36*** 1423.83 234310.72*** 45933.34*** 0.12*** 12.15*** 48687.77*** 96193.90*** 7.47*** 77730.76***
Genotype 229 12.47*** 4743.49*** 364.13*** 438.01*** 0.01*** 0.02** 123.31*** 153.41*** 0.01*** 378.27***
Treatment 1 14357.62*** 8102.70 1370.66*** 2700.33*** 0.03*** 83.82*** 37945.15*** 111929.83*** 216.50*** 85072.63***
GxE 458 12.77*** 4754.42*** 266.72*** 255.66*** 0.01*** 0.02*** 126.61*** 158.67*** 0.01*** 338.66***
E xT 2 47615.05*** 9633.95* 349490.77*** 77896.06*** 0.08*** 4.54*** 8094.54*** 10112.51*** 7.47*** 28029.15***
G xT 229 7.21*** 4771.21*** 261.40*** 217.76*** 0.01*** 0.03*** 108.92*** 125.19*** 0.01*** 368.58***
GxExT 458 8.36*** 4770.53*** 236.53*** 222.47*** 0.01*** 0.02*** 111.41*** 133.36*** 0.01*** 368.38***
Residual 5.26 2871.50 17.54 49.84 0.003 0.004 230.02 27.35 0.01 159.01
***P≤0.001; **P≤0.01; *P≤0.05; DF=degrees of freedom; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; EPP=ears per plant;
RL=root lodging; SL=stem lodging; SEN=senescence; ER=ear rot; Env=environment; G x E=genotype by environment interaction; G x T=genotype by treatment; G x E x T=genotype
by environment by treatment.

216
 Table 7.11 Performance of early maturing testcrosses for grain yield and other agronomic traits under drought and
optimum conditions in the 2010/11 season
Genotype GYD AD ASI PH EH EPO EPP RL SL SEN ER
Ten best testcrosses G111 4.17 75.8 1.3 175.4 95.4 0.56 0.69 10.4 13.7 0.28 22.9
G131 3.87 75.2 0.6 165.0 90.4 0.56 0.65 5.9 18.3 0.28 27.7
G211 3.79 73.5 1.9 166.3 90.0 0.54 0.67 5.1 13.7 0.26 23.4
G88 3.79 75.8 1.5 167.9 97.9 0.58 0.63 9.5 11.1 0.28 26.8
G25 3.78 75.1 1.3 167.5 84.2 0.52 0.68 7.6 15.6 0.28 30.7
G11 3.71 74.9 1.8 171.3 98.3 0.58 0.65 4.3 12.8 0.28 13.7
G79 3.67 74.9 1.3 180.4 99.6 0.54 0.66 3.9 17.3 0.28 20.1
G219 3.64 75.2 2.2 163.8 92.5 0.57 0.62 11.8 5.6 0.28 20.3
G102 3.63 75.5 1.8 163.3 90.8 0.57 0.69 6.2 10.5 0.27 32.1
G209 3.62 74.9 1.4 171.7 96.7 0.57 0.67 12.0 10.5 0.29 14.1
Ten poorest
testcrosses G58 2.32 75.6 2.4 178.8 100.4 0.56 0.51 5.9 8.5 0.28 20.9
G164 2.31 74.5 1.9 170.0 90.8 0.54 0.63 8.2 12.3 0.26 17.9
G104 2.31 75.8 2.0 169.2 91.7 0.55 0.6 18.1 9.1 0.26 20.3
G65 2.28 75.5 1.9 176.7 105.0 0.61 0.61 9.1 5.0 0.28 23.3
G213 2.28 75.8 1.3 169.6 95.4 0.58 0.61 6.5 17.3 0.29 27.8
G228 2.08 82.4 -6.5 176.7 98.3 0.57 0.65 8.6 13.7 0.28 22.1
G21 2.04 76.3 0.7 172.1 92.1 0.53 0.56 7.4 8.1 0.28 22.8
G154 1.88 76.3 2.8 167.1 86.7 0.53 0.57 6.8 11.0 0.27 29.5
G157 1.88 74.5 4.2 161.3 87.1 0.53 0.53 10.6 9.1 0.28 19.4
G69 1.84 75.0 1.3 172.9 99.2 0.58 0.50 8.4 16.6 0.32 22.3
Mean 2.98 75.1 0.58 170.0 94.3 0.56 0.62 8.28 11.4 0.28 22.5
LSD 0.19 1.54 0.60 2.80 4.74 0.04 0.04 3.22 3.51 0.05 8.47
CV% 8.01 3.05 9.10 2.46 7.48 9.38 10.75 57.98 45.84 29.00 55.91
R2 0.99 0.98 0.69 0.99 0.92 0.80 0.97 0.93 0.94 0.97 0.83
-1
GYD=grain yield (t ha ); AD=Anthesis days; anthesis silking interval (days); PH=plant height (cm); Ear height (cm); EPO=ear position (0-1); EPP=ears per plant (#); RL=root lodging (%); SL=stem
lodging (%); SEN=senescence (1-10); ER=ear rot (%); LSD=least significant difference; CV=coefficient of variation; R2=coefficient of determination.

217
 4.5

3.5

3
Grain yield t ha-1

2.5

1.5

0.5

0
Optimum Drought Combined
Treatment

Figure 7.1 Mean grain yield for early maturing testcrosses under well-watered,
drought and combined sites.

120

100
Number of testcrosses

80

60

40

20

0
0.5-1.0 1.1-1.5 1.6-2.0 2.0-2.5 2.5+ Mean (1.53)
Yield t ha-1

Figure 7.2 Mean grain yield for early maturing testcrosses under drought conditions
across two sites in the 2011 winter season.

218
Table 7.12 Genetic and phenotypic variance, repeatability and genetic gain for early
maturing testcrosses for the measured traits

Trait σ2g σ2 p b ∆g

GYD 0.03 0.21 0.14 0.01

AD 0.11 6.28 0.02 0.03
ASI 1.45 599.21 0.002 0.24
PH 9.07 44.27 0.20 1.77
EH 23.39 66.18 0.35 4.6
EPP 0.00125 0.004 0.31 0.0002
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH= ear height; EPP=ears per plant; σ2g=
genetic variance; σ2p=phenotypic variance; b=repeatability; ∆g=genetic gain.

Table 7.13 Correlation coefficients between grain yield and secondary traits under
managed drought conditions
GYD AD ASI EPP
AD -0.1
ASI -0.178** -0.159*
EPP 0.632** -0.36 -0.086
SEN -0.38 -0.022 -0.143* -0.177**
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant; SEN=senescence.

7.3.4 Performance of late maturing testcrosses under drought conditions

The ANOVA for grain yield and other agronomic traits for late maturing testcrosses are
presented in Table 7.14. Environments were significantly different (P≤0.001) for grain yield,
anthesis days, ear height, ear position, ears per plant, ear rot and ear aspect. Genotype and
the G x E interaction mean squares were significant (P≤0.001) for all traits except anthesis
days and anthesis silking interval. The late maturing testcrosses were mainly dominated by
SC5522 derived lines. G107 (CZH0837), a CIMMYT experimental three-way hybrid used as
a check variety, was the best performer in terms of grain yield under drought conditions with
a mean yield of 3.21 t ha-1 (Table 7.15). The second best performing testcross was G29
((DTPWC9-F104-5-4-1-1-BBB/SC5522)-B-2//CML539/CML442) with a mean yield of 3.05
t ha-1 and it performed 40.5% above the trial mean of 2.17 t ha-1. The rest of the testcrosses
within the ten best performing genotypes all performed above the trial mean.

219
Table 7.14 Analysis of variance for grain yield and other agronomic traits for late maturing testcrosses under drought
conditions in the 2011 winter season

Source DF GYD AD ASI EH EPO EPP ER EA

Site 2 180.48*** 29188.95*** 18.64 32277.30*** 0.19*** 1.89*** 14050.50*** 83.17***

Genotype 107 0.94*** 128.43 123.24 506.50*** 0.009*** 0.04*** 159.74*** 0.44***

GxE 214 0.68*** 119.42 122.59 180.63*** 0.005*** 0.024*** 131.23*** 0.47***

Residual 219 0.09 109.94 118.69 118.04 0.004 0.015 68.68 0.24
***P≤0.001; DF=degrees of freedom; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; EH=ear height; EPO=ear position; EPP=ears per plant; ER=ear rot; EA=ear
aspect; Env=environment; G x E=genotype by environment interaction.

220
Table 7.15 Performance of late maturing testcrosses for grain yield and other agronomic traits under drought conditions
Genotype GYD EH EPO EPP ER EA
Ten best testcrosses G107 3.21 100 0.52 0.73 16.1 4.1
G29 3.05 95.8 0.50 0.68 15.0 3.7
G92 2.87 120.8 0.60 0.67 9.4 3.8
G14 2.86 95.0 0.49 0.69 14.3 3.8
G32 2.81 95.0 0.49 0.64 8.7 4.0
G26 2.78 103.3 0.54 0.75 8.1 3.5
G38 2.75 88.3 0.51 0.67 10.9 3.3
G39 2.74 91.7 0.50 0.75 9.4 4.0
G88 2.69 115.0 0.59 0.79 9.5 3.8
G51 2.69 109.2 0.54 0.64 10.7 3.7

Ten poorest testcrosses G105 1.62 95.0 0.51 0.64 15.9 3.8
G48 1.60 101.7 0.54 0.53 19.5 4.3
G50 1.58 85.8 0.49 0.49 15.7 4.2
G99 1.56 104.2 0.55 0.58 12.1 3.8
G95 1.54 106.7 0.56 0.52 10.1 3.2
G66 1.50 113.3 0.60 0.52 9.3 4.3
G11 1.34 100.0 0.53 0.47 5.7 4.2
G103 1.28 105.8 0.56 0.33 29.2 3.3
G56 1.21 80.8 0.44 0.56 27.9 4.6
G85 1.16 115.0 0.61 0.51 10.7 4.2
Mean 2.17 102.0 0.54 0.63 13.6 3.9
LSD 0.29 10.36 0.05 0.12 7.90 0.47
CV% 14.05 10.65 10.96 19.53 61.00 12.48
R2 0.97 0.88 0.79 0.85 0.86 0.88
-1
GYD=grain yield (t ha ); EH=ear height (cm); EPO=ear position (0-1); EPP=ears per plant (#); ER=ear rot (%); EA=ear aspect (1-5); LSD=least significant difference; CV=coefficient
of variation; R2=coefficient of determination.

221
The poorest performing testcross was G85 ((DTPWC9-F115-1-2-1-2-BBB/NAW5885)-B-
7//CML444/CML395) with a mean yield of 1.16 t ha-1 and the second poorest performing
testcross was G56 ((MAS[206/312]-23-2-1-1-B*7/RA214P)-B-1//CML539/CML442) with a
mean yield of 1.21 t ha-1. Testcrosses generally showed low ear position values. The ear
aspect scores for most testcrosses were above 3.5.

7.3.5 Performance of late maturing testcrosses under optimum conditions

The testcrosses were significantly different (P≤0.001) for all traits except for ear aspect
(Tables 7.16 and 7.17). Environments were significantly different (P≤0.001) for all
measured traits. The G x E interaction was significant for all traits except ear aspect. The best
performing testcross in terms of grain yield under optimum conditions was G12 ((DTPWC9-
F16-1-1-1-1-BBB/SC5522)-B-6//CML539/CML442) with a mean yield of 5.41 t ha-1
followed by G76 ((DTPWC9-F16-1-1-1-1-BBB/NAW5885)-B-9//CML444/CML395) with a
mean yield of 5.38 t ha-1 (Table 7.18). The two testcrosses generally showed good values for
anthesis silking interval, root and stem lodging and ear rot. The poorest performing genotype
was G56 ((MAS[206/312]-23-2-1-1-B*7/RA214P)-B-1//CML539/CML442) with a mean
yield of 1.88 t ha-1 and the same genotype was the second poorest performer under drought
conditions. Testcrosses amongst the ten best performers had mean grain yields above the trial
mean of 3.94 t ha-1. Genotypes G102 (013WH03) and G103 (013WH01), DR&SS three-way
experimental hybrids used as check varieties, were among the poorest performing genotypes
and recorded high ear rot scores. The trial mean for grain texture was 2.2, indicating that
genotypes were mainly semi-flint in texture. High R2 values were recorded for all traits. Both
genotypes and the environment played a critical role in the expression of plant and ear height
as indicated by high genetic and phenotypic variances.

222
 Table 7.16 Analysis of variance for grain yield for late maturing testcrosses under
well-watered conditions in the 2011 winter season

Source DF MS

Site 1 14.58***

Genotype 107 2.04***

GxE 107 1.53***

Residual 146 0.38

***P≤0.001; DF=degrees of freedom; MS=mean square; Env=environment; G x E=genotype by environment.

7.3.6 Performance of late maturing testcrosses under combined environments

The combined ANOVA showed that testcrosses were significantly different (P≤0.001) for all
traits (Tables 7.19 and 7.20). Sites were significantly different for all traits except anthesis
silking interval. The G x E interaction was highly significant (P≤0.001) for all traits. The
other interactions were also significant. The best performing genotype in terms of grain
yield in the combined analysis was G26 ((DTPWC9-F16-1-1-1-1-BBB/RS61P)-B-
32//CML539/CML442) with a mean yield of 4.24 t ha-1 (Table 7.21). The genotype
performed above the trial mean which was 3.06 t ha-1. The testcrosses G26 ((DTPWC9-F16-
1-1-1-1-BBB/RS61P)-B-32//CML539/CML442) and G29 ((DTPWC9-F104-5-4-1-1-
BBB/SC5522)-B-2//CML539/CML442) were among the ten best performers under drought
and optimum conditions as well as under combined environment analysis. These testcrosses
also exhibited good ear position and root and stem lodging scores. The poorest performing
testcross was G56 ((MAS[206/312]-23-2-1-1-B*7/RA214P)-B-1//CML539/CML442) with a
mean yield of 1.32 t ha-1 with a ears per plant value of less than 0.50 and anthesis silking
interval of 4.5. The R2 values were again very high for all measured traits.

223
Table 7.17 Analysis of variance for anthesis silking interval and other agronomic traits for late maturing testcrosses under
well-watered conditions in the 2011 winter season

Source DF ASI PH EH EPO EPP RL SL ER EA TEX

Site 2 11.96*** 306516.29*** 63852.28*** 0.09*** 41.40*** 12977.14*** 11332.01*** 8172.72*** 951.03*** 778.19***

Genotype 107 3.94*** 422.57*** 439.58*** 0.009*** 0.016*** 74.16*** 71.62*** 85.52*** 4.23 0.18***

GxE 214 2.16*** 229.11*** 198.42*** 0.005*** 0.01** 59.53*** 54.99*** 53.00*** 4.29 0.21***

Residual 219 1.18 87.79 53.76 0.003 0.008 33.67 11.88 32.09 4.02 0.05
***P≤0.001; **P≤0.01; DF=degrees of freedom; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; EPP=ears per plant; RL=root lodging; SL=stem
lodging; ER=ear rot; EA=ear aspect; TEX=texture; Env=environment; G x E=genotype by environment interaction.

224
Table 7.18 Performance of late maturing testcrosses for grain yield and other agronomic traits under well-watered
conditions
Genotype GYD AD ASI PH EH EPO EPP RL SL ER TEX
Ten best testcrosses G12 5.41 70.0 1.2 176.7 96.7 0.56 0.5 4.3 10.5 8.7 2.3
G76 5.38 70.5 0.5 178.3 103.3 0.58 0.57 1.1 3.6 3.9 2.0
G26 5.38 70.0 1.2 175.0 92.5 0.54 0.62 3.1 4.1 5.1 2.0
G29 5.22 70.0 1.2 175.0 82.5 0.47 0.56 9.4 6.8 8.8 2.2
G92 5.14 71.0 1.5 187.5 123.3 0.65 0.54 10.0 4.2 7.2 2.3
G21 5.14 71.0 1.2 187.5 96.7 0.51 0.56 5.5 6.4 4.7 2.0
G22 5.06 71.0 1.3 186.7 102.5 0.56 0.51 1.9 2.9 4.9 2.1
G19 5.03 71.5 0.3 160.0 95.8 0.66 0.53 0.0 4.9 5.4 2.3
G75 4.90 69.0 1.8 168.3 106.7 0.63 0.55 3.3 10.6 6.2 2.3
G38 4.89 69.0 1.3 171.7 92.5 0.54 0.51 4.1 8.4 4.9 2.0

Ten poorest testcrosses G71 2.99 70.5 3.2 179.2 92.5 0.52 0.47 8.2 6.2 9.5 2.7
G11 2.91 74.5 2.5 170.8 96.7 0.57 0.44 8.8 6.6 9.1 2.3
G45 2.91 71.5 0.8 173.3 90.8 0.53 0.52 0.9 1.1 6.2 2.3
G58 2.84 70.0 0.8 179.2 98.3 0.55 0.43 3.0 5.2 13 2.3
G50 2.83 70.5 0.7 170.0 92.5 0.55 0.34 9.5 5.9 2.7 2.5
G47 2.83 71.5 0.8 176.7 109.2 0.62 0.35 4.2 4.2 1.9 2.4
G55 2.75 71.0 4.8 165.8 77.5 0.47 0.47 5.4 6.9 8.9 2.3
G102 2.67 66.5 1.7 180.8 94.2 0.52 0.35 16.6 6.2 33.2 2.6
G103 2.59 72.0 2.0 187.5 100.8 0.54 0.46 7.9 10.1 18.1 2.6
G56 1.88 71.5 3.0 176.7 84.2 0.48 0.38 2.3 3.3 7.3 2.2
Mean 3.94 70.87 1.4 178.4 97.6 0.55 0.51 5.9 6.5 6.9 2.2
LSD 0.72 1.46 1.03 8.93 6.99 0.05 0.08 5.53 3.29 5.4 0.21
Heritability 0.25 0.13 0.45 0.46 0.55 0.38 0.32 0.19 0.23 0.38 0.17
CV% 15.66 1.24 29.12 5.25 7.51 10.08 17.14 38.24 33.26 31.44 9.94
R2 0.92 0.94 0.82 0.98 0.96 0.81 0.98 0.89 0.95 0.87 0.99
-1
GYD=grain yield (t ha ); AD=anthesis days; ASI=anthesis silking interval (days); PH=plant height (cm); EH=ear height (cm); EPO=ear position (0-1); EPP= ears per plant (#);
RL=root lodging (%); SL=stem lodging (%); ER=ear rot (%); TEX=texture (1-5); LSD=least significant difference; CV=coefficient of variation; R2=coefficient of determination.

225
Table 7.19 Analysis of variance for grain yield for late testcrosses under both drought
and well-watered conditions in the 2011 winter season

Source DF MS

Site 1 115.1***

Genotype 107 2.21***

Treatment 1 661.61***

GxE 107 1.25***

ExT 1 260.13***

G xT 107 0.89***

GxExT 107 0.97***

Residual 431 0.13

***P≤0.001; DF=degrees of freedom; MS=mean square; Env=environment; G x E=genotype by environment interaction;
G x T=genotype by treatment interaction; G x E x T=genotype by environment by treatment.

The mean grain yield for late maturing testcrosses under drought conditions was 55% of the
mean yield under well-watered conditions (Figure 7.3). The testcrosses showed good
performance under drought conditions with more than 50% of them performing above the
trial mean (2.71 t ha-1) (Figure 7.4).

7.3.7 Variance components for late maturing testcrosses

The phenotypic variances were higher than the genetic variances for all traits (Table 7.22).
Grain yield, plant height, ear height and ears per plant showed good repeatability values. The
expected genetic gains were higher for plant and ear height and lower for the other traits.
Plant and ear height had repeatability estimates of above 0.50; however grain yield and ears
per plant also had good repeatability values of 0.41 and 0.49 respectively. Expected genetic
gain was higher for plant (11.3) and ear (15.25) height and for grain yield it was 30 g grain
yield per plant, which converted to 3% genetic gain.

7.3.8 Correlation between grain yield and secondary traits under managed drought
for late maturing testcrosses
Grain yield was significantly correlated with all secondary traits under managed drought
conditions (Table 7.23). The correlation with anthesis days and anthesis silking interval was
negative whilst the correlation with ears per plant was positive. Both anthesis days and
anthesis silking interval were significantly but negatively correlated with ears per plant.

226
Table 7.20 Analysis of variance for anthesis days and other agronomic traits for late maturing testcrosses under drought
and well-watered conditions in the 2011 winter season

Source DF AD ASI PH EH EPO EPP RL SL

Site 2 44202.88*** 0.45 264997.19*** 88260.69*** 0.01*** 16.35*** 25733.39*** 86739.47***

Genotype 107 72.11*** 63.00*** 652.57*** 795.96*** 0.01*** 0.04*** 113.08*** 250.90***
Treatment 1 990.50*** 553.56*** 32320.05*** 6219.08*** 0.02*** 4.76*** 677.88*** 181594.35***
GxE 214 61.57*** 61.96*** 214.65*** 208.87*** 0.01*** 0.02*** 74.69*** 174.91***
ExT 2 1532.06*** 30.14 84746.58*** 7868.89*** 0.27** 26.95*** 231.84*** 40979.91***
GxT 107 65.06*** 64.17*** 197.54*** 150.12*** 0.01*** 0.02*** 59.55*** 179.43***
GxExT 214 59.98*** 62.79*** 213.75*** 170.18*** 0.01*** 0.02*** 57.91*** 166.17***
Residual 646 37.26 40.64 50.67 58.24 0.002 0.01 19.96 19.38
***P≤0.001; **P≤0.01; DF=degrees of freedom; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; EPP=ears per plant; RL=root
lodging; SL=stem lodging; Env=environment; G x E=genotype by environment; G x T= genotype by treatment; G x E x T= genotype by environment by treatment.

227
Table 7.21 Performance of late maturing testcrosses for grain yield and other agronomic traits under drought and well-
watered conditions in the 2011 winter season

Genotype GYD AD ASI PH EH EPO EPP RL SL

Ten best testcrosses G26 4.24 73.8 1.3 184.2 97.9 0.54 0.69 5.1 18.1
G29 4.21 73.1 1.8 181.7 89.2 0.49 0.62 7.3 20.9
G92 4.13 76.4 2.1 195.0 122.1 0.63 0.61 10.8 12.3
G76 4.00 75.3 0.5 186.7 106.3 0.57 0.61 5.5 14.5
G38 3.94 73.8 1.2 171.7 90.4 0.53 0.59 3.1 22.6
G75 3.87 73.5 2.3 180.4 110.8 0.61 0.63 8.5 20.3
G51 3.82 74.8 1.3 191.3 106.3 0.56 0.58 4.7 12.9
G28 3.78 74.6 1.4 182.9 105.8 0.58 0.58 6.9 14.1
G12 3.77 73.5 1.5 180.4 94.6 0.53 0.56 4.4 21.2
G21 3.69 73.3 1.8 187.1 95.0 0.51 0.54 6.2 19.6

Ten poorest testcrosses G48 2.36 74.9 4.5 185.8 102.1 0.55 0.52 3.8 17.0
G47 2.34 74.2 1.7 180.8 108.3 0.60 0.46 5.6 23.6
G55 2.31 73.3 2.6 168.8 84.6 0.50 0.56 4.4 13.1
G57 2.25 74.6 1.6 172.9 84.2 0.49 0.51 4.2 20.9
G85 2.21 76.0 2.4 183.8 112.5 0.61 0.51 7.8 22.4
G58 2.10 74.4 2.0 183.8 93.8 0.51 0.42 4.1 22.8
G11 2.05 75.9 3.3 180.4 98.3 0.55 0.46 9.5 15.1
G50 2.00 73.8 1.8 170.4 89.2 0.52 0.41 10.8 18.9
G103 1.93 74.9 3.8 187.1 103.3 0.55 0.39 11.3 19.7
G56 1.32 74.9 4.5 179.2 82.5 0.46 0.47 1.7 21.5
Mean 3.06 74.1 2.0 183.4 99.8 0.55 0.56 6.6 18.3
LSD 0.29 4.10 4.20 4.70 5.10 0.03 0.05 3.00 2.96
CV% 11.53 8.24 31.4 3.88 7.64 8.66 15.42 37.39 24.04
2
R 0.97 0.86 0.67 0.96 0.92 0.80 0.96 0.91 0.98
GYD=grain yield (t ha-1); AD=anthesis days; ASI=anthesis silking interval (days); PH=plant height (cm); EH=ear height (cm); EPO=ear position (0-1); EPP=ears per plant (#); RL=root
lodging (%); SL=stem lodging (%); LSD=least significant difference; CV=coefficient of variation; R2=coefficient of determination.

228
 4.5
4
3.5
Mean grain yield t ha-1

3
2.5
2
1.5
1
0.5
0
Optimum Drought Combined
Treatment

Figure 7.3 Mean grain yield for late maturing testcrosses under optimum, drought
and combined environments in the 2011 winter season.

60

50
Number of testcrosses

40

30

20

10

0
1.1-1.5 1.6-2.0 2.1-2.5 2.6-3.0 3.1-3.5 Mean (2.17)
Yield (t )
ha-1

Figure 7.4 Mean grain yield of late maturing testcrosses under drought conditions
across three environments in the 2011 winter season.

229
Table 7.22 Genetic and phenotypic variances, repeatability and genetic gain for grain
yield and other agronomic traits measured in late maturing testcrosses

Trait σ2g σ2p b ∆g

GYD 0.13 0.32 0.41 0.03

AD 0.68 9.01 0.08 0.14
ASI 0.04 8.48 0.0047 0.80
PH 56.77 85.63 0.66 11.30
EH 75.89 104.43 0.73 15.25
EPP 0.0025 0.0051 0.49 0. 10
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPP=ears per plant;
σ2g=genetic variance; σ2p=phenotypic variance; b=repeatability; ∆g=genetic gain.

Table 7.23 Correlation of grain yield and secondary traits for late maturing testcrosses
under managed drought

GYD AD ASI

AD -0.274**

ASI -0.360** 0.254**

EPP 0.616** -0.234* -0.341**

**P≤0.01; *P≤0.05; GYD =grain yield; AD=anthesis days; ASI=anthesis silking interval; EPP=ears per plant.

7.4 Discussion
It is critical to testcross segregating lines in a breeding programme so that inferior lines can
be discarded in early stages of breeding and this saves time and resources. Results indicated
that environments, genotypes and G x E interaction effects were highly significant for the
given growing conditions. Highly significant differences observed among testcrosses in this
study were an indication of genetic variability available that can be exploited for further
improvement. However, breeders need the information on the magnitude of genetic variation
that exists for a given trait in a set of germplasm to justify selection for that trait (Nachit et
al., 1992). This can be estimated by partitioning of the total variation (total sum of squares)
into its various components from the ANOVA and a higher magnitude implies a greater
genetic potential for improvement for the trait. The mean grain yield reported in combined
analysis for early maturing testcrosses was 2.98 t ha-1 and 3.06 t ha-1 for late maturing
testcrosses. The yield reductions under drought conditions were 40% for early maturing

230
testcrosses and 55% for late maturing testcrosses and the stress could be classified as severe.
The yield reduction in early maturing testcrosses is in agreement with results reported by
Azeez et al. (2005). The reported yield reduction of 55% for late maturing testcrosses was
within the value range reported by Campos et al. (2006) of 45-60% yield losses when
drought occurred at silk emergence. However, Banziger et al. (2000) reported yield
reductions of 15-20% under moderate drought stress. According to Azeez et al. (2005) yield
reductions reported in the current study could be due to reduced ears per plant, reduction in
individual kernel weight and reduced translocation of the photosynthates to the grain.
Reduction in other yield determining traits significantly affected grain yield of the testcrosses
such as high ear rot percentages and poor ear aspect scores though high maize yield under
drought stress has been reported to be associated with a high number of fertile ears per plant
(Bolanos and Edmeades, 1996; Bolanos et al., 1993).

The presence of significant G x E interactions for most traits recorded in this study is an
indication that the testcrosses did not have consistent performance in different environments.
Results are contrary to findings by Grauffret et al. (2000) and Menkir et al. (2007).
Morphological, phenological and physiological traits of varieties contribute to G x E
interactions (Nachit et al., 1992). The presence of significant G x E interaction under drought
and non-drought environments underlines the importance of conducting multi-location trials
in representative environments to identify yield stable hybrids under both conditions. The
variation in performance of genotypes from environment to environment, especially changes
in rank, hinder identification of superior stable hybrids (Hyrkas and Carena, 2005). In any
plant breeding programme selection of genotypes without crossover G x E interaction and
identification of sites with similar or different characteristics is of principal importance
(Setimela et al., 2007). Significant G x E interaction under drought stress impedes breeding
progress (Ribaut et al., 2009). It therefore appears that G x E effects would present
challenges in breeding for drought tolerance in the given germplasm.

Testcrosses constituted from inbred lines K64r, RS61P and N3.2.3.3 with DTPWC9 based
donors generally showed superior performance over other testcrosses. The testcross
(DTPWC9-F104-5-4-1-1-BBB/K64r)-B-33//CML444/CML395 was the overall best

231
performing testcross in the early maturing category whilst (DTPWC9-F16-1-1-1-1-
BBB/RS61P)-B-32//CML539/CML442 was the overall best performer in the late maturing
category. The testcrosses ranked the best mainly because of the good performance of the
inbred lines K64r and RS61P reported earlier in the study. The best testcrosses were
generally consistent in their performance as they were also among the best performing
genotypes under drought and optimum conditions. Of the nine drought tolerant donors used,
the DTPWC9 based donors showed superior performance, especially against G16BNSeqC4
and LaPostaSeqC7 based donors. The identified superior lines can therefore be selected for
further generation advancement for the development of homozygous lines. Two main goals
of any maize population improvement programme include improving the mean of a
quantitative trait by concentrating favourable alleles and maintaining genetic variability
(Hyrkas and Carena, 2005). The superior individual lines identified after crossing with a
tester can be inbred lines for potential use as parents of synthetic or hybrid cultivars (Fehr,
1987).

Correlation coefficient analysis has been used in selection of secondary traits affecting yield
(Menkir, 2008). Grain yield was significantly correlated with the secondary traits anthesis
days, anthesis silking interval and ears per plant in this study. Ears per plant showed good
heritability estimates compared to anthesis days and anthesis silking interval that showed
very low heritability estimates. Indirect selection for secondary traits correlated with grain
yield rather than grain yield alone has been shown to increase selection efficiency by about
20% in maize grown under low soil nitrogen stress (Banziger and Lafitte, 1997). Results are
consistent with findings by other researchers (Bolanos and Edmeades, 1996; Magorokosho et
al., 2003; Zaidi et al., 2004; Monneveux and Ribaut, 2006; Ribaut et al., 2009). Therefore
the exploitation of highly heritable components, which are highly correlated to grain yield, is
therefore a more effective option than direct selection of yield per se (Kashiani and Saleh,
2010). In this study anthesis days was negatively correlated to grain yield for both early and
late maturing testcrosses and similar results were reported by Magorokosho et al. (2003).
This may have been due to high temperatures experienced during the growing season at the
Lowveld research stations, which subjected testcrosses to drought stress during grain filling.

232
The genetic variances and repeatability estimates in this study were generally low. The
genetic variance for grain yield recorded for early maturing testcrosses was 0.03 and for late
maturing testcrosses was 0.13 and the repeatability estimates were 0.14 and 0.41
respectively. Results are in agreement with findings by other researchers that genetic
variance and repeatability of grain yield often decline with increasing moisture stress
(Bolanos and Edmeades, 1996; Magorokosho et al., 2003). Correlation of ears per plant with
grain yield under drought stress was high, but because of the small variances, selection for
this trait may not be effective. Similar results were reported by Magorokosho et al. (2003);
however, these findings are contrary to findings by Edmeades et al. (1995) and Bolanos and
Edmeades (1996). Plant and ear height had higher genetic variances and consequently higher
repeatability estimates in this study. The genetic gains for these traits were also higher
compared to other traits. Falconer and Mackay (1996) stated that genetic gain of selection for
given traits depend on the heritability estimates. Traits with high repeatability can be selected
on individual plant basis (El-Badawy, 2011), whilst on the other hand single plant selection
would be inefficient for low repeatability traits and a type of family selection would be
required. Good repeatability values were recorded for grain yield, plant height, ear height and
ears per plant for the late maturing testcrosses in the study. Large heritability for late hybrids
support direct selection for yield whilst the reverse is true for early testcrosses. El-Badawy
(2011) postulated that high repeatability values indicate the possibility of predicting the real
individual value with a relatively small number of measurements. Therefore the knowledge
of the repeatability coefficient allows an efficient use of resources and time in the evaluation
phase.

7.5 Conclusions
Testcross performance evaluation is critical in that it enables the breeder to screen and select
superior lines even at the early stages of recurrent selection. Significant differences amongst
the genotypes in both sets of trials for the majority of traits are an indication that there is
existence of variability that can be exploited during selection. However, existence of highly
significant G x E interaction is evidence that the performance of testcrosses was not
consistent across the production environments therefore complicating the selection process.
Generally the best performing testcrosses were identified to be mostly constituted of DR&SS

233
inbred lines K64r, RS61P and N3.2.3.3 with the CIMMYT drought tolerant donors based on
DTPWC9. The best performing testcrosses included testcrosses from both the CIMMYT
group A (CML539/CML442) and group B (CML444/CML395) testers. The information can
now be used in selecting segregating lines for further generation advancement. Highly
significant correlation of grain yield with important secondary traits such as anthesis days
and anthesis silking interval under drought conditions was recorded in the study. This
therefore means that these traits can be exploited in the selection process instead of just
directly selecting for yield per se. High and positive correlations of grain yield with ears per
plant under drought conditions were recorded for both sets of trials, however because of
small genetic variances, selecting for the trait may not be effective. On the contrary, ears per
plant showed good repeatability, which means that if the trait is selected for there are higher
chances that similar results may be realised again. The genetic variances and repeatability
estimates for grain yield were low compared to traits such as plant and ear height, which
implies that a slow genetic progress will be realised in terms of selecting for the trait.
Repeatability for grain yield for the late maturing testcrosses under drought conditions was
good, which therefore means that there are higher chances that the performance will be
consistent over years compared to the early maturing testcrosses and the finding supports
direct selection.

7.6 References
Agronomix Software, Inc. 2005. AGROBASE Generation II User’s Manual. Version II,
Revised Edition. www.agronomix.com. Agronomix Software, Winnipeg, M.B.,
Canada.
Allard, R.W. 1960. Hybrid varieties: Principles of Plant Breeding. John Wiley and Sons, Inc.
USA.
Azeez, J.O., D. Chikoye, A.Y. Kamara, A. Menkir and M.T. Adetunji. 2005. Effect of
drought and weed management on maize genotypes and the tensiometroc soil water
content of an eutric nitisol in south western Nigeria. Plant and Soil 276: 61-68.
Banziger, M. and H.R. Latfitte. 1997. Efficiency of secondary traits for improving maize for
low-nitrogen target environments. Crop Science 37: 1110-1117.

234
Banziger, M., G.O. Edmeades, D. Beck and M. Bellon. 2000. Breeding for drought and
nitrogen stress tolerance in maize. From theory to practise. Mexico, D.F: CIMMYT.
pp 25-30.
Becelaere, G.V., E.L. Lubbers, A.H. Paterson and P.W. Chee. 2005. Pedigree vs DNA
marker-based genetic similarity estimates in cotton. Crop Science 45: 2281-2287.
Bolanos, J. and G.O. Edmeades. 1996. The importance of the anthesis-silking interval in
breeding for drought tolerance in tropical maize. Field Crops Research 48: 65-80.
Bolanos, J., G.O. Edmeades and L. Martinez. 1993. Eight cycles of selection for drought
tolerance in lowland tropical maize. III. Responses in drought adaptive physiological
and morphological traits. Field Crops Research 31: 269-286.
Campos, H., M. Cooper, G.O. Edmeades, C. Loffler, J.R. Schussler and M. Ibanez. 2006.
Changes in drought tolerance in maize associated with fifty years of breeding for
yield in the U.S. Corn Belt. Maydica 51: 369-381.
Cox, T.S., Y.T. Kiang, M.B. Gorman and D.M. Rogers. 1985. Relationship between the
coefficient of parentage and genetic similarity in soybean. Crop Science 25: 529-
532.
Dreisigacker, S., P. Zhang, M.L. Warbuton, M. Van Ginkel, D. Hoisington, M. Bohn and
A.E. Melchinger. 2004. SSR and pedigree analyses of genetic diversity among
CIMMYT wheat lines targeted to different mega-environments. Crop Science 44:
381-388.
Edmeades, G.O., S.C. Chapman, J. Bolanos, M. Banziger and H.R. Lafitte. 1995. Recent
evaluation of progress in selection for drought tolerance in tropical maize.
Proceedings of the 4th Eastern and Southern African Regional Maize Conference,
Harare, Zimbabwe. pp 94-100.
El-Badawy, M.EI.M. 2011. Estimation of genetic variance and its components in new
synthetic “Moshtohor2” of white maize. Journal of Applied Sciences Research 7:
2489-2494.
Falconer, D.S. 1960. Introduction to Quantitative Genetics. 1st Edition. The Ronald Press
Company. New York.
Falconer, D.S. and T.F.C. Mackay. 1996. Introduction to Quantitative Genetics. 4th edition.
Longman. Essex. England.

235
FAOSTAT. 2008. Available at http://faostat.fao.org. Accessed on line 25 March 2009.
Fehr, W.R. 1987. Principles of Cultivar Development. Vol 1. Theory and Technique.
Macmillian Publishing Company. New York.
Grauffret, C., J. Lothrop, D. Dorvillez, B. Gouesnaid and M. Derieux. 2000. Genotype x
environment interaction in maize hybrids from temperate or highland tropical origin.
Crop Science 40: 1004-1012.
Hallauer, A.R. 1980. Relation of quantitative genetics to applied maize breeding. Brazilian
Journal of Genetics 3: 207-233.
Hyrkas, A. and M.J. Carena. 2005. Response to long-term selection in early maturing maize
synthetic varieties. Euphytica 143: 43-49.
Jenkins, M.T. 1935. The effect of inbreeding and of selection within inbred lines of maize
upon the hybrids made after successive generations of selfing. Iowa State College
Journal of Science 3: 677-721.
Kashiani, P. and G. Saleh. 2010. Estimation of genetic correlations on sweet corn in inbred
lines using SAS Mixed Model. American Journal of Agricultural and Biological
Sciences 5: 309-314.
Magorokosho, C., K.V. Pixley and P. Tongoona. 2003. Selection for drought tolerance in two
tropical maize populations. African Crop Science 11: 151-161.
Menkir, A. 2008. Genetic variation for grain mineral content in tropical adapted maize inbred
lines. Journal of Food Chemistry 110: 454-464.
Menkir, A., I. Ingelbrecht and C. The. 2007. Testcross performance and diversity analysis of
white maize lines derived from backcrosses containing exotix germplasm. Euphytica
155: 417-428.
Messmer, M.M., A.E. Melchinger, R.G. Herrmann and J. Boppenmaier. 1993. Relationships
among early European maize inbreds: II. Comparison of pedigree and RFLP data.
Crop Science 33: 944-950.
Mihaljevic, R., C.C. Schon, H.F. Utz and A.E. Melchinger. 2005. Correlations and QTL
correspondence between line per se and testcross performance for agronomic traits
in four populations of European maize. Crop Science 45: 114-122.

236
Monneveux, P. and J.M. Ribaut. 2006. Secondary traits for drought in cereals. In: Ribaut,
J.M. (ed) Drought Adaptation in Cereals. The Haworth Press, Inc., Binghamton. pp
97-143.
Nachit, M.N., M.E. Sorrels, R.W. Zobel, H.G. Gauch, R.A. Fischer and W.R. Coffman.
1992. Association of environmental variables with sites’ mean grain yield and
components of genotype-environment interaction in durum wheat. Journal of
Genetics and Breeding 46: 369-372.
Pixley, K.V., T. Dhliwayo and P. Tongoona. 2006. Improvement of maize populations by
full-sib selection alone versus full-sib selection with selection during inbreeding.
Crop Science 46: 1130-1136.
Ribaut, J.M., J. Betran, P. Monneveux and T. Setter. 2009. Drought tolerance in maize. In:
Bennetzon, J.L. and S.C. Hale (eds) Handbook of Maize: Its biology. Springer NY.
pp 311-344.
Rukuni, M., P. Tawonezvi and C. Eicher. 2006. Zimbabwe’s Agricultural Revolution
Revisited. University of Zimbabwe Publications. Harare. pp 119-140.
Setimela, P.S., B. Vivek, M. Banziger, J. Crossa and F. Maideni. 2007. Evaluation of early to
medium maturing open pollinated maize varieties in SADC region using GGE biplot
based on the SREG model. Field Crops Research 103: 161-169.
SPSS 15.0 for Windows. 2006. www.spss.com SPSS Inc., Chicago III.
Wricke, G. and W.E. Weber. 1986. Quantitative Genetics and Selection in Plant Breeding.
Walter de Gruyter. New York. pp 406.
Zaidi, P.H., G. Srinivasan, H.S. Cordova and C. Sanchez. 2004. Gains from improvement for
mid-season drought tolerance in tropical maize (Zea mays L.). Field Crops Research
89: 135-152.

237
 CHAPTER 8

Performance and yield prediction of three-way hybrids from drought

tolerant single cross hybrids

Abstract
Drought is one of the most limiting factors in maize production in Zimbabwe and the use of
poorly adapted varieties has been of great concern. The objectives of this study were to
evaluate the agronomic performance of three-way hybrids predicted from drought tolerant
single cross hybrids and to investigate the correlation between the predicted and observed
means. Eleven single cross hybrids were crossed to different males using a North Carolina
Design I. A total of 77 three-way hybrids were successfully produced. Hybrids were then
evaluated across three sites under optimum and managed drought conditions in the 2011
winter season. Results revealed varied performance of hybrids for grain yield and other yield
related traits. Grain yield was significantly (P≤0.01) and negatively correlated with anthesis
days (r=-40) under drought conditions and there was a tendency that the earlier maturing
hybrids produced higher yields compared to the later maturing hybrids. The correlation
between grain yield and ears per plant was significant (P≤0.01) and higher under drought
conditions (r=0.76) compared to optimum conditions. The study further confirmed the utility
of anthesis silking interval as indirect selection criteria for grain yield under drought
conditions. However, the utility of ears per plant was not conclusive because of low genetic
variance and negligible broad sense heritability estimates. A significant (P≤0.05) but weak
correlation (r=0.27) was obtained between the predicted and the observed grain yield means
and this could be explained by the epistatic and significant G x E interaction effects, which
were not taken into account in the prediction equation used. Results show that the three-way
hybrids with superior predicted yields can be evaluated in multi-location trials and superior
hybrids identified and released for commercial use. Three hybrids,
RA214P/CML538//RS61P, RS61P/CML444//CML538 and RS61P/CML444//CML539 were
identified as having superior performance. However, there is still need for further evaluation
of hybrids in multi-location advanced variety trials before considering them for release.
There is also a need to evaluate the hybrids under low nitrogen conditions as this was not
done in the current study.

238
8.1 Introduction

Maize ranks first in terms of the number of hectares grown and total cereal production in
Zimbabwe. It is the staple food and an important cash crop. Zimbabwe maize breeding has
been a success story over the years with hybrids being developed for both small scale and
commercial production. The maize breeding programme evolved through four phases,
namely, OPV, double cross hybrid, single cross hybrid, and the three-way hybrid
development phases. The three-way hybrid phase started in the late 1970s and the first three-
way hybrids, R201 and R215, were registered in 1988. These two hybrids became very
popular with farmers but their weakness was that they were not bred for drought tolerance as
a result they would succumb to drought.

The main maize production constraint in the country has been the use of poorly adapted
varieties since most of the previous maize breeding work was focusing on high input
environments. According to CIMMYT-Zimbabwe (2000) improved yields, variety yield
stability, pest and disease resistance, tolerance to drought and low soil fertility, generally
produce yield improvements of 30-50%. Breeding for abiotic stress tolerance (drought and
low N) is now being incorporated into the inbred line and variety development process. The
expected genetic variance and predicted yield potential decline from single, three-way,
double to top crosses (Cockerham, 1961). Performance evaluation trials performed by
Weatherspoon (1970) showed that single crosses showed higher grain yields followed by
three-way crosses, while the double crosses showed lower grain yields. Single crosses result
in maximum hybrid vigour, but low seed yields of inbred lines make the cost of single cross
hybrid seed prohibitive. Single crosses are more sensitive or responsive to environmental
conditions, whilst stable high average yield is important for producer consistency in
performance across years and locations. According to Allard and Bradshaw (1964) the
uniformity of single crosses causes a lack of population buffering as they possess only
individual buffering, whilst three-way crosses have both population and individual buffering.

Maize breeders have continued to develop a large number of inbred lines to facilitate
efficient hybrid variety development and in recent years this has been expedited by doubled
haploid technology (Seitz, 2005). Predicting the performance of hybrids from per se

239
performance of their parental inbred lines has not been effective due to masking dominance
effects (Smith, 1986; Hallauer, 1990). With many inbred lines it is often impractical to test
and compare all possible three-way cross hybrids. In commercial maize breeding
programmes, identification of single cross hybrids with superior yield performance is of
fundamental importance. Many methods of prediction have been proposed and some are
currently in use (Eberhart, 1964; Eberhart and Gardner, 1966; Hinkelmann, 1968).

Predicting the performance of all hybrid combinations between a number of inbred lines is a
practical problem because the number of these combinations usually exceeds the practical
limits of field evaluation. Questions arise, however, as to the best methods of prediction and
the accuracy of predictions. In various studies Jenkins’ method B (Jenkins, 1934), which
employs the mean of the non-parental single crosses, proved more suitable and is
consequently used in hybrid maize breeding for prediction of three-way and double crosses
(Melchinger et al., 1987). The single crosses used for prediction are anticipated to have
originated from a comprehensive factorial mating design. The number of single crosses is
considerably fewer and it is logical that they can be used to estimate or predict the
performance of three-way cross hybrids. It is common to predict performance of three-way
hybrids from single cross test results (Melchinger et al., 1987). However, results obtained
using the different methods did not always agree sufficiently with the observed values. In an
earlier study (Chapter 3) superior single cross hybrids developed using a NCDII were
identified and they are the ones used for predicting three-way hybrid performance in the
current study. Therefore the objectives of this study were to evaluate the agronomic
performance of three-way hybrids predicted from superior drought tolerant single cross
hybrids and to determine the relationship between the observed and estimated means.

8.2 Materials and methods

8.2.1 Germplasm

In the previous study (Chapter 3) 11 single cross hybrids were identified in terms of good
SCA and 12 inbred lines were identified in terms of good GCA effects (Table 8.1). The

240
three-way hybrids were then constituted in the field using North Carolina Design I. A total of
77 three-way hybrids were successfully produced.

Table 8.1 Germplasm used in constituting the three-way hybrids

Single cross hybrids Code Inbred lines Code

K64r/CML444 SC1 NAW5885 L1

K64r/CML537 SC2 CML442 L2
K64r/CML539 SC3 RA214P L3
N3.2.3.3/CML545 SC4 N3233 L4
N3.2.3.3/CML444 SC5 CML539 L5
NAW5885/CML442 SC6 CML538 L6
NAW5885/CZL03007 SC7 CML548 L7
RA214P/CML538 SC8 CML545 L8
RA214P/CML548 SC9 CML444 L9
RS61P/CML444 SC10 RS61P L10
RS61P/CZL03007 SC11 CZL03007 L11
K64r L12

8.2.2 Evaluation sites

The three-way hybrids were evaluated under well-watered and managed drought conditions
across three sites namely Chiredzi, Chisumbanje and Save Valley Research Stations. The
sites are as described in section 7.2.2.

8.2.3 Trial management

A 0.1 alpha lattice design was used in planting the trials. Trials were planted in 4.0 m one
row plots with an inter-row spacing of 0.75 m and in-row spacing of 0.25 m. Two seeds were
planted per station and later thinned to one plant per station to achieve a plant population of
53 000 plants ha-1. Compound D basal dressing of 400 kg ha-1 and ammonium nitrate
topdressing of 350 kg ha-1 were applied to the crop. The topdressing was split applied at four
and eight weeks after crop emergence. Dipterex 2.5% granules were applied for stalkborer
control. Weed control was mainly done through hand weeding.

241
8.2.4 Management of drought site
The crop was raised offseason (winter) through the use of irrigation. Irrigation was applied in
such a way that drought at flowering was severe enough to delay silking and cause ear
abortion. Ideally anthesis silking interval should average about four to eight days and ears per
plant 0.3 to 0.7, whilst grain yield should be around 1-2 t ha-1 (15-20% of well watered
yields) for a drought managed site.

8.2.5 Data collection

Data was collected for grain yield (t ha-1), days to 50% anthesis, days to 50% silking, plant
and ear height (cm), root and stem lodging, ear rot, senescence, ear aspect and texture.
Derived traits such as ear position, ears per plant and anthesis silking interval were
calculated. Traits were measured and derived as described in section 3.2.4 (Table 3.2).

8.2.6 Statistical analysis

The data from individual sites was subjected to ANOVA and later the analysis was combined
over locations using AGROBASE (2005). Broad-sense heritability for traits was estimated
using the ANOVA components of variance procedure as suggested by Becker (1984) and as
described in sections 3.2.5 and 5.2.6.

The three-way hybrid prediction from the single crosses was done according to Fehr (1993).
TWC12.3 = ½ [SC13 + SC23)
Where: TWC12.3 = predicted three-way cross mean
SC13 = single cross mean between inbred line 1 and inbred line 3
SC23 = single cross mean between inbred line 2 and inbred line 3
Pearson correlation coefficients (r) amongst the different traits were calculated from means
across environments. The statistical computations were performed with SPSS 15.0 version
for Windows (2006).

242
8.3 Results

8.3.1 Performance of hybrids under managed drought conditions

Results presented are for the 10 best and 10 poorest hybrids in terms of grain yield and the
means presented in the tables are for all entries. Results for all entries under optimum,
drought and combined environments are presented in Appendices 13-15. Sites were
significantly different (P≤0.001) for anthesis days, anthesis silking interval, plant and ear
height, root and stem lodging and senescence (Table 8.2).

Table 8.2 Analysis of variance for grain yield and other agronomic traits under
managed drought conditions in the 2011 winter season
Source DF GYD AD ASI PH EH EPP RL SL SEN

Site 2 6.32** 48134.59*** 170.096*** 27497.19*** 14735.64*** 0.166* 6438.63*** 6879.46*** 5.211***

Genotypes 84 1.21*** 18.89*** 17.82*** 480.80** 482.96*** 0.051ns 156.70*** 390.25*** 0.010***

GxE 168 4.68* 2.84*** 9.078ns 244.58** 142.38*** 0.044ns 96.93* 189.65*** 0.006***

Residual 156 0.182 0.085 7.133 115.34 39.62 0.041 75.25 58.92 0.001
***P≤0.001; **P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear
height; EPP=ears per plant; RL=root lodging; SL=stem lodging; SEN=senescence; Env=environment; G x E=genotype x
environment interaction; DF=degrees of freedom.

Genotypes were significantly different for all traits except ears per plant. The G x E
interaction was significant for grain yield and root lodging (P≤0.05), plant height (P≤0.01)
and for anthesis days, ear height, stem lodging and senescence (P≤0.001). The minimum
grain yield obtained under managed drought conditions was 0.68 t ha-1 from genotype
013WH63 and the maximum yield was 2.89 t ha-1 from two genotypes SC7//L5
and SC8//L10 (Table 8.3). The ten best performing hybrids had mean yields above the trial
mean. Hybrids generally showed good anthesis silking interval with a trial mean of 2.6 days
and this is desirable under drought conditions. The root and stem lodging trial means were
also indicative of good stand ability of hybrids. The number of ears per plant was influenced
by a high level of barrenness amongst hybrids with a trial mean of 0.59. The poorest
performing hybrids generally had ear rot means that were above the trial mean and hence
they had poor ear aspect scores. The Pearson’s coefficient of correlation showed that there
was significant (P≤0.01) correlation between grain yield and other traits such as anthesis
days, anthesis silking interval, plant and ear height, ear position and ears per plant under
managed drought conditions (Table 8.4).

243
Table 8.3 Mean performance for grain yield and other agronomic traits under managed drought in the 2011 winter season
Entry GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA
Ten Best hybrids SC7//L5 2.89 63.3 2.8 178.3 78.0 0.43 10.6 19.8 0.76 5.4 0.5 1.5 2.4
SC8//L10 2.89 63.9 1.2 174.7 101.0 0.53 11.1 5.1 0.65 16.0 0.5 2.7 3.3
SC3//L2 2.88 66.4 1.6 167.7 82.7 0.45 15.8 22.0 0.65 30.8 0.6 3.5 3.6
SC2//L5 2.87 64.9 1.5 170.5 77.4 0.46 11.7 13.1 0.75 6.3 0.6 1.0 2.7
SC10//L6 2.86 66.4 1.0 177.1 88.9 0.52 -1.5 12.1 0.69 16.3 0.5 2.7 3.0
SC3//L1 2.63 65.2 5.4 176.3 81.3 0.41 8.0 12.2 0.56 21.9 0.5 3.0 3.0
SC2//L9 2.61 67.9 1.6 180.3 96.7 0.51 11.7 10.8 0.69 26.0 0.6 3.0 3.8
SC10//L5 2.59 63.7 0.8 178.4 90.4 0.47 10.4 29.4 0.76 4.9 0.6 3.0 3.2
CZH0616 2.57 64.7 4.0 174.8 85.8 0.50 6.3 9.9 0.69 11.4 0.5 3.0 2.9
023WH31 2.54 69.4 1.8 185.9 100.9 0.55 11.8 11.4 0.67 23.3 0.5 2.8 3.2

Ten Poorest hybrids SC4//L9 1.27 67.6 3.0 165.9 108.2 0.60 1.7 8.6 0.42 40.6 0.5 2.9 4.2
SC7//L3 1.26 65.9 4.7 184.5 95.1 0.51 8.5 1.6 0.42 18.4 0.5 2.9 3.7
SC8//L11 1.22 67.6 3.6 179.6 102.3 0.54 6.4 14.6 0.36 31.1 0.6 2.7 3.5
SC6//L5 1.19 68.2 2.9 181.4 102.5 0.55 7.9 12.3 0.49 21.5 0.5 3.3 4.2
SC7//L4 1.18 65.7 5.9 195.1 103.0 0.52 9.8 5.6 0.39 34.1 0.5 2.5 3.5
SC5//L2 1.15 69.6 6.3 187.5 98.1 0.51 8.7 9.0 0.41 27.0 0.5 4.1 4.1
SC5//L11 0.83 68.9 2.6 168.5 117.7 0.65 0.2 7.3 0.37 26.8 0.6 3.3 4.1
SC10//L4 0.83 68.1 7.4 190.5 107.3 0.56 6.8 11.9 0.26 48.0 0.5 3.4 4.2
SC11//L4 0.80 66.1 3.4 178.6 102.2 0.51 22.3 16.8 0.51 22.9 0.5 3.0 3.3
013WH63 0.68 66.9 5.1 190.2 97.0 0.53 17.2 6.7 0.40 40.6 0.6 3.9 4.3
Mean 1.95 66.0 2.6 175.5 94.5 0.52 13.1 13.7 0.59 26.6 0.6 3.0 3.5
LSD
1.03 2.0 2.4 17.8 13.1 0.10 20.3 13.0 0.16 46.7 0.1 0.9 0.9
(0.05)
MSE 0.53 2.1 4.4 80.5 130.6 0.00 104.5 128.9 0.02 1651.1 0.0 0.2 0.4
Min 0.68 61.8 -1.1 133.1 66.5 0.37 -2.3 -1.1 0.26 2.3 0.5 1.0 2.4
Max 2.89 70.6 7.4 198.5 117.7 0.70 50.7 33.6 0.83 180.9 0.7 4.1 4.6
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem lodging; EPP=ears per plant;
ER=ear rot; SEN=senescence; TEX=texture; EA=ear aspect; LSD=least significant difference; MSE=Mean square error; Min=minimum; Max=maximum.

244
Table 8.4 Pearson’s coefficient of correlation between grain yield and other agronomic
traits under managed drought conditions

GYD AD ASI PH EH EPO RL SL EPP ER

AD -0.40**
ASI -0.44** 0.30**
PH -0.288** 0.25* 0.32**
EH -0.45** 0.43** 0.28* 0.27*
EPO -0.33** 0.36** 0.19 -0.10 0.78**
RL 0.10 -0.40** -0.18 0.05 -0.05 -0.11
SL 0.17 -0.27* -0.32** -0.42** -0.40** -0.35** 0.05
EPP 0.76** -0.60** -0.65** -0.36** -0.52** -0.38** 0.24* 0.31**
ER -0.18 -0.13 0.03 0.02 0.10 0.13 -0.05 -0.02 -0.30**
SEN -0.03 -0.24* -0.37** -0.17 -0.08 -0.01 0.27* 0.35** 0.11 0.16
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear
position; RL=root lodging; SL=stem lodging; EPP=ears per plant; ER=ear rot; SEN=senescence.

The correlation was, however, negative with all these traits except ears per plant. A highly
significant positive (0.76) correlation of grain yield with ears per plant was realised. Ear
position was significantly and positively correlated (0.78) with ear height. There was
negative and significant correlation of ears per plant with ear rot. The genotypic variances
were lower than the phenotypic variances under managed drought conditions and hence the
broad sense heritability estimates were also low. Grain yield, stem lodging, ear height and
anthesis days had broad sense heritability estimates were above 0.50 (Table 8.5). Ears per
plant had low variances compared to all the other traits and the broad sense heritability
estimate was negligible.

245
Table 8.5 Genotypic and phenotypic variances and broad sense heritability estimates
for measured traits under managed drought conditions

Trait σ2g σ2 p h2 B
AD 0.1573 0.1852 0.85
ASI 0.0857 0.175 0.49
PH 2.3159 4.714 0.49
EH 3.339 4.7349 0.71
EPP 0.0001 0.0005 0.14
RL 0.5859 1.5362 0.38
SL 1.9667 3.826 0.51
SEN 0.000041 0.0001 0.41
GYD 0.234 0.456 0.51
σ2g=genetic variance; σ2p=phenotypic variance; h2B=broad sense heritability; AD=anthesis days; ASI=anthesis silking interval;
PH=plant height; EH=ear height; EPP=ears per plant; RL=root lodging; SL=stem lodging; SEN=senescence; GYD=grain yield.

8.3.2 Performance of three-way hybrids under optimum conditions

Environments were again significantly different for all traits (P≤0.001 and P≤0.01) under
optimum conditions (Table 8.6). Genotypes were significantly different (P≤0.001) for grain
yield, anthesis silking interval, plant and ear height, ear position, ear aspect, ear rot, root and
stem lodging and texture and significantly at P≤0.01 for anthesis days. The G x E interaction
was also significant for all traits. The best performing genotype was SC9//L10 with a mean
yield of 4.79 t ha-1 whilst the poorest performing hybrid was 013WH01 with a mean yield of
1.95 t ha-1 (Table 8.7).

The overall trial mean was 3.51 t ha-1 and all ten best performing genotypes performed above
the mean. Genotypes generally outperformed the check varieties. The ear position values
showed that genotypes had good ear placement and the mean was 0.55. The stand ability
properties of genotypes were good with root and stem lodging means of 5.4 and 16.6
respectively. Prolificacy (ears per plant) also appeared to be good with a trial mean of 0.78
with SC9//L9 having a mean of 1.25 ears per plant. The maturity of hybrids was more
inclined toward medium maturity with a trial mean of 71.8 anthesis days. The Pearson
correlation coefficient among traits showed a positive and significant correlation between
grain yield and ear height, ear position and ears per plant (Table 8.8).

246
Table 8.6 ANOVA for grain yield and other agronomic traits under optimum conditions in the 2011 winter season
Source DF GYD AD ASI PH EH EPO EPP EA ER RL SL TEX
Site 2 108.59*** 303.22** 117.531*** 315379.460*** 350.66** 0.141*** 0.291*** 24.095*** 286.024*** 18209.290*** 19156.42*** 40.687***
Genotypes 84 164.52*** 50.56** 30.902*** 500.838*** 363.80*** 0.009*** 0.039** 0.527*** 147.860*** 185.640*** 121.95*** 0.895***
GxE 168 16.56** 21.78*** 26.264*** 171.971*** 140.59** 0.004*** 0.036** 0.356*** 81.910*** 162.603*** 120.73*** 0.576***
Residual 156 0.035 10.65 15.702 40.375 2.70 0.001 0.025 0.131 24.980 24.462 56.86 0.150
***P≤0.001; **P≤0.01; DF=degrees of freedom; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; EPP=ears per
plant; EA=ear aspect; ER=ear rot; RL=root lodging; SL=stem lodging; TEX=texture; Env=environment; G x E=genotype x environment interaction.

Table 8.7 Mean performance of three-way hybrids for grain yield and other agronomic traits under optimum conditions
in the 2011 winter season
Entry GYD AD ASI PH EH EPO RL SL EPP ER TEX EA
Ten best hybrids SC9//L10 4.79 69.9 2.1 162.9 90.9 0.57 34.6 17.8 0.66 15.4 3.6 4.1
SC10//L5 4.65 71.2 2.9 174.0 90.7 0.54 5.2 16.5 0.94 7.8 3.1 3.7
SC4//L10 4.64 70.8 1.4 171.2 103.1 0.60 12.0 9.6 0.75 8.6 3.7 3.9
SC6//L10 4.64 71.2 2.0 176.5 98.3 0.56 30.4 7.6 0.83 8.6 3.8 3.8
SC8///L10 4.63 71.6 1.8 179.6 98.5 0.55 2.2 6.9 0.79 11.0 3.6 4.1
SC3//L8 4.63 69.7 4.0 157.6 82.2 0.52 0.9 20.2 0.84 2.8 3.3 3.8
SC11//L5 4.60 69.8 4.0 156.8 89.7 0.58 2.5 22.3 0.91 3.3 2.7 3.5
SC6//L5 4.58 73.6 0.9 187.0 104.2 0.56 9.7 9.8 0.69 12.1 3.7 3.3
SC9//L9 4.52 75.1 0.7 177.7 99.6 0.57 2.3 43.4 1.25 6.5 4.8 2.6
SC1//L6 4.50 71.6 2.1 172.6 89.1 0.53 9.4 11.2 0.79 3.2 3.3 3.6

Ten poorest SC9//L12 2.61 72.0 1.0 165.2 80.5 0.49 14.0 16.3 0.79 25.2 3.8 4.4
hybrids 013WH63 2.56 72.4 -0.2 171.0 93.0 0.54 24.2 18.9 0.60 5.8 3.8 4.6
SC3//L6 2.54 70.8 3.3 164.3 83.3 0.52 2.2 12.3 0.77 8.4 3.8 4.0
SC3//L7 2.43 70.2 3.3 149.5 69.3 0.48 1.1 25.9 0.81 2.9 3.2 3.9
SC2//L3 2.37 71.6 1.8 174.2 83.6 0.49 0.2 21.9 0.85 18.6 3.9 4.0
SC7//L4 2.36 73.2 0.3 185.2 104.5 0.58 38.7 20.3 0.65 9.7 4.0 4.2
SC3//L1 2.24 69.7 7.0 171.0 88.0 0.52 3.7 20.4 0.84 21.6 3.9 4.6
SC1//L5 2.20 71.3 3.8 158.3 81.3 0.52 5.6 19.4 0.90 10.2 3.0 3.3
SC1//L4 1.99 72.7 0.0 178.1 100.3 0.57 26.4 16.6 0.66 18.3 4.2 4.6
013WH01 1.95 74.7 0.5 183.8 106.6 0.59 31.1 10.6 0.65 7.6 3.6 4.1
Mean 3.51 71.8 2.4 172.6 93.0 0.55 5.4 16.6 0.78 12.2 3.5 3.9
LSD 1.33 1.6 2.6 13.0 11.6 0.06 29.4 17.5 0.21 12.8 0.7 0.8
MSE 0.90 2.0 1.7 128.8 101.5 0.00 218.8 154.1 0.02 41.3 0.3 0.4
Min 1.95 68.5 -0.6 149.5 69.3 0.48 11.5 2.3 0.51 1.1 2.6 2.6
Max 4.79 75.2 7.0 190.7 116.4 0.66 57.8 43.4 1.25 27.1 4.8 4.7
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem lodging; EPP=ears per plant; ER=ear
rot; TEX=texture; EA=ear aspect; LSD=least significant difference (0.05); MSE=mean square error; Min=minimum; Max=maximum.

247
Table 8.8 Pearson’s coefficient of correlation of grain yield with other agronomic traits
under optimum conditions

GYD AD ASI PH EH EPO RL SL EPP ER TEX

AD 0.00
ASI -0.13 -0.64**
PH 0.02 0.59** -0.30**
EH 0.24* 0.47** -0.45** 0.65**
EPO 0.27* 0.17 -0.36** 0.11 0.80**
RL -0.17 0.10 -0.24* 0.02 0.17 0.23*
SL -0.19 0.00 0.10 -0.16 -0.19 -0.11 -0.71
EPP 0.22* -0.21 0.29** -0.23* -0.23* -0.12 -0.40** 0.35**
ER -0.34** -0.10 0.17 0.08 -0.09 -0.17 -0.10 0.05 -0.14
TEX -0.27* 0.29** -0.10 0.31** 0.12 -0.09 0.10 0.08 -0.08 0.23
EA -0.48** -0.12 0.01 0.61 0.01 -0.01 0.10 -0.06 -0.31** 0.38** 0.26*
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear
position; RL=root lodging; SL=stem lodging; EPP=ears per plant; ER=ear rot; TEX=texture; EA=ear aspect.

The correlation of grain yield with ear rot, texture and ear aspect was negative and
significant. The highest correlation of 0.80 (P≤0.01) was observed between ear height and ear
position. The heritability estimate for grain yield under optimum conditions was 0.62 (Table
8.9). Other traits with heritability estimates above 0.50 were plant and ear height and ear
position. Traits such as ears per plant and root and stem lodging had negligible heritability
estimates.

8.3.3 Combined analysis

A combined ANOVA for optimum and managed drought environments was done for grain
yield and other yield related agronomic traits (Table 8.10). Environments, genotypes as well
as G x E interaction mean squares were significant (P≤0.001) for all traits. The variance
components were estimated from ANOVA mean squares in order to estimate heritability.
Generally the heritability estimates combined for all trials were lower than estimates realised
under drought and optimum conditions separately. The highest heritability estimate of 0.86
was realised for ear height and the lowest heritability estimate of 0.19 for anthesis silking
interval (Table 8.11). Grain yield had negative and significant correlation with anthesis days,
anthesis silking interval and plant height. The correlation of grain yield with ears per plant
was positive and significant (0.59). The highest positive and significant correlation was
realised for ear height and ear position (0.83) (Table 8.12).

248
Table 8.9 Genotypic and phenotypic variance estimates and broad sense heritability of
the agronomic traits under optimum conditions

Trait σ2g σ2 p h2B

ASI 0.0455 0.303 0.15
PH 3.224 4.91 0.66
EPO 0.0001 0.0001 0.60
EPP 0.000006 0.0001 0.06
EA 0.0017 0.0052 0.33
ER 0.6466 1.4496 0.45
RL 0.2258 1.82 0.12
SL 0.012 1.1956 0.01
TEX 0.0031 0.0088 0.34
EH 10.62 13.32 0.79
GYD 0.0566 0.0917 0.62
σ2g=genotypic variance; σ2p=phenotypic variance; h2B=broad sense heritability; ASI=anthesis silking interval; PH=plant height;
EPO=ear position; EPP=ears per plant; EA=ear aspect; ER=ear rot; RL=root lodging; SL=stem lodging; TEX=texture; EH=ear
height; GYD=grain yield.

Table 8.10 Combined analysis of variance for agronomic traits in the 2011 winter
season

Source DF GYD AD ASI PH EH EPO RL SL EPP

Site 2 3.37*** 25.83*** 189.52*** 221531.61*** 46892.45*** 0.27*** 2879.91*** 14157.11*** 1.09***
Genotype 84 0.93*** 1.17*** 26.81*** 790.67*** 829.79*** 0.04*** 220.75** 323.64*** 0.09***
Treatment 1 5.44*** 8.66*** 301.99*** 26581.23*** 1669.63*** 0.042 824.76*** 216.12*** 6.72***
GxE 168 0.47*** 0.62*** 16.77*** 228.35*** 147.56*** 0.02*** 149.99*** 150.29*** 0.03***
ExT 2 26.14*** 27.56*** 98.11*** 121345.05*** 31478.83*** 0.035 21768.01*** 878.77*** 0.24***
GxT 84 0.56*** 0.40*** 21.91*** 190.97*** 162.44*** 0.02*** 121.59*** 188.57*** 0.04***

GxExT 168 0.38*** 0.39*** 18.58*** 188.20*** 138.00*** 0.03*** 109.55*** 160.09*** 0.03***

Residual 508 0.08 0.11 7.01 47.82 11.06 0.01 30.62 35.57 0.008
***P≤0.01; DF=degrees of freedom; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear
height; EPO=ear position; RL=root lodging; SL=stem lodging; EPP=ears per plant; Env=environment; G x E=genotype x
environment interaction; E x T=environment x treatment; G x T=genotype x treatment G x E x Y=genotype x environment x
treatment.

249
Table 8.11 Genotypic and phenotypic variances and broad sense heritability for critical
agronomic traits in combined analysis
Trait σ2g σ2p h2B
GYD 0.0708 0.1508 0.47
AD 0.088 0.1983 0.45
ASI 1.65 8.66 0.19
PH 61.9042 109.7241 0.56
EH 68.2275 79.2875 0.86
EPO 0.0025 0.0125 0.20
EPP 0.0068 0.0148 0.46
σ2g=genotypic variance; σ2p=phenotypic variance; h2B=broad sense heritability; GYD=grain yield; AD=anthesis days; ASI=anthesis
silking interval; PH=plant height; EH=ear height; EPO=ear position; EPP=ears per plant.

In combined analysis genotype SC8//L10 was the best performer with a mean yield of 3.76 t
ha-1 and the poorest performing genotype was 013WH63 (Table 8.13). The new three-way
hybrids generally outperformed the locally grown varieties that were included in evaluations
as check varieties such as SC635 and SC513. Hybrids displayed good stand ability properties
and good ear placement. The ear per plant mean of 0.66 was obtained in combined analysis
indicating that hybrids had good prolificacy.

Table 8.12 Pearson’s correlation coefficients among agronomic variables in the

combined analysis
GYD AD ASI PH EH EPO
AD -0.25*
ASI -0.39** 0.16
PH -0.27* 0.56** 0.35**
EH -0.07 0.51** 0.10 0.65**
EPO 0.10 0.32** -0.76 0.21 0.83**
EPP 0.59** -0.59** -0.52** -0.57** -0.47** -0.25*
**P≤0.01; *P≤0.05; GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear
position; EPP=ears per plant.

250
Table 8.13 Mean performance of hybrids for grain yield and other agronomic traits in combined analysis in the 2011
winter season
Entry GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA
Ten best hybrids SC8//L10 3.76 68.5 1.3 178.3 99.8 0.54 6.6 5.8 0.70 14.8 0.5 3.4 3.7
SC10//L6 3.68 69.5 1.1 169.5 90.7 0.54 4.2 10.9 0.71 13.8 0.5 3.5 3.5
SC10//L5 3.62 68.2 1.3 175.1 90.6 0.51 2.6 24.2 0.83 5.6 0.6 3.1 3.4
SC4//L10 3.58 67.7 1.4 170.9 101.3 0.58 22.5 3.2 0.74 19.1 0.6 3.7 3.7
SC2//L9 3.50 70.7 1.5 181.5 96.3 0.54 0.1 12.8 0.72 22.5 0.6 3.2 3.8
SC3//L8 3.44 66.9 1.2 159.6 82.0 0.51 3.8 23.2 0.66 136.4 0.6 3.1 3.8
SC11//L5 3.39 67.4 1.6 160.3 92.1 0.53 7.5 26.1 0.75 15.5 0.6 2.4 3.1
SC1//L6 3.36 70.1 1.5 170.8 90.4 0.52 7.1 6.0 0.68 19.0 0.6 3.2 3.5
013WH29 3.32 69.2 1.8 164.7 95.0 0.57 6.6 5.5 0.72 14.8 0.5 3.1 3.5
SC7//L5 3.31 67.1 2.9 167.3 78.9 0.48 0.3 20.6 0.80 8.4 0.5 2.3 3.2

Ten poorest hybrids SC7//L3 2.19 69.1 4.8 178.4 90.2 0.50 11.7 5.2 0.51 19.4 0.5 3.2 4.0
SC2//L1 2.09 69.0 6.6 181.0 91.6 0.51 12.3 9.0 0.56 35.4 0.6 3.6 4.1
SC3//L6 2.08 68.9 1.5 167.1 80.9 0.50 6.4 15.0 0.65 11.8 0.6 3.5 3.7
SC10//L4 2.08 72.4 5.5 190.1 109.7 0.59 9.9 15.5 0.43 37.3 0.5 3.6 4.0
SC1//L7 2.05 70.0 2.5 166.0 86.1 0.50 2.3 26.0 0.61 12.6 0.6 3.3 3.9
SC11//L4 1.91 70.2 2.9 184.4 104.1 0.56 38.1 19.0 0.56 19.6 0.5 3.5 3.6
SC1//L4 1.79 70.2 3.1 181.3 98.2 0.54 14.0 12.7 0.58 28.5 0.6 3.9 3.9
SC7//L4 1.77 70.2 4.5 187.7 103.7 0.56 24.2 11.4 0.50 28.0 0.5 3.6 3.9
013WH01 1.70 72.2 4.0 187.5 107.9 0.57 19.5 7.0 0.46 32.1 0.6 3.3 3.6
013WH63 1.62 70.2 3.8 175.8 95.0 0.54 20.7 11.6 0.48 31.9 0.6 3.9 4.4
Mean 2.73 69.5 2.5 173.3 93.7 0.53 9.3 14.9 0.66 23.0 0.6 3.4 3.7
LSD (0.05) 0.84 1.3 1.9 10.7 8.7 0.05 17.9 10.5 0.13 35.1 0.1 0.5 0.6
MSE 0.72 2.0 3.8 116.7 116.1 0.00 161.6 139.0 0.02 1248.7 0.0 0.3 0.4
Min 1.62 65.9 -0.2 148.1 75.8 0.46 0.1 2.0 0.43 2.4 0.5 2.3 2.9
Max 3.76 72.8 6.6 190.1 117.0 0.64 40.5 29.4 0.84 136.4 0.7 4.3 4.5
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; EPO=ear position; RL=root lodging; SL=stem lodging; EPP= ears per plant;
ER=ear rot; SEN=senescence; TEX=texture; EA=ear aspect; LSD=least significant difference; MSE= mean square error; Min=minimum; Max=maximum.

251
8.3.4 Correlation between the predicted and observed mean yield
The predicted mean yields were positively and significantly (0.27) correlated with observed
mean yield although the correlation was relatively small (data not shown). The predicted
means followed a similar trend with the observed means (Figure 8.1). The arrow shows the
drop in yield within the predicted and observed means where yield drops as the graph
approaches the poorest performing hybrids. The ten best performing hybrids in the combined
analysis also had high predicted means compared to the poorest performing hybrids.

12.00

10.00

8.00
Grain yield t ha-1

6.00
PR
OB
4.00

2.00

0.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Hybrid number

Figure 8.1 Predicted and observed mean yield for the best 10 and poorest 10 hybrids
for combined environments.
PR=predicted; OB=observed.

252
8.4 Discussion
Hybrid varieties evaluated in this study revealed varied performance for yield and yield
components. The final grain yield is a function of the combined individual yield components,
which are likely to be influenced by genetic as well as environmental factors. The good
performance demonstrated by genotypes was an indication that they were able to adapt to the
environments in which they were evaluated. Results therefore suggest different genetic
backgrounds among hybrids studied, as was also indicated by Khan et al. (2002). This is
ideal as it provides an opportunity for selecting genotypes based on their performance for the
different measured traits. In this study grain yield under drought conditions was 55% of the
grain yield realised under optimum conditions and this was equivalent to severe stress. The
new three-way hybrids were evaluated against check varieties, which included the current
experimental hybrids and locally grown maize varieties. The new hybrids generally
outperformed the check varieties both under drought and optimum conditions. Three hybrids
namely RA214P/CML538//RS61P, RS61P/CML444//CML538 and RS61P/CML444//
CML539 showed stable performance across different conditions and were amongst the ten
best performing hybrids under drought, optimum and combined analysis. This therefore
means that there is no yield penalty if good rainfall occurs in a drought prone area. Hybrids
outperformed the locally grown varieties such as SC635 and SC513 and also outperformed
the recently released hybrid from the national programme, 013WH63. RA214/CML538//
RS61P outperformed the check varieties SC635, SC513 and 013WH63 by 1.7%, 19.5% and
88.9% respectively whilst RS61P/CML444//CML538 by 22.2%, 43.8% and 127.2%
respectively and RS61P/CML444//CML539 by 20.2%, 41.4% and 123.5% respectively. The
results show that there was a significant progress made in terms of producing superior
hybrids.

The maturity range of hybrids in the study was early to medium and displayed short
anthesis- silking intervals. The environmental mean for anthesis-silking interval ranged from
-0.2 to 6.6 and results are similar to results reported by Betran et al. (2003) and Salami et al.
(2007). The shortened anthesis-silking interval observed in this study is desirable and
according to Edmeades et al. (1993) low anthesis-silking intervals enhance maize tolerance
to stresses during flowering and ensures good grain filling. Results in this study showed that

253
hybrids had low root and stem lodging means, which is an indication of good stand ability
properties. The ability of maize hybrids to withstand root and stem lodging at optimum plant
populations is vital to obtaining high harvestable yields. Prolificacy (ears per plant) ranged
from 0.51-1.25 under optimum conditions to 0.26-0.83 under drought conditions. Betran et
al. (2003) and Derera et al. (2008) also reported reduced ears per plant under drought
conditions. Salami et al. (2007) reported no prolificacy since none of the cultivars evaluated
had ears per plant exceeding one. Decreased number of ears per plant may occur as a result
of failure of fertilisation caused by large anthesis-silking intervals or increased rate of kernel
abortion due to water stress (Westgate and Bassetti, 1990).

Significant negative correlation between grain yield and maturity under drought conditions
was observed and there was a tendency for hybrids with shorter maturity periods to produce
higher grain yield. This might have been due to the shorter growing period where the later
maturing hybrids needed a longer growing season for different developmental stages. In
some cases earliness is associated with drought escape as hybrids tend to reach the grain
filling stage before being severely stressed. On the contrary Salami et al. (2007) reported
negative and non-significant correlation between grain yield and days to 50% anthesis. The
effectiveness of stress around flowering and the controlling of drought intensity are important
in order to reveal genetic variability for anthesis-silking interval and ears per plant. Grain
yield under drought conditions showed strong correlation with anthesis-silking interval (-
0.44) and ears per plant (0.76). Similar results were reported by Bolanos and Edmeades
(1996), Betran et al. (2003) and Magorokosho et al. (2003). In this study, the correlation of
grain yield with anthesis-silking interval and ears per plant became stronger under drought
conditions and increased from -0.13 and 0.22 respectively under optimum conditions to -0.44
and 0.76 respectively under drought conditions. These results further confirm that
anthesis- silking interval and ears per plant are useful secondary traits to select for grain yield
under drought conditions but that they are less useful under optimum conditions. Positive
correlations between yield and ears per plant are expected because yield is a dependent
variable of ears per plant. The importance of secondary traits in screening germplasm for
drought tolerance has been determined by observing genetic correlations with grain yield
(Betran et al., 2003).

254
Heritability estimates provide guidelines for the development of effective breeding strategies
(Smalley et al., 2004). In this study lower broad sense heritability estimates for grain yield
(0.47) than for ear (0.86) and plant (0.56) height were recorded. Results are consistent with
previous findings by Hallauer and Miranda (1988) and Smalley et al. (2004). Traits that are
more closely related to reproductive fitness have lower heritability and grain yield is critical
to reproductive fitness in maize whilst ear and plant heights are considered of less importance
to reproductive fitness. Broad sense heritability estimates for grain yield decreased from 0.62
under optimum conditions to 0.51 under drought conditions, whilst broad sense heritability
estimates for anthesis silking interval increased from 0.15 under optimum conditions to 0.49
under drought conditions. Heritability of 0.51 for grain yield under drought was large and
this implies that direct selection would be effective. Similar results were reported by Rosielle
and Hamblin (1981) and Magorokosho et al. (2003). These results further confirm the
importance of anthesis silking interval as a secondary trait used to increase grain yield when
selecting under drought conditions.

The performance of three-way cross hybrids can be predicted from single cross hybrids that
have been evaluated in as many environments as feasible. In this study the correlation
between the predicted and observed means was significant but the correlation was weak. This
might have been due to significant G x E interaction. Otsuka et al. (1972) found that G x E
interaction affected the correlation of observed and predicted performance much more than
did epistatic effects. The epistatic and G x E interaction effects were not considered during
yield prediction of the three-way hybrids in this study. Otsuka et al. (1972) also showed that
heritability of three-way crosses was higher than the correlation between the genotypic and
predicted three-way cross values probably because of epistasis. Contrary to results obtained
in this study Melchinger et al. (1987) reported high (0.86) correlation between the predicted
and observed three-way cross means. Results show that the three-way hybrids with superior
predicted yields can be evaluated in several environments and superior ones released for
commercial use. Genetic gain can be increased through employing higher selection intensity
among the predicted hybrids (Melchinger et al., 1987).

255
8.5 Conclusions
Generally the performance of hybrids indicated that they have good genetic potential, as
some showed high yield performance and good performance in other traits. Three hybrids,
RA214P/CML538//RS61P, RS61P/CML444//CML538 and RS61P/CML444//CML539 were
identified to have better performance compared to other hybrids. However, there still exists a
need for further evaluation of these hybrids on a larger scale in advanced variety trials before
considering them for release. There also exists a need to evaluate hybrids under low nitrogen
conditions as well. The amount of genetic variance, heritability and genetic correlation with
grain yield determine the relative usefulness of secondary traits such as ears per plant and
anthesis silking interval as indirect selection criteria for grain yield. This study therefore
identified the potential of anthesis silking interval as an indirect selection criterion for grain
yield under drought stress conditions. The usefulness of ears per plant as an indirect selection
criterion for grain yield under drought conditions was not convincingly determined in this
study due to low genetic variance and a negligible heritability estimate. The low and
negligible heritability estimates for ears per plant might have been an indication that the trait
was influenced by environmental factors. However heritability for grain yield under drought
conditions was larger relative to secondary traits such as ears per plant hence direct selection
would be the most appropriate. Results have shown that the number of three-way hybrids
handled by a breeding programme can be reduced by predicting their performance from
superior single cross hybrids followed by evaluating them in multi-location trials to identify
superior hybrids for release.

8.6 References
Agronomix Software, Inc. 2005. AGROBASE Generation II User’s Manual. Version II,
Revised Edition. www.agronomix.com. Agronomix Software, Winnipeg, M. B,
Canada.
Allard, R.W. and A.D. Bradshaw. 1964. Implications of genotype-environment interactions
in applied plant breeding. Crop Science 4: 503-508.
Becker, W.A. 1984. Manual of Quantitative Genetics. 4th edition, Academic Enterprises,
Pullman, Washington. pp 188.

256
Betran, F.J., D. Beck, M. Banziger and G.O. Edmeades. 2003. Secondary traits in parental
inbreds and hybrids under stress and non-stress environments in tropical maize.
Field Crops Research 83: 51-65.
Bolanos, J. and G.O. Edmeades. 1996. The importance of anthesis silking interval in
breeding for drought tolerance in tropical maize. Field Crops Research 48: 65-80.
CIMMYT-Zimbabwe. 2000. CIMMYT-Zimbabwe: 2000 Research Highlights. Harare.
Zimbabwe. pp 3-7.
Cockerham, C.C. 1961. Implications of genetic variances in a hybrid breeding program. Crop
Science 1: 47-52.
Derera, J., P. Tongoona, B.S. Vivek and M.D. Laing. 2008. Gene action controlling grain
yield and secondary traits in southern African maize hybrids under drought and non-
drought conditions. Euphytica 162: 411-422.
Eberhart, S.A. 1964. Theoretical relations among single, three way and double cross hybrids.
Biometrics 20: 522-539.
Eberhart, S.A. and C.O. Gardner. 1966. A general model for genetic effects. Biometrics 22:
864-881.
Edmeades, G.O., J. Bolanos, M. Hernandez and S. Bello. 1993. Causes for silk delay in a
lowland tropical maize population. Crop Science 33: 1029-1035.
Fehr, W.R. 1993. Principles of Cultivar Development: Theory and Technique. 1st edition.
Macmillan Publishing Company. USA. pp 250-294.
Hallauer, A.R. 1990. Methods used in developing maize inbreds. Maydica 35: 1-16.
Hallauer, A.R., and J.B. Miranda. 1988. Quantitative Genetics in Maize Breeding. 2nd
edition. Lowa State University Press. Ames Lowa.
Hinkelmann, K. 1968. Partial tetra-allele crosses. Theoretical and Applied Genetics 38: 85-
89.
Jenkins, M.T. 1934. Methods of estimating the performance of double crosses in corn.
Journal of American Society of Agronomy 26: 199-204.
Khan, Z., R.A. Hassanali, W. Overholt, T.M. Khamis, A.M. Hooper, J.A. Pickett, L.J.
Wadhams and C.M. Woodcock. 2002. Control of witchweed Striga hermonthica by
intercropping with Desmodium spp and the mechanism defined as allelopathic zone.
Chemical Ecology 28: 1871-1885.

257
Magorokosho, C., K.V. Pixley and P. Tongoona. 2003. Selection for drought tolerance in two
tropical maize populations. African Crop Science 11: 151-161.
Melchinger, A.E., H.H. Geiger, G. Seitz and G.A. Schmidt. 1987. Optimum prediction of
three-way crosses from single crosses in forage maize (Zea mays L.) Theoretical and
Applied Genetics 74: 339-345.
Otsuka, Y., S.A. Eberhart and W.A. Russel. 1972. Comparisons of prediction formulas for
maize hybrids. Crop Science 12: 325-330.
Rosielle, A.A. and J. Hamblin. 1981. Theoretical aspects of selection for yield in stress and
non-stress environments. Crop Science 21: 943-946.
Salami, A.E., S.A.O. Adegoke and O.A. Adegbite. 2007. Genetic variability among maize
cultivars grown in Ekiti-State, Nigeria. Middle East Journal of Scientific Research 2:
9-13.
Seitz, G. 2005. The use of doubled haploids in corn breeding. In: Proceedings of the 41st
Annual Illinois Corn Breeders’ school. Urbana-Champaign. pp 1-7.
Smalley, M.D., J.L. Daub and A.R. Hallauer. 2004. Estimation of heritability in maize by
parent-offspring regression. Maydica 49: 221-229.
Smith, O.S. 1986. Covariance between line per se and testcross performance. Crop Science
26: 540-543.
SPSS 15.0 for Windows. 2006. www.spss.com SPSS Inc., Chicago III.
Weatherspoon, J.H. 1970. Comparative yields of single, three-way and double crosses of
maize. Crop Science 10: 157-159.
Westgate, M.E. and P. Bassetti. 1990. Heat and drought stress in corn: what really happens to
the corn plant at pollination? In: Wilkinson, D. (ed) Proceedings of Annual Corn and
Sorghum Research Conference 45th, Chicago, ASTA. Washington, D.C. pp 12-28.

258
 CHAPTER 9
General conclusions and recommendations

Since maize is the staple food crop in Zimbabwe, the country requires 1.8 million ton for
consumption and 300 000 ton as national strategic reserve per annum. However, biotic
(insect pests and diseases) and abiotic (drought and low nitrogen) stresses pose many
challenges for maize production in the country. The crop is produced by large and small scale
commercial as well as communal farmers. The communal sector is, however, the largest
producer of maize in the country. The yield of maize in the communal sector has remained
below 0.5 t ha-1 and the sector is faced with major challenges, amongst them occurrence of
dry spells, unavailability of inputs and poorly adapted varieties. Therefore germplasm
improvement for drought tolerance remains a high priority in the country. The Zimbabwe
National Breeding Programme under the DR&SS has partnered with CIMMYT in producing
drought tolerant maize varieties. Drought tolerance is required for farmers to achieve high
and stable maize yields, especially for the communal farmers who are mostly located in the
drier parts of the country. This study was thus conducted to (i) estimate combining ability
and heterosis for grain yield and other agronomic traits between DR&SS and CIMMYT
white maize inbred lines under stress and non-stress environments, (ii) analyse G x E
interaction and stability of single cross hybrids for grain yield, (iii) examine genetic diversity
among DR&SS and CIMMYT white maize inbred lines using morphological traits and SNP
markers, (iv) assess the relationship between genetic diversity of DR&SS and CIMMYT
parental inbred lines and F1 performance, heterosis and SCA effects of hybrids under abiotic
and optimal environments, (v) estimate test-cross performance of F3 segregating populations
developed from CIMMYT drought tolerant donors and DR&SS elite inbred lines under
drought and optimal conditions and (vi) estimate performance and yield prediction of three-
way hybrids from drought tolerant single cross hybrids.

Outputs from the study indicated that an average of 112.29% MPH and 76.40% HPH were
realised across environments and was an indication of the potential of these inbred lines for
hybrid development. The negative heterosis (-5.1% MPH and -7.4% HPH) for days to
anthesis was an indication that hybrids were earlier compared to their parental inbred lines.

259
Significant positive correlations and regressions were recorded for SCA with MPH, HPH and
per se performance of hybrids under optimum and drought conditions. Results imply that an
improvement in selection for SCA, which is a good predictor of grain yield, will result in
indirect improvement of MPH and HPH under both optimum and drought conditions. The
HPH and MPH also showed significant positive association and linear regression along with
high coefficient of determination with per se performance of hybrids especially under
drought conditions.

Pearson’s correlation coefficients between grain yield and anthesis silking interval were
smaller under optimum conditions and they became larger under drought and low N
conditions. A strong and negative correlation of grain yield and anthesis silking interval
under stress conditions showed that genotypes with negative anthesis silking interval would
also have good grain yield performance. Reduced or negative anthesis silking interval under
stress environments is desirable as silking delay has been seen as the cause for poor
pollination and grain filling in maize. Similarly the relationship between grain yield and ears
per plant became stronger under stress environments. Positive and strong correlation of grain
yield and ears per plant under stress environments showed that genotypes with a high number
of ears per plant would also have high grain yield. Results confirmed ears per plant and
anthesis silking interval as important secondary traits to select for under stress environments.
Therefore these two traits can be used together with grain yield in selecting superior
genotypes under stress environments. Anthesis days were negatively correlated with grain
yield under both drought and low N. Early maturing genotypes tended to produce higher
yields as stress did not coincide with their critical stage of development. Results have shown
that selection for earliness in the breeding programme have produced genotypes that perform
well under short rainy seasons or where mid-season dry spells coincide with flowering of late
maturing genotypes. In some cases genotypes that performed well under optimum conditions
also performed well under drought conditions but this was not always true. It is thus critical
for the programme to focus on breeding for varieties that are drought tolerant and also
perform well under optimum conditions.

260
The genetic variability of lines was studied using morphological data and SNP markers. Both
methods indicated that there was variability amongst lines. UPGMA cluster analysis based
on morphological data grouped lines into five clusters mainly based on grain yield, anthesis
days, anthesis silking interval and plant height. SNP markers grouped lines into two major
groups and a number of subgroups. Within subgroups some lines clustered in accordance
with known heterotic groups and pedigree relationships. DR&SS and CIMMYT have been
using heterotic groups based on combining ability studies from diallel and NCDII
experiments over the years. Morphological traits have been used in defining heterotic groups.
These heterotic groups have proved to be useful as high levels of heterosis have been realised
from crossing lines from different defined groups. As a result a number of good hybrids have
been registered and availed to farmers for production. New testers that have been identified
can be used together with the old testers in predicting heterotic groups. However to
complement the currently used method, SNP markers can also be used in further defining the
heterotic groups. Information generated using SNP markers can confirm the existing groups
or can assist in defining new groups that could otherwise not be possible to define using only
morphological traits. The National Breeding Programme is recommended to use both
morphological and SNP markers in defining efficient and effective heterotic groups in order
to maximise on utilisation of existing and newly acquired germplasm.

Therefore superior germplasm identified in this study can be put into use in germplasm
improvement for stress environments. Lines RS61P, NAW5885 (from DR&SS) and
CML444, CML539, CML442, CML537 and CML548 (from CIMMYT) were identified as
having desirable GCA effects under both drought and low N conditions. The single cross
RS61P/CML444 was identified as a potential tester for the SC heterotic group, whilst
2N3d/CML548 was identified as a potential tester for the N3 heterotic group. The testcrosses
containing lines derived from DR&SS lines K64r, RS61P, NAW5885, SC5522 and
CIMMYT drought tolerant donors based on DTPWC9 were amongst the best performing
testcrosses in early and late maturing trials under drought, optimum and across
environments. Three-way cross hybrids namely RA214P/CML538//RS61P, RS61P/CML444
//CML538 and RS61P/CML444//CML539 were identified as having superior performance.

261
The two single crosses identified as potential testers were also identified as the most stable
genotypes across test environments. Therefore the national programme can now engage in
validation studies for these testers before they are used in predicting heterotic groups.
Identification of good single cross hybrids with good SCA effects and lines with good GCA
effects proved to be an important intervention for the breeding programme. The single
crosses and inbred lines were used to produce three-way hybrids which showed good
performance under both optimum and drought conditions. By doing so the breeding
programme has managed to produce final products within a short space of time. Results
revealed that additive and non-additive gene effects were important in expression of traits.
However, non-additive gene action assumed a more important role in expression of grain
yield and secondary traits (anthesis silking interval, ears per plant and senescence) under both
drought and low N conditions. Exotic germplasm was furthermore acquired from CIMMYT
Mexico and used to enhance the National Breeding Programme germplasm. By doing this, a
segregating population was developed to facilitate development of new inbred lines. Early
testcrossing has been used to assist in selecting superior segregating lines for further
generation advancement. Genetic variances and genetic gains for plant and ear height were
generally higher than those for other traits. Grain yield had a repeatability value of 0.41 for
late maturing testcrosses and good repeatability values were also recorded for plant and ear
height and ears per plant. It is important for breeding programmes to constantly acquire
germplasm from other breeding programmes to improve on genetic diversity and enhance
quality of breeding material.

Finally it is recommended that the national programme utilises lines identified as having
good GCA as parental lines in the hybrid breeding programme. Genetic relationships
determined using morphological and SNP data will further assist breeders in identifying
divergent parents and designing an effective crossing programme. Identified single cross
testers can be used in grouping new inbred lines into heterotic groups, however there exists a
need to validate testers through further combining ability studies before they can be put to
use in the breeding programme. The information generated from testcross evaluation will
assist breeders in selecting segregating lines for further generation advancement until
homozygosity is attained. The identified three-way hybrids can be included in multi-location

262
national advanced variety trials before they are considered for release. There is also need to
evaluate the testcrosses and the three-way hybrids under low N conditions. With the current
climate change, improving germplasm for drought tolerance is an important intervention
therefore the National Breeding Programme will in the near future be able to register drought
tolerant lines and hybrids as products of this study.

263
 SUMMARY

Breeding for drought and low N tolerant genotypes in Zimbabwe is an important

intervention that will see the country curbing the food in-security problem. Both CIMMYT
and DR&SS maize germplasm was used in this study that was conducted in Zimbabwe in
the 2009/10 and 2010/11 seasons. Evaluations were done under optimum, drought and low
N conditions. One of the objectives was to estimate combining ability and heterosis for
grain yield and other agronomic traits of white maize inbred lines under stress and optimal
environments. Line x tester analysis of 23 inbred lines identified RS61P, NAW5885 (from
DR&SS) and CML444, CML539, CML442, CML537 and CML548 (from CIMMYT) as
having desirable GCA effects under both drought and low N conditions. Additive and non-
additive gene effects were important in the expression of traits across all environments;
however non-additive gene effect assumed a more important role in the expression of traits
under stress environments. The single crosses RS61P/CML444 and 2N3d/CML548 were
identified as potential testers for the SC and N3 heterotic groups respectively. In the
analyses of G x E and stability using AMMI and GGE biplot the same single crosses were
identified as the most stable. There were three mega environments identified within the test
environments and Agricultural Research Trust farm site was the most powerful in
discriminating genotypes. Genetic diversity amongst the 23 inbred lines was examined
using 14 morphological traits and 1 129 SNP markers. The morphological data revealed
variability amongst inbred lines that could be manipulated through selection and
hybridisation. Variability was further substantiated using PCA where the overall diversity
could not be explained by a few eigenvectors and the major contributors were grain yield,
texture, ear aspect, common rust, GLS and anthesis days. Euclidean and Rogers’
dissimilarity matrices based on morphological and SNP data respectively clustered lines
related by pedigree together in some cases. The SNP dendrogram had a high goodness of fit
value (r=0.87) compared to the morphological dendrogram, which showed that it grouped
the lines efficiently, although at times it was not in agreement with the known heterotic
grouping that was previously established using testers. The assessment of correlation
between genetic distances, F1 performance, heterosis and SCA revealed significant positive
correlations and regressions between SCA, MPH, HPH and per se performance of hybrids.

264
 The HPH and MPH also showed significant positive association and linear regression along
with high coefficient of determination with per se performance of hybrids, especially under
drought conditions. Correlations of genetic distances with MPH and HPH were too low to
be of predictive value. An average of 112.29% MPH and 76.40% HPH were realised across
environments and this was an indication of the potential of these inbred lines for hybrid
development. The segregating lines at F3 stage were testcrossed to group A
(CML539/CML442) and B (CML444/CML395) testers and testcrosses containing lines
derived from DR&SS lines K64r, RS61P, NAW5885, SC5522 and CIMMYT drought
tolerant donors based on DTPWC9 were generally amongst the best performing testcrosses
in early and late maturing trials. Three-way hybrid performance was predicted from 11
single cross hybrids and results showed that there was significant but weak correlation
between the predicted and the observed grain yield means and this could be explained by
epistatic and significant G x E interaction, which were not taken into account in the
prediction equation. Three-way cross hybrids identified as having superior performance
under drought and well-watered conditions included RA214P/CML538//RS61P, RS61P/
CML444//CML538 and RS61P/CML444//CML539. However, there is still need to evaluate
these hybrids under low N conditions before they can be recommended for release.

Key words: combining ability, genetic diversity, heterosis.

265
 OPSOMMING
Die ontwikkeling van genotipes vir Zimbabwe wat droogte en lae N tolerant is kan ‘n
belangrike bydrae maak om die probleem van voedselsekuriteit aan te spreek. Beide
CIMMYT en DR&SS mieliekiemplasma is in hierdie studie gebruik wat uitgevoer is in
Zimbabwe in die 2009/10 en 2010/11 seisoene. Die proewe is uitgevoer onder optimum,
droogte en lae N toestande. Een van die belangrikste doelwitte was om kombineervermoë
en heterose vir graanopbrengs en ander agronomiese eienskappe van wit mielie ingeteelde
lyne onder stremmings en optimale toestande te bepaal. Lyn x toetser analise van 23
ingeteelde lyne het RS61P, NAW5885 (van DR&SS) en CML444, CML539, CML442,
CML537 en CML548 (van CIMMYT) geïdentifiseer as lyne met die beste GCA effekte
onder beide droogte en lae N toestande. Beide additiewe en nie-additiewe geeneffekte was
belangrik by die uitdrukking van eienskappe oor alle omgewings; maar nie-additiewe
geeneffekte was meer belangrik by die uitdrukking van eienskappe onder
stremmingstoestande. Die enkelkruise RS61P/CML444 en 2N3d/CML548 is geïdentifiseer
as potensiële toetsers vir die SC en N3 heterotiese groepe onderskeidelik. In die analise van
G x E en stabiliteit met die gebruik van AMMI en GGE biplotte is dieselfde enkelkruise as
die mees stabiel geïdentifiseer. Daar was drie mega-omgewings binne die toetsomgewings
en die “Agricultural Research Trust” plaas omgewing was die mees effektief om te
onderskei tusssen genotipes. Genetiese diversiteit tussen die 23 ingeteelde lyne is ondersoek
met die gebruik van 14 morfologiese eienskappe en 1 129 SNP merkers. Die morfologiese
data data het variasie tussen ingeteelde lyne gewys wat gemanipuleer kan word deur
seleksie en hibridisasie. Variasie is verder bevestig met die gebruik van PCA waar totale
variasie nie verklaar kon word deur enkele eigenvektore nie, en waar die meeste variasie
verklaar is deur graanopbrengs, tekstuur, kopaspek, gewone roes, GLS en dae tot antese.
Euklidiese en Rogers se matrikse van verskille gebasseer op morfologiese en SNP data
onderskeidelik het in sommige gevalle die lyne gegroepeer volgens stambome. Die SNP
dendrogram het die hoogste akkuraatheid (r=0.87) getoon in vergelyking met die
morfologiese dendrogram, wat gewys het dat lyne effektief gegroepeer is, alhoewel dit
soms nie in ooreenstemming was met bekende heterotiese groeperings wat vroeër bepaal is
met toetsers nie. Die evaluasie van korrelasies tussen genetiese afstande, F1 prestasie,
heterose en SCA het betekenisvolle positiewe korrelasie en regressie tussen SCA, MPH,

266
 HPH en per se prestasie van basters getoon. Die HPH en MPH was ook betekenisvol
positief geassosieër en liniêre regressie sowel as ‘n hoë koeffisiënt van bepaling is gesien
met die per se prestasie van basters, veral onder droogtetoestande. Korrelasies van
genetiese afstande met MPH en HPH was te laag om voorspellingswaarde te hê. ‘n
Gemiddeld van 112.29% MPH en 76.40% HPH is gesien oor omgewings. Dit is ‘n
aanduiding van die potensiaal van hierdie ingeteelde lyne vir basterontwikkeling. Die
segregerende F3 lyne is getoetskruis met groep A (CML539/CML442) en B
(CML444/CML395) toetsers en toetskruise van lyne afkomstig van DR&SS lines K64r,
RS61P, NAW5885, SC5522 en CIMMYT droogtetolerante skenkers gebasseer op
DTPWC9 was oor die algemeen die beste presterende toetskruise in die vroeë en laat
rypheidstyd proewe. Drierigting basterprestasie is voorspel vanaf 11 enkelkruise en
resultate het getoon dat daar betekenisvolle maar lae korrelasies was tussen die voorspelde
en die werklike graanopbrengs. Dit kon verklaar word deur die epistatiese en betekenisvolle
G x E interaksies wat nie in ag geneem is in die voorspellingsformule nie.
Drierigtingkruisbasters wat geïdentifiseer is wat baie goed presteer het onder beide droogte
en optimum toestande, was RA214P/CML538//RS61P, RS61P/CML444//CML538 and RS
61P/CML444//CML539. Dit is egter nog steeds nodig om hierdie basters onder lae N
toestande te evalueer voor hulle aanbeveel kan word vir vrystelling.

Sleutelwoorde: kombineervermoë, genetiese diversiteit, heterose

267
 Appendices
Appendix 1 Single cross hybrids produced

Entry Pedigree
1 N3233-B/CML536
2 CML395-B/N3233-B
3 CML442-B/N3233-B
4 CML442-B/CML539
5 CML444-BB/N3233-B
6 CML537/N3233-B
7 CML538/N3233-B
8 CML539/N3233-B
9 CML548/N3233-B
10 CML545/N3233-B
11 CZL03007/N3233-B
12 SC5522-B/CML536
13 CML395-B/SC5522-B
14 CML444-BB/SC5522-B
15 CML545/C5522-B
16 CZL03007/SC5522-B
17 2Kba-B/CML444-BB
18 CML395-B/2Kba-B
19 CML442-B/2Kba-B
20 CML537/2Kba-B
21 CML538/2Kba-B
22 CML539/2Kba-B
23 CML545/2Kba-B
24 CZL052/2Kba-B
25 CZL03007/2Kba-B
26 K64r-B/CML312-B
27 K64r-B/CML536
28 K64r-B/CZL052
29 CML395-B/K64r-B
30 CML442-B/K64r-B
31 CML444-BB/K64r-B
32 CML537/K64r-B
33 CML538/K64r-B
34 CML539/K64r-B
35 CML548/K64r-B
36 CML545/K64r-B
37 CZL052/K64r-B
38 CML442-B/NAW5885-B
39 CML444-BB/NAW5885-B
40 CML537/NAW5885-B

268
 Appendix 1 72 single cross hybrids produced

41 CML538/NAW5885-B
42 CML539/NAW5885-B
43 CML548/NAW5885-B
44 CML545/NAW5885-B
45 CZL03007/NAW5885-B
46 CML395-B/2N3d-B
47 CML442-B/2N3d-B
48 CML444-BB/2N3d-B
49 CML537/2N3d-B
50 CML538/2N3d-B
51 CML539/2N3d-B
52 CML548/2N3d-B
53 CML545/2N3d-B
54 CML395-B/RS61P-B
55 CML442-B/RS61P-B
56 CML442-B/CML537
57 CML444-BB/RS61P-B
58 CML537/RS61P-B
59 CML538/RS61P-B
60 CML539/RS61P-B
61 CML548/RS61P-B
62 CML545/RS61P-B
63 CZL03007/RS61P-B
64 RA214P-B/CML444-BB
65 CML395-B/RA214P-B
66 CML442-B/RA214P-B
67 CML537/RA214P-B
68 CML538/RA214P-B
69 CML539/RA214P-B
70 CML548/RA214P-B
71 CML545/RA214P-B
72 CZL03007/RA214P-B

269
Appendix 2 Performance of genotypes for grain yield and other agronomic traits across 14 environments in the 2009/10
and 2010/11 seasons
ENTRY GYD AD ASI PH EH RL SL EPP ER GLS RUST ET SEN
t ha-1 d d cm cm % % # % 1-5 1-5 1-5 1-10
74 5.62 75.3 1.4 266.7 138.6 5.7 5.0 0.80 8.4 1.5 1.3 2.2 1.9
61 5.31 69.7 0.6 230.7 119.5 11.0 3.7 0.92 5.0 2.7 1.0 1.5 2.1
57 4.94 72.4 1.5 240.5 133.3 6.9 6.7 0.91 5.2 1.6 1.0 2.6 1.9
54 4.89 70.2 1.6 242.2 138.7 8.3 4.6 0.87 6.9 2.1 0.9 1.8 1.9
48 4.88 73.0 1.1 267.9 147.0 7.7 3.7 0.96 8.1 3.9 1.4 1.1 1.8
63 4.78 69.8 1.2 227.3 121.1 19.4 6.7 0.85 4.7 1.3 1.0 1.6 2.1
52 4.76 71.0 1.0 244.4 125.3 9.4 1.8 0.86 12.7 3.1 1.3 2.0 2.2
79 4.74 75.1 0.9 243.3 135.6 5.4 5.5 0.83 4.4 1.9 1.0 1.5 1.8
45 4.74 70.2 1.9 239.6 122.5 7.7 7.9 0.84 6.3 2.7 1.0 1.5 2.1
68 4.69 70.7 1.5 235.5 114.6 10.0 0.6 0.89 7.4 1.8 1.0 1.5 2.3
10 4.69 69.6 1.2 235.2 124.3 5.5 2.5 0.82 7.6 3.3 1.0 2.2 2.0
72 4.63 69.9 -0.3 229.0 112.0 6.5 3.1 0.93 4.6 1.5 1.0 1.5 2.1
34 4.63 66.7 0.2 213.8 101.2 11.9 15.7 0.94 5.8 3.0 1.0 2.1 2.2
38 4.63 69.1 2.4 242.4 121.4 7.2 7.1 0.88 5.4 3.4 1.1 2.0 1.7
60 4.62 66.9 0.5 223.7 117.8 7.7 3.1 0.98 6.6 2.7 0.9 1.5 2.0
7 4.61 70.0 2.0 244.2 120.3 10.7 5.6 0.77 8.3 1.8 0.9 1.8 2.0
58 4.61 69.4 1.5 226.4 113.5 8.6 5.7 0.95 5.0 2.2 1.0 1.9 1.9
59 4.56 69.3 0.0 220.6 119.2 2.6 8.4 0.88 6.9 1.4 0.9 2.3 2.1
70 4.55 72.9 3.0 244.1 128.4 4.5 4.8 0.73 8.3 1.6 1.0 1.1 1.9
8 4.49 69.8 1.8 236.2 122.3 11.5 7.0 0.90 4.8 4.2 0.9 1.7 2.3
51 4.48 70.3 0.8 237.4 115.7 11.2 7.6 0.84 8.1 3.6 1.0 1.6 2.0
62 4.47 68.0 0.0 229.5 125.3 10.0 2.7 1.00 8.8 1.4 0.9 1.1 1.9
5 4.41 73.0 1.5 250.4 138.7 12.7 13.1 0.79 7.3 3.4 1.1 2.3 2.0
47 4.41 71.3 0.8 242.5 120.3 6.8 7.6 0.85 8.2 4.3 1.5 2.1 1.8
77 4.32 72.3 1.6 235.9 116.5 13.0 -0.6 0.82 9.7 1.9 0.9 2.2 2.1
4 4.30 70.0 1.5 217.8 105.4 20.8 3.5 0.99 5.6 2.7 1.3 2.1 2.1
27 4.29 71.1 -0.4 241.1 119.6 12.1 3.3 0.82 7.2 2.5 1.4 2.0 1.9
28 4.28 68.3 1.1 227.9 114.7 10.7 5.7 0.86 4.8 1.7 1.7 3.0 2.2
26 4.27 69.8 1.2 235.2 122.2 3.8 3.2 0.88 7.2 1.9 0.9 1.6 2.0
78 4.27 72.6 1.5 241.2 122.1 7.3 4.9 0.79 7.9 2.0 1.0 2.5 1.9
31 4.21 71.7 1.2 239.9 126.6 13.1 5.1 0.82 10.3 2.5 1.5 2.2 2.0
71 4.18 69.2 0.5 232.9 115.9 7.7 1.9 0.89 9.3 2.3 0.9 1.1 2.2
6 4.17 71.6 1.6 254.1 130.1 7.6 1.5 0.83 10.2 3.7 1.0 2.0 2.2
32 4.15 69.1 0.6 232.1 113.9 13.7 2.2 0.86 6.2 3.2 1.1 2.7 2.3
1 4.12 72.1 1.7 253.6 138.3 4.3 5.8 0.90 6.3 2.1 1.6 2.0 2.0
80 4.10 70.5 1.3 229.9 110.6 5.5 2.6 0.84 3.8 1.7 1.0 2.0 1.8
3 4.09 70.1 2.4 246.4 123.3 17.1 7.7 0.84 5.3 3.9 1.2 1.5 2.1
16 4.08 72.0 1.0 244.7 130.3 6.6 14.5 0.85 7.8 2.6 0.9 2.1 2.0
55 4.07 70.4 0.5 232.6 126.9 7.7 2.9 0.96 5.2 2.6 1.0 2.2 1.7

270
 Appendix 2 Performance of genotypes for grain yield and other agronomic traits across 14 environments in the the 2009/10 and
2010/11 seasons
ENTRY GYD AD ASI PH EH RL SL EPP ER GLS RUST ET SEN
t ha-1 d d cm cm % % # % 1-5 1-5 1-5 1-10
11 4.07 70.1 2.0 245.3 128.2 24.7 16.9 0.81 6.1 1.8 1.3 1.6 2.3
49 4.05 72.3 1.8 248.0 126.6 5.3 9.1 0.95 9.3 3.4 1.0 1.9 2.2
15 4.04 70.4 0.7 235.9 126.4 7.2 0.4 0.89 7.3 2.0 1.1 1.7 2.3
36 4.03 66.9 -0.1 228.1 111.8 9.7 4.1 0.90 11.7 2.2 1.2 2.0 2.1
35 4.02 69.1 1.2 230.1 108.8 10.7 6.2 0.86 6.0 2.3 1.2 1.9 2.5
66 4.01 72.2 2.3 239.5 116.5 9.3 0.1 0.82 6.8 1.8 1.0 2.0 2.2
19 4.01 68.1 1.3 231.2 113.5 13.5 4.5 0.85 6.3 3.7 1.6 2.1 1.9
67 4.01 73.5 1.8 256.2 131.1 7.8 1.1 0.88 8.2 2.3 1.1 1.9 2.3
9 3.99 70.4 1.5 237.9 125.4 12.5 9.8 0.83 10.5 3.3 1.0 1.7 2.3
21 3.98 68.2 1.1 233.1 114.2 10.3 4.0 0.82 5.6 1.3 1.0 1.4 2.1
22 3.97 66.6 0.7 230.2 114.1 10.2 4.4 0.89 4.6 3.0 1.0 1.0 2.3
20 3.96 69.3 1.4 233.7 120.6 13.3 2.2 0.83 8.4 2.4 0.9 1.4 2.2
53 3.95 69.6 0.1 240.6 127.2 6.7 13.0 0.92 11.2 4.1 0.9 2.0 2.2
46 3.91 72.9 1.9 261.8 135.9 14.5 11.0 0.88 9.2 2.8 1.1 2.4 2.3
76 3.90 69.0 1.3 240.0 119.3 10.1 0.9 0.99 6.1 2.4 1.2 1.5 2.0
33 3.89 68.2 0.0 219.7 108.2 12.4 4.0 0.78 7.7 2.2 1.0 2.8 2.0
65 3.88 74.9 2.8 258.7 138.5 12.9 5.8 0.75 5.1 2.0 1.1 2.5 2.1
40 3.85 72.4 1.9 241.1 118.7 13.2 3.3 0.80 12.0 2.3 0.9 1.0 2.0
44 3.85 68.6 2.3 241.9 117.6 7.1 3.9 0.77 8.4 2.4 1.1 1.6 1.9
2 3.83 71.3 2.2 249.2 138.1 16.4 9.9 0.69 15.5 3.7 1.6 2.7 2.2
30 3.81 68.3 0.8 224.0 112.1 10.5 2.0 0.86 8.0 3.8 1.2 2.5 2.1
75 3.81 72.0 2.0 253.5 120.1 7.3 8.1 0.93 5.4 2.3 1.5 1.3 2.0
43 3.74 72.1 2.7 240.9 124.1 8.1 4.4 0.81 9.5 1.7 1.4 1.1 2.2
64 3.69 75.4 1.5 259.5 138.1 0.7 5.9 0.86 7.4 1.4 1.2 1.4 2.0
41 3.67 71.0 1.9 245.4 122.7 11.4 8.5 0.76 8.9 2.9 0.9 1.8 1.9
23 3.62 66.7 0.4 227.5 117.4 10.7 9.0 0.88 9.0 2.1 1.3 2.3 2.1
25 3.61 68.4 1.1 229.1 111.0 13.6 13.3 0.77 6.9 2.0 1.0 2.2 2.4
24 3.59 66.7 0.8 228.2 116.7 10.9 4.8 0.87 7.5 2.7 1.0 1.7 2.2
56 3.50 71.4 1.8 224.2 104.3 14.1 9.0 0.83 5.5 3.0 1.0 1.5 2.1
13 3.49 72.7 2.0 253.2 137.8 19.6 6.7 0.68 8.3 2.4 1.9 2.8 2.2
42 3.49 68.2 2.2 229.0 105.4 8.8 1.7 0.81 11.2 2.5 1.3 0.9 2.2
69 3.48 71.7 1.6 191.4 91.6 6.2 20.8 0.72 7.3 2.6 1.0 1.0 2.4
18 3.45 71.0 1.7 249.4 133.7 1.9 6.6 0.86 6.4 2.3 1.4 1.9 2.3
39 3.41 71.4 0.7 251.1 132.2 6.0 0.7 0.87 8.2 3.3 1.5 1.0 1.7
50 3.41 70.7 1.5 232.2 116.2 14.7 3.3 0.69 12.7 3.1 1.3 1.6 2.2
14 3.28 75.6 2.6 257.7 142.8 7.0 11.4 0.74 6.1 1.6 1.2 1.8 1.9
37 3.07 68.4 0.4 222.6 107.8 17.3 14.5 0.70 9.0 1.7 1.1 1.8 2.2
73 2.91 72.3 4.2 250.8 136.2 20.7 2.9 0.55 13.2 2.7 1.2 2.0 2.3
12 2.79 73.7 2.4 251.5 134.7 16.2 4.7 0.70 7.1 2.9 1.0 2.1 2.2
29 2.37 71.2 3.3 238.3 127.3 18.6 21.5 0.74 9.8 2.8 1.3 2.7 2.0
17 2.29 74.4 2.8 239.8 124.9 16.3 11.9 0.67 5.9 1.9 1.4 1.2 2.0

Mean 4.07 70.6 1.4 238.5 122.4 10.3 6.1 0.84 7.6 2.5 1.1 1.8 2.1
LSD 0.68 0.9 0.9 10.1 8.0 9.2 11.0 0.11 4.4 0.9 0.5 0.9 0.4
MSE 1.04 2.9 2.1 285.0 163.4 129.1 60.6 0.02 35.0 0.5 0.1 0.2 0.1
Min 2.29 66.6 -0.4 191.4 91.6 0.7 -0.6 0.55 3.8 1.3 0.9 0.9 1.7
Max 5.62 75.6 4.2 267.9 147.0 24.7 21.5 1.00 15.5 4.3 1.9 3.0 2.5
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging; SL=stem lodging; EPP=ears per plant; ER=ear rot; GLS=grey leaf spot; RUST=common rust; ET=leaf
blight turcicum; SEN=senescence LSD=least significant difference; MSE=mean square error; Min=minimum; Max=maximum.
271
Appendix 3 Performance of genotypes for grain yield and other agronomic traits across optimum sites in the 2009/10 and
2010/11 seasons
ENTRY GYD AD ASI PH EH RL SL EPP HC ER GLS RUST ET
t ha-1 d d cm cm % % # % % 1-5 1-5 1-5
74 7.86 73.1 0.6 296.5 148.5 4.9 5.0 0.97 22.4 4.5 1.5 1.3 2.2
68 6.79 67.8 0.1 257.1 120.6 10.0 0.6 0.95 5.2 7.3 1.8 1.0 1.5
57 6.63 70.0 0.5 262.5 142.5 11.1 6.7 1.00 11.1 2.7 1.6 1.0 2.6
48 6.63 69.2 0.5 291.1 151.6 6.0 3.7 1.08 8.7 9.2 3.9 1.4 1.1
61 6.62 67.5 -0.1 247.8 120.5 13.1 3.7 0.98 19.3 3.7 2.7 1.0 1.5
7 6.61 67.5 1.0 266.5 122.7 9.8 5.6 0.86 8.8 6.5 1.8 0.9 1.8
79 6.55 71.5 0.5 264.6 141.0 9.0 5.5 0.96 3.3 3.4 1.9 1.0 1.5
5 6.54 69.6 0.7 272.4 142.7 10.8 13.1 1.01 13.1 2.9 3.4 1.1 2.3
51 6.52 67.1 -0.9 265.6 116.9 8.0 7.6 0.98 16.0 7.3 3.6 1.0 1.6
52 6.45 68.5 0.2 261.9 125.6 6.0 1.8 0.95 24.8 4.4 3.1 1.3 2.0
45 6.34 67.3 0.6 259.6 124.8 5.7 7.9 0.94 22.4 1.4 2.7 1.0 1.5
4 6.30 67.5 0.8 233.3 107.2 17.2 3.5 1.07 15.8 2.6 2.7 1.3 2.1
60 6.17 64.3 -0.5 240.8 118.4 9.7 3.1 0.98 17.4 4.8 2.7 0.9 1.5
72 6.09 67.3 -0.4 247.8 116.9 8.4 3.1 1.02 21.5 3.0 1.5 1.0 1.5
47 6.06 68.5 0.6 267.1 123.6 2.7 7.6 0.96 21.5 7.7 4.3 1.5 2.1
10 6.05 66.3 0.2 262.1 122.6 4.3 2.5 0.91 14.6 7.8 3.3 1.0 2.2
8 6.05 67.4 1.0 257.7 131.4 8.1 7.0 0.96 6.3 3.0 4.2 0.9 1.7
38 6.02 66.0 1.8 264.1 119.9 6.4 7.1 0.92 18.3 3.9 3.4 1.1 2.0
49 6.00 68.6 0.7 265.1 131.2 7.7 9.1 1.16 17.8 5.6 3.4 1.0 1.9
54 5.98 67.3 1.4 259.7 137.6 8.7 4.6 0.95 6.3 1.2 2.1 0.9 1.8
77 5.95 69.5 0.6 257.4 124.4 12.3 -0.6 0.84 16.3 9.1 1.9 0.9 2.2
28 5.94 64.7 0.9 248.7 117.1 8.7 5.7 0.93 3.0 3.1 1.7 1.7 3.0
26 5.91 66.9 0.1 253.9 126.3 0.5 3.2 1.01 21.5 4.5 1.9 0.9 1.6
70 5.89 69.5 1.6 273.4 131.2 2.1 4.8 0.93 18.7 3.1 1.6 1.0 1.1
34 5.89 63.9 -0.4 233.9 108.2 8.1 15.7 0.99 15.7 5.9 3.0 1.0 2.1
58 5.87 66.1 0.7 243.3 117.6 10.1 5.7 1.03 15.2 1.9 2.2 1.0 1.9
62 5.81 64.5 -0.3 240.5 124.7 11.3 2.7 1.03 15.1 6.3 1.4 0.9 1.1
71 5.78 66.0 0.1 246.8 117.0 6.0 1.9 0.96 8.5 6.2 2.3 0.9 1.1
9 5.74 66.6 1.0 264.6 127.6 11.4 9.8 0.96 15.5 5.2 3.3 1.0 1.7
76 5.73 66.0 0.8 264.1 125.0 11.3 0.9 1.18 11.2 7.5 2.4 1.2 1.5
43 5.73 69.4 1.5 259.2 127.8 0.6 4.4 0.97 29.4 3.7 1.7 1.4 1.1
6 5.65 68.9 0.7 275.3 131.1 9.1 1.5 0.96 12.8 4.8 3.7 1.0 2.0
1 5.64 69.0 1.3 281.0 139.9 5.4 5.8 0.92 4.3 3.8 2.1 1.6 2.0
27 5.64 68.0 -1.1 260.5 127.1 10.9 3.3 0.89 10.3 4.1 2.5 1.4 2.0
80 5.56 67.0 0.3 250.6 117.2 5.7 2.6 0.93 4.2 1.9 1.7 1.0 2.0
67 5.55 70.6 0.8 282.6 136.2 11.2 1.1 1.04 15.4 7.1 2.3 1.1 1.9
63 5.53 67.6 0.2 251.2 122.5 25.2 6.7 0.97 5.4 2.4 1.3 1.0 1.6
40 5.52 68.7 1.3 270.5 129.5 17.0 3.3 0.91 25.1 6.1 2.3 0.9 1.0
16 5.52 69.1 0.1 266.2 128.1 6.9 14.5 0.98 14.2 4.8 2.6 0.9 2.1
59 5.50 66.5 -0.1 238.8 122.3 -0.5 8.4 0.85 19.5 4.5 1.4 0.9 2.3
78 5.48 69.1 0.9 263.8 121.3 4.9 4.9 0.94 17.5 4.1 2.0 1.0 2.5
31 5.44 68.8 0.4 268.7 136.2 17.0 5.1 0.89 25.6 9.9 2.5 1.5 2.2
11 5.40 66.9 1.6 264.3 134.6 30.7 16.9 0.87 9.7 3.5 1.8 1.3 1.6

272
 Appendix 3 Performance of genotypes for grain yield and other agronomic traits across optimum sites in the 2009/10 and
2010/11 seasons
ENTRY GYD AD ASI PH EH RL SL EPP HC ER GLS RUST ET
t/ha d d cm cm % % # % % 1-5 1-5 1-5
3 5.38 67.5 1.4 268.2 124.9 13.8 7.7 0.94 5.3 5.7 3.9 1.2 1.5
53 5.28 67.2 -0.8 264.3 124.4 1.5 13.0 1.12 34.0 9.1 4.1 0.9 2.0
44 5.28 65.6 0.9 267.9 118.8 9.0 3.9 0.87 18.3 2.5 2.4 1.1 1.6
32 5.26 65.9 -0.4 250.5 107.9 11.4 2.2 0.94 12.0 4.8 3.2 1.1 2.7
66 5.22 69.4 1.4 264.0 119.7 4.9 0.1 0.89 9.3 5.0 1.8 1.0 2.0
46 5.17 70.0 1.0 281.5 142.7 13.3 11.0 0.93 17.7 7.4 2.8 1.1 2.4
75 5.09 69.3 0.7 272.1 127.0 10.6 8.1 1.08 3.4 3.7 2.3 1.5 1.3
65 5.05 72.0 2.3 281.2 143.0 14.6 5.8 0.90 7.0 2.7 2.0 1.1 2.5
15 5.03 67.4 0.1 258.8 129.7 6.3 0.4 0.97 9.5 5.7 2.0 1.1 1.7
42 4.99 65.2 0.8 250.8 109.8 11.1 1.7 0.94 18.1 2.8 2.5 1.3 0.9
35 4.98 66.6 0.0 246.6 108.8 16.9 6.2 0.93 14.6 3.8 2.3 1.2 1.9
2 4.96 68.7 1.9 278.4 140.9 18.5 9.9 0.81 7.1 9.9 3.7 1.6 2.7
36 4.91 63.8 -0.2 249.3 116.5 8.7 4.1 0.94 14.2 12.0 2.2 1.2 2.0
64 4.86 72.9 1.1 277.1 142.4 3.2 5.9 1.01 5.9 7.3 1.4 1.2 1.4
19 4.82 65.2 0.7 249.7 114.3 15.5 4.5 0.93 8.8 5.6 3.7 1.6 2.1
39 4.81 68.1 0.3 276.4 132.9 6.3 0.7 0.90 15.7 8.1 3.3 1.5 1.0
14 4.79 72.7 2.1 285.6 156.3 13.9 11.4 0.88 3.5 5.7 1.6 1.2 1.8
33 4.78 65.4 -1.0 237.0 105.9 8.8 4.0 0.85 7.6 6.2 2.2 1.0 2.8
41 4.77 68.4 1.4 268.2 125.0 7.9 8.5 0.85 9.1 8.9 2.9 0.9 1.8
21 4.75 65.5 0.3 252.7 115.8 7.3 4.0 0.84 2.9 5.0 1.3 1.0 1.4
50 4.73 67.2 0.6 255.8 118.9 7.6 3.3 0.85 16.6 7.7 3.1 1.3 1.6
69 4.71 68.7 0.3 209.7 96.2 8.3 20.8 0.87 3.5 4.3 2.6 1.0 1.0
24 4.71 63.9 0.1 248.8 119.3 3.9 4.8 0.96 4.5 4.3 2.7 1.0 1.7
20 4.64 66.6 0.9 249.8 122.8 7.7 2.2 0.88 19.4 4.9 2.4 0.9 1.4
55 4.63 67.4 0.2 254.1 131.5 6.3 2.9 0.97 13.0 2.2 2.6 1.0 2.2
56 4.56 68.0 1.0 242.5 110.2 3.6 9.0 0.97 3.5 3.4 3.0 1.0 1.5
25 4.52 65.7 -0.3 250.8 111.3 16.5 13.3 0.82 4.2 3.3 2.0 1.0 2.2
13 4.49 70.2 1.6 276.6 136.1 20.3 6.7 0.85 5.8 4.8 2.4 1.9 2.8
23 4.46 63.8 -0.1 249.9 119.5 4.5 9.0 0.92 16.1 12.9 2.1 1.3 2.3
30 4.44 65.4 -0.1 246.1 110.1 6.9 2.0 0.86 16.1 5.9 3.8 1.2 2.5
22 4.33 64.3 0.1 250.0 120.2 12.1 4.4 0.92 16.5 3.0 3.0 1.0 1.0
73 4.26 68.9 3.0 278.2 142.7 23.7 2.9 0.66 9.0 7.7 2.7 1.2 2.0
18 4.17 68.8 0.6 270.5 134.1 -1.1 6.6 0.98 22.1 6.8 2.3 1.4 1.9
37 3.67 65.3 0.1 240.9 111.6 25.2 14.5 0.87 7.9 6.7 1.7 1.1 1.8
12 3.54 70.4 0.9 282.9 137.1 17.3 4.7 0.78 3.7 4.1 2.9 1.0 2.1
17 3.10 71.2 1.6 270.4 125.2 14.0 11.9 0.90 19.5 8.2 1.9 1.4 1.2
29 2.91 67.8 2.1 257.7 129.1 17.9 21.5 0.81 7.0 6.1 2.8 1.3 2.7

Mean 5.39 67.7 0.6 260.2 125.5 9.9 6.1 0.94 13.0 5.3 2.5 1.1 1.8
LSD 0.99 1.2 1.0 14.5 9.4 14.0 11.0 0.12 9.6 4.2 0.9 0.5 0.9
MSE 1.24 2.1 1.4 263.7 112.6 147.6 60.6 0.01 92.1 13.4 0.5 0.1 0.2
Min 2.91 63.8 -1.1 209.7 96.2 0.46 -1.1 -0.6 0.66 1.2 1.3 0.9 0.9
Max 7.86 73.1 3.0 296.5 156.3 0.59 30.7 21.5 1.18 12.9 4.3 1.9 3.0
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging; SL=stem lodging; EPP=ears per plant; HC=husk cover; ER=ear rot; GLS=grey leaf spot;
RUST=common rust; ET=leaf blight turcicum; LSD=least significant difference; MSE=mean square error; Min=minimum; Max=maximum.

273
Appendix 4 Performance of genotypes for grain yield and other agronomic traits
across managed drought sites in the 2009/10 and 2010/11 seasons
ENTRY GYD AD ASI PH EH RL EPP SEN
t ha-1 d d cm cm % # 1-10
52 3.26 98.0 1.1 236.9 121.6 5.5 0.95 3.7
27 3.07 95.8 0.9 216.6 100.7 0.2 1.07 3.3
19 3.01 91.7 1.9 230.5 105.7 0.7 0.90 3.2
38 3.01 94.6 3.0 222.9 110.3 7.5 0.92 2.9
59 2.84 94.8 -0.1 210.4 102.3 3.5 0.91 3.8
36 2.79 90.6 -0.7 215.9 98.2 0.6 0.96 3.6
61 2.79 94.1 1.9 230.1 114.4 2.3 0.97 3.6
66 2.77 98.7 4.4 220.8 105.7 5.2 0.84 3.9
3 2.74 97.1 3.0 228.5 119.4 31.4 0.79 3.7
30 2.71 93.7 1.3 226.4 112.9 2.1 0.96 3.5
55 2.65 95.9 0.5 221.1 111.9 0.5 0.99 3.1
34 2.64 90.2 -0.4 195.6 83.5 3.4 1.08 3.8
63 2.62 92.5 0.9 201.7 100.4 -0.7 0.92 3.7
35 2.55 93.6 1.3 220.3 102.4 -0.5 1.07 4.5
60 2.51 89.4 0.8 196.0 93.3 -0.6 0.97 3.4
2 2.50 96.0 0.1 223.6 117.3 0.4 0.59 3.7
74 2.50 104.8 1.9 236.7 108.8 -0.5 0.42 3.3
79 2.49 104.7 0.7 233.5 121.2 7.8 0.72 3.1
11 2.47 95.6 -0.4 231.0 108.4 12.6 0.85 3.9
62 2.46 92.5 -0.7 215.3 116.2 -0.8 1.06 3.3
16 2.45 97.6 1.6 232.5 125.1 2.0 0.94 3.5
33 2.44 92.7 0.2 220.7 111.3 -1.8 0.99 3.4
58 2.43 92.7 2.8 206.7 96.7 -0.7 0.85 3.2
10 2.41 95.0 1.7 224.4 114.2 0.1 0.81 3.5
22 2.39 88.4 1.0 224.1 108.0 2.7 0.86 3.9
70 2.38 100.0 2.5 228.1 116.4 1.0 0.59 3.3
29 2.35 97.8 4.4 233.4 115.6 28.0 0.85 3.4
54 2.33 94.9 1.5 227.4 117.0 0.6 0.82 3.4
72 2.31 95.2 -2.8 211.7 100.7 13.6 1.04 3.6
40 2.28 98.8 2.6 222.0 97.1 -1.8 0.88 3.4
28 2.28 94.5 0.1 227.1 118.4 -1.4 0.94 3.7
32 2.24 93.3 2.9 222.0 97.4 -1.9 0.90 4.0
24 2.23 88.9 1.2 215.5 99.6 0.5 0.88 4.0
77 2.21 101.0 1.9 216.8 89.0 18.5 0.81 3.6
23 2.15 89.3 1.1 218.2 108.0 4.8 0.95 3.6
18 2.14 94.9 2.9 230.3 115.7 -1.6 0.76 4.0
31 2.09 98.7 1.0 219.6 108.4 1.8 0.86 3.4
21 2.05 93.5 2.1 220.8 108.1 7.4 0.81 3.6
9 2.04 96.0 1.3 224.7 119.7 0.1 0.73 4.0
25 2.04 90.0 0.9 216.8 96.3 3.3 1.06 4.1
41 2.04 98.5 1.1 227.8 120.1 1.3 0.62 3.2
26 2.04 95.9 2.1 213.0 93.5 12.5 0.90 3.5
5 2.03 101.5 2.4 241.3 129.8 10.7 0.38 3.4
67 2.03 100.7 2.3 238.7 117.4 0.6 0.73 4.0
48 1.99 102.5 0.8 260.3 148.3 6.5 0.97 3.1
76 1.99 93.3 2.3 217.5 101.3 7.8 0.85 3.4

274
 Appendix 4 Performance of genotypes for grain yield and other agronomic traits across
managed drought sites in the 2009/10 and 2010/11 seasons

ENTRY GYD AD ASI PH EH RL EPP SEN

t ha-1 d d cm cm % # 1-10
46 1.99 100.9 2.0 239.9 114.6 4.7 0.86 4.0
6 1.98 101.4 1.9 235.5 126.3 7.6 0.49 3.8
15 1.95 94.7 1.7 226.4 107.4 2.9 0.92 4.2
47 1.95 97.5 1.7 230.7 114.0 4.3 0.69 3.1
75 1.94 97.7 4.6 232.8 105.1 0.6 0.66 3.5
45 1.92 96.9 1.3 230.7 116.7 3.3 0.73 3.5
44 1.91 92.7 2.8 214.9 107.2 -0.8 0.73 3.3
37 1.91 90.6 -2.4 206.5 97.6 4.9 0.68 3.7
12 1.87 97.3 3.8 229.6 114.2 9.5 0.55 3.8
39 1.84 99.1 -0.5 227.3 119.1 -0.4 1.49 2.9
4 1.83 95.2 1.7 209.9 91.7 1.3 0.63 3.7
1 1.83 99.8 3.0 230.2 121.2 -1.0 1.28 3.5
20 1.80 94.6 1.4 216.9 109.0 -0.9 0.93 4.0
50 1.79 96.5 2.1 228.9 114.5 16.0 0.58 3.8
49 1.78 101.4 1.4 234.8 117.3 10.9 0.40 3.8
78 1.78 101.1 1.5 238.0 119.0 3.2 0.51 3.4
57 1.78 97.9 2.1 212.5 101.4 1.2 0.72 3.2
42 1.77 91.0 3.8 200.5 82.7 -2.6 0.78 3.9
7 1.71 96.4 3.7 225.5 113.6 16.5 0.36 3.5
8 1.71 95.4 2.8 220.5 104.0 19.5 0.69 3.9
53 1.69 96.7 0.1 234.5 133.3 0.4 0.94 3.7
64 1.63 104.1 2.2 233.5 122.0 -0.2 0.57 3.6
71 1.47 94.2 0.5 231.5 94.5 0.4 0.88 3.8
80 1.42 96.1 1.7 214.9 100.2 1.4 0.67 3.0
17 1.40 102.3 4.5 214.2 115.9 19.4 0.30 3.5
68 1.40 96.8 4.0 219.3 102.1 -1.1 0.76 4.1
65 1.36 102.8 3.9 238.7 120.6 0.4 0.19 3.8
43 1.36 97.7 3.6 232.4 111.7 5.6 0.67 3.9
13 1.35 97.7 0.1 238.2 112.6 4.5 0.10 3.8
14 1.18 104.0 1.2 237.9 114.1 6.4 0.43 3.3
51 1.17 97.2 1.0 227.6 104.3 9.5 0.68 3.5
73 1.15 99.2 5.8 247.4 124.2 -0.7 0.21 4.0
56 1.04 99.3 2.8 209.3 96.1 23.3 0.44 3.7
69 0.59 95.6 2.9 166.9 66.4 -3.1 0.30 4.1
Mean 2.09 96.4 1.7 223.7 109.3 4.5 0.77 3.6
LSD 0.92 2.6 2.3 20.9 23.0 15.7 0.45 0.73
MSE 0.42 3.3 2.8 220.0 267.5 62.5 0.05 0.3
Min 0.59 88.4 -2.8 166.9 66.4 -3.1 0.10 2.9
Max 3.26 104.8 5.8 260.3 148.3 31.4 1.49 4.5
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; EH=ear height; RL=root lodging; EPP=ears per plant;
SEN=senescence; LSD=least significant difference; MSE=mean square error; Min=minimum; Max=maximum.

275
Appendix 5 Performance of genotypes for grain yield and other agronomic traits across low nitrogen sites
ENTRY GYD AD ASI PH RL EPP ER
t ha-1 d d cm % # %
76 0.81 71.4 4.7 210.9 2.6 0.76 8.8
16 0.77 73.4 5.9 228.0 1.9 0.56 12.6
3 0.76 71.5 10.3 217.7 -2.0 0.76 10.1
1 0.75 73.4 4.6 222.3 -0.8 0.79 11.7
54 0.73 73.7 4.1 210.4 5.0 0.70 24.6
61 0.73 70.7 2.3 177.7 7.2 0.76 9.3
8 0.72 70.3 6.8 217.1 -1.0 0.78 12.5
78 0.71 74.1 7.9 192.3 6.3 0.65 27.9
57 0.68 73.6 8.6 221.9 -1.7 0.79 16.2
26 0.67 70.4 6.3 239.7 -3.1 0.71 26.0
39 0.67 73.8 4.4 238.9 -1.5 0.67 18.4
75 0.67 74.2 7.7 247.7 -5.8 0.70 15.7
59 0.65 70.6 1.2 192.1 -2.1 0.86 20.0
24 0.64 69.6 4.8 211.4 3.6 0.64 12.5
53 0.62 71.0 5.6 220.7 16.2 0.52 12.8
60 0.61 69.2 5.8 228.7 -2.2 0.94 7.1
34 0.60 69.5 7.6 195.0 0.3 0.79 14.1
47 0.60 73.2 3.4 214.5 -5.4 0.64 11.6
36 0.60 71.4 1.1 215.2 -0.4 0.77 16.7
7 0.59 71.7 8.9 222.6 0.8 0.82 25.1
14 0.59 76.6 10.6 211.8 -1.5 0.57 15.5
77 0.57 71.3 7.6 214.9 6.1 0.56 9.6
80 0.56 74.0 9.1 229.7 -1.9 0.77 11.5
4 0.55 71.4 7.3 204.4 5.9 0.82 27.3
42 0.54 71.4 6.9 205.8 -0.8 0.65 48.8
72 0.54 71.0 5.7 207.7 1.8 0.68 12.9
64 0.54 75.7 4.5 268.8 -1.4 0.62 10.1
55 0.54 72.4 3.3 206.4 1.1 0.87 15.9
62 0.53 71.7 2.9 227.4 -3.8 0.76 14.4
46 0.53 74.4 10.3 264.1 5.3 0.72 15.8
18 0.52 73.3 6.5 228.6 -0.3 0.70 17.4
48 0.52 77.3 8.7 251.5 5.4 0.78 8.8
23 0.51 70.6 2.9 189.8 -2.2 0.79 17.5
65 0.50 77.5 8.7 248.4 3.6 0.67 9.2
27 0.50 74.1 0.6 218.3 -1.0 0.58 8.6
17 0.50 76.7 8.4 200.5 17.9 0.50 15.8
79 0.50 78.5 4.5 228.2 -6.0 0.71 9.3
28 0.50 71.1 4.0 221.0 8.3 0.78 13.4
40 0.50 75.0 7.3 207.6 4.6 0.55 32.9
71 0.50 72.7 4.1 218.9 -0.3 0.70 9.8
68 0.49 72.7 8.1 226.9 7.9 0.73 9.0
21 0.49 70.1 6.4 204.9 10.6 0.71 7.4
19 0.48 71.3 3.9 198.0 12.3 0.68 10.9
58 0.48 73.0 6.9 223.0 -3.0 0.80 8.8
41 0.47 72.8 7.0 212.0 19.8 0.56 22.4
15 0.46 73.7 3.7 213.3 1.5 0.66 12.6

276
 Appendix 5 Performance of genotypes for grain yield and other agronomic traits across low nitrogen sites

ENTRY GYD AD ASI PH RL EPP ER

t ha-1 d d cm % # %
74 0.46 75.1 9.8 266.9 5.3 0.68 16.3
66 0.46 71.7 8.6 205.1 9.6 0.68 14.2
13 0.43 74.8 9.7 212.9 35.7 0.54 19.6
20 0.43 70.7 6.7 217.9 33.3 0.62 29.9
63 0.42 70.7 8.0 191.5 25.4 0.61 13.9
38 0.41 70.8 6.2 207.2 2.7 0.77 14.4
35 0.39 70.6 10.0 221.1 1.0 0.65 19.7
49 0.39 73.7 8.5 240.8 -4.1 0.94 21.9
22 0.39 69.6 3.8 229.8 -5.5 0.76 13.9
67 0.38 77.2 7.7 229.4 -3.6 0.65 10.5
32 0.38 73.5 4.7 216.6 15.7 0.61 18.0
51 0.38 72.0 12.4 185.9 2.2 0.68 12.0
31 0.37 74.2 9.1 216.3 -2.3 0.63 22.3
52 0.37 71.9 8.3 238.2 1.7 0.74 47.3
29 0.36 75.2 9.7 203.9 3.1 0.57 23.7
12 0.36 77.0 10.3 219.5 6.0 0.68 14.5
9 0.35 74.5 7.8 204.3 6.4 0.70 29.0
44 0.35 71.7 9.6 228.1 0.3 0.62 14.9
6 0.34 70.9 10.2 246.3 -6.7 0.75 43.3
5 0.31 76.2 9.2 228.8 20.8 0.61 21.3
10 0.30 72.8 8.7 207.1 1.8 0.56 17.9
33 0.27 71.6 8.3 186.5 9.7 0.42 29.0
43 0.27 72.9 9.7 234.9 13.1 0.57 43.4
56 0.26 73.0 7.9 210.2 22.6 0.72 15.0
30 0.26 69.7 8.8 185.1 7.1 0.86 26.8
70 0.22 77.2 17.6 205.8 10.4 0.36 43.9
73 0.22 74.8 12.2 215.0 15.2 0.59 29.6
45 0.21 71.4 11.4 220.8 6.2 0.66 14.4
2 0.20 74.6 8.6 184.9 4.7 0.44 59.9
50 0.16 74.4 7.8 207.6 6.5 0.47 38.7
37 0.15 73.7 9.5 185.7 8.3 0.38 30.6
25 0.14 72.5 12.4 205.3 -0.3 0.44 21.3
11 0.10 71.9 12.6 234.5 31.6 0.67 16.8
69 0.08 75.5 8.8 208.3 6.0 0.51 24.3
Mean 0.47 72.9 7.3 217.0 5.0 0.67 19.3
LSD 0.34 3.9 5.0 46.6 22.3 0.22 21.8
MSE 0.03 7.5 6.3 546.9 125.0 0.02 119.6
DF 79
GYD=grain yield; AD=anthesis days; ASI=anthesis silking interval; PH=plant height; RL=root lodging; EPP=ears per plant; ER=ear rot; LSD=least significant difference; MSE=mean square error; DF=degrees
of freedom.

277
Appendix 6 Line general combining ability effects for grain yield across different
environments

Line Across Optimum Drought Low N

K64r -0.120 -0.377 0.372 -0.074
N3.2.3.3 0.172 0.593 -0.076 0.141
RS61P 0.717 0.635 0.369 0.582
NAW5885 0.131 0.355 0.119 0.213
2N3d -0.900 -0.924 -0.215 -0.702
2Kba -0.267 -0.699 -0.152 -0.122
SC5522 -0.953 -1.319 -0.537 -0.639
RA214P 0.180 0.373 -0.264 -0.208
LSD 0.007 0.024 0.013 0.08
LSD=least significant difference; Low N=low nitrogen.

Appendix 7 Tester general combining ability effects for grain yield across different
environments
Tester Across Optimum Drought Low N
1 -0.92 -1.19 -0.12 -0.68
2 -0.64 -0.97 0.28 -0.46
3 0.16 0.22 -0.13 0.30
4 0.19 -0.02 0.59 0.27
5 0.24 0.16 -0.04 0.18
6 0.07 0.02 -0.03 -0.21
7 0.01 0.10 -0.16 -0.07
8 0.13 0.25 -0.10 -0.01
9 -0.11 0.01 -0.42 0.11
10 -0.84 -1.34 0.27 -0.64
11 0.34 0.48 0.39 0.09
12 -0.92 -1.10 0.01 -0.64
LSD 0.007 0.024 0.013 0.08
LSD=least significant difference; Low N=low nitrogen.

278
Appendix 8 Mean grain yield (t ha-1) for 80 genotypes across seven environments
E1 E2 E3 E4 E5 E6 E7 Mean
Genotype t ha-1
G1 7.22 1.44 4.69 1.83 4.40 2.63 3.31 3.65
G2 6.40 1.39 3.98 2.50 4.00 3.05 3.50 3.55
G3 7.18 1.72 4.01 2.74 4.04 2.05 3.35 3.58
G4 8.37 2.36 5.32 1.83 3.71 4.00 2.95 4.08
G5 8.23 1.67 5.27 2.03 4.73 3.14 2.92 4.00
G6 7.09 1.96 4.07 1.98 4.77 3.78 4.07 3.96
G7 7.82 1.66 5.24 1.71 5.23 2.73 3.31 3.96
G8 7.25 2.01 5.60 1.71 4.51 3.15 4.31 4.08
G9 6.04 1.51 5.49 2.04 4.38 2.80 2.64 3.56
G10 7.49 1.37 5.00 2.41 4.69 3.31 4.70 4.14
G11 7.32 1.37 3.55 2.47 4.18 3.56 3.54 3.71
G12 5.25 1.44 3.06 1.87 2.48 3.05 2.88 2.86
G13 5.33 1.59 3.45 1.35 3.92 3.78 4.12 3.36
G14 5.61 1.74 3.85 1.18 4.39 3.73 3.09 3.37
G15 6.74 1.48 3.94 1.95 4.06 2.72 5.09 3.71
G16 6.78 1.14 4.58 2.45 4.08 2.55 3.40 3.57
G17 4.42 0.60 2.33 1.40 2.97 2.10 2.34 2.31
G18 6.05 1.50 2.63 2.14 3.37 3.13 4.27 3.30
G19 5.53 1.36 4.52 3.01 3.71 2.99 4.09 3.60
G20 5.95 2.02 3.25 1.80 4.13 3.27 5.96 3.77
G21 5.71 1.19 4.33 2.05 3.37 3.27 4.54 3.49
G22 5.91 1.87 3.44 2.39 3.27 3.70 5.83 3.77
G23 5.75 1.00 4.13 2.15 2.95 3.11 4.11 3.31
G24 5.65 1.22 4.35 2.23 3.55 3.35 3.16 3.36
G25 4.95 1.04 4.80 2.04 2.85 3.40 3.81 3.27
G26 8.31 1.73 4.84 2.04 3.84 3.47 3.73 3.99
G27 7.50 1.57 4.59 3.07 3.70 3.20 3.07 3.82
G28 7.11 1.56 5.54 2.28 3.89 3.23 3.26 3.84
G29 4.19 0.89 2.44 2.35 2.70 1.98 2.21 2.39
G30 4.99 2.05 4.63 2.71 3.09 3.02 4.44 3.56
G31 6.83 1.51 4.77 2.09 3.70 2.87 4.71 3.78
G32 6.58 1.65 4.73 2.24 3.41 3.62 4.32 3.79
G33 5.19 0.95 4.75 2.44 3.20 3.71 4.18 3.49
G34 7.07 1.68 5.98 2.64 3.60 3.65 4.30 4.13
G35 5.35 1.74 5.38 2.55 3.54 2.64 4.55 3.68
G36 6.09 1.35 4.57 2.79 3.54 2.93 4.01 3.61
G37 4.74 0.64 3.24 1.91 2.57 2.90 3.59 2.80
G38 6.99 1.88 5.82 3.01 3.98 4.16 3.89 4.25
G39 6.98 1.89 3.10 1.84 3.29 2.11 2.53 3.11
G40 7.27 1.10 4.39 2.28 3.78 2.40 2.62 3.41
G41 6.44 1.06 3.10 2.04 4.38 3.08 3.54 3.38
G42 5.75 1.50 4.68 1.77 4.15 2.58 2.86 3.33
G43 7.83 1.62 4.69 1.36 3.36 2.84 2.40 3.44
G44 6.45 1.51 4.52 1.91 3.93 3.26 3.04 3.52
G45 7.51 1.97 6.03 1.92 4.21 3.00 4.59 4.18
G46 6.35 2.40 3.71 1.99 4.42 2.99 3.89 3.68
G47 7.39 1.28 5.11 1.95 4.58 3.47 3.97 3.96
G48 8.37 1.63 4.60 1.99 5.39 3.23 4.61 4.26
G49 7.47 2.12 4.53 1.78 4.88 1.91 2.48 3.60

279
Appendix 8 Mean grain yield (t ha-1) for 80 genotypes across seven environments

E1 E2 E3 E4 E5 E6 E7 Mean
Genotype t ha-1
G50 5.50 1.29 4.53 1.79 3.43 2.75 2.80 3.16
G51 7.64 1.38 6.02 1.17 4.70 3.16 3.82 3.99
G52 8.91 2.35 5.04 3.26 4.04 3.06 3.26 4.27
G53 7.41 0.88 3.73 1.69 4.01 2.35 3.62 3.38
G54 7.72 1.54 4.26 2.33 4.90 3.63 6.12 4.36
G55 5.76 1.77 3.96 2.65 3.61 3.29 5.55 3.80
G56 5.60 1.82 3.44 1.04 3.99 2.46 4.31 3.24
G57 8.20 2.44 6.42 1.78 4.29 4.26 5.19 4.66
G58 7.22 1.96 5.51 2.43 3.89 3.93 5.02 4.28
G59 6.78 2.08 5.00 2.84 3.89 3.35 5.15 4.16
G60 6.69 2.83 6.37 2.51 4.18 3.91 4.31 4.40
G61 8.33 2.31 6.68 2.79 3.65 3.65 5.76 4.74
G62 7.84 1.49 4.67 2.46 4.33 3.53 4.18 4.07
G63 6.87 1.30 5.40 2.62 3.23 4.08 6.51 4.29
G64 8.10 1.28 2.59 1.63 3.39 2.94 3.75 3.38
G65 6.42 2.02 4.00 1.36 4.26 3.09 4.53 3.67
G66 6.68 1.31 3.95 2.77 4.13 3.70 3.30 3.69
G67 6.94 1.19 4.55 2.03 4.44 4.45 4.23 3.98
G68 9.99 1.70 4.82 1.40 4.22 3.77 3.72 4.23
G69 5.90 0.94 3.84 0.79 3.19 3.18 3.86 3.10
G70 8.02 1.03 4.62 2.38 4.05 3.80 4.90 4.11
G71 7.56 1.22 4.31 1.47 4.43 3.73 4.01 3.82
G72 7.46 1.51 5.66 2.31 4.18 3.93 4.16 4.17
G73 5.50 1.29 2.91 1.15 3.46 2.00 2.24 2.65
G74 10.43 4.08 5.29 2.50 6.16 2.50 4.85 5.12
G75 7.96 1.64 3.10 1.94 4.21 3.43 3.56 3.69
G76 7.25 2.28 4.51 1.99 4.56 3.16 2.53 3.75
G77 6.83 2.11 5.13 2.21 4.31 2.79 3.08 3.78
G78 6.96 1.94 4.30 1.78 4.20 3.33 5.12 3.95
G79 8.39 1.28 5.51 2.49 4.51 3.82 3.56 4.22
G80 6.93 2.02 4.56 1.42 4.12 2.30 4.00 3.62
MEAN 6.83 1.62 4.50 2.09 3.96 3.17 3.92
E1=Agricultural Research Trust farm; E2=Harare low N; E3=Kadoma; E4=Chiredzi winter; E5=Rattray Arnold Research Station;
E6=Chiredzi summer; E7=Chisumbanje; Underlined and bold values= highest yielder in the given environment.

280
Appendix 9 Minor allele frequency and corresponding number of single nucleotide
polymorphism markers
Minor allele frequency Number of SNPs
0.50 30
0.48 27
0.47 18
0.46 2
0.45 46
0.44 9
0.43 32
0.42 11
0.41 13
0.40 15
0.39 20
0.38 19
0.37 11
0.36 27
0.35 14
0.34 17
0.33 33
0.32 15
0.31 15
0.30 22
0.29 22
0.28 19
0.27 13
0.26 21
0.25 24
0.24 40
0.23 22
0.22 4
0.21 21
0.20 27
0.19 11
0.18 42
0.17 32
0.16 20
0.15 18
0.14 33
0.13 29
0.12 16
0.11 33
0.10 34
0.09 33
0.08 16
0.07 43
0.06 3
0.05 50
0.04 30
0.03 13
0.02 39
0.00 25
Total 1129
SNP=single nucleotide polymorphism.

281
Appendix 10 Polymorphic information content values and corresponding number of
single nucleotide polymorphism markers
PIC Number of SNPs
0.38 30
0.37 164
0.36 69
0.35 71
0.34 37
0.33 32
0.32 47
0.31 21
0.30 64
0.29 24
0.28 20
0.27 29
0.26 16
0.25 44
0.24 25
0.23 20
0.22 18
0.21 33
0.20 14
0.19 31
0.18 13
0.17 20
0.16 34
0.15 33
0.13 16
0.12 32
0.11 11
0.10 3
0.09 33
0.08 47
0.06 1
0.05 19
0.04 32
0.00 26
Total 1129
PIC=polymorphic information content; SNP=single nucleotide polymorphism.

282
Appendix 11 F1 mean grain yield (t ha-1), specific combining ability, mid- and high-
parent heterosis and genetic distance under optimum conditions

Hybrid GYD MPH HPH GD SCA

L2/T10 5.64 161.86 138.10 0.35 0.34

L2/T10 4.96 92.97 54.98 0.35 0.38
L2/T4 5.38 121.69 84.74 0.36 0.94
L2/T9 6.54 187.98 151.43 0.33 -0.33
L2/T5 5.65 121.51 78.75 0.35 0.38
L2/T8 6.61 168.56 121.69 0.34 -0.97
L2/T3 6.05 170.08 138.18 0.36 0.13
L2/T11 5.74 97.42 48.19 0.36 -0.37
L2/T7 6.05 143.13 99.14 0.33 0.62
L2/T6 5.40 143.84 116.91 0.32 0.42
L7/T10 3.54 136.19 49.48 0.36 -0.05
L7/T1 4.49 134.68 40.44 0.34 0.34
L7/T9 4.79 196.74 84.32 0.28 -0.28
L7/T7 5.03 174.04 65.41 0.24 0.16
L7/T6 5.52 253.53 121.49 0.31 -0.07
L6/T9 3.10 49.51 19.32 0.33 -0.35
L6/T1 4.17 75.63 30.35 0.34 -0.55
L6/T4 4.82 115.98 65.51 0.35 0.37
L6/T5 4.64 97.21 46.97 0.35 0.51
L6/T8 4.75 109.70 59.39 0.34 0.67
L6/T3 4.33 111.91 70.61 0.35 -0.15
L6/T7 4.46 94.35 46.72 0.36 -0.48
L6/T12 4.71 133.05 89.06 0.33 -0.13
L6/T6 4.52 123.93 81.66 0.34 -0.01
L1/T2 5.91 194.14 162.77 0.32 -1.10
L1/T10 5.64 172.30 137.83 0.37 -0.05
L1/T12 5.94 178.73 138.43 0.33 -0.14
L1/T1 2.91 16.91 -9.21 0.23 0.12
L1/T4 4.44 89.80 52.62 0.27 -0.49
L1/T9 5.44 148.92 109.19 0.33 0.99
L1/T5 5.26 113.26 66.36 0.31 0.12
L1/T8 4.78 101.13 60.30 0.31 -0.44
L1/T3 5.89 173.19 131.78 0.33 0.57
L1/T11 4.98 76.43 28.56 0.30 -0.18
L1/T7 4.91 104.06 61.43 0.36 0.70
L1/T6 3.67 72.36 47.44 0.33 -0.26
L4/T4 6.02 163.02 106.98 0.32 -0.57
L4/T9 4.81 125.11 84.85 0.29 -0.68

283
 Appendix 11 F1 mean grain yield (t ha-1), specific combining ability, mid- and high-parent
heterosis and genetic distance under optimum conditions

Hybrid GYD MPH HPH GD SCA

L4/T5 5.52 128.46 74.60 0.34 -0.36

L4/T8 4.77 105.21 60.11 0.31 0.99
L4/T3 4.99 136.99 96.40 0.29 0.35
L4/T11 5.73 175.41 130.06 0.29 0.42
L4/T7 5.28 124.19 73.67 0.32 -0.26
L4/T6 6.34 204.87 154.67 0.29 -0.71
L5/T1 5.17 76.68 61.49 0.36 -0.18
L5/T4 6.06 117.89 108.16 0.34 -0.06
L5/T9 6.63 152.70 155.13 0.29 -0.55
L5/T5 6.00 106.68 90.00 0.20 0.26
L5/T8 4.73 67.99 58.69 0.33 -0.19
L5/T3 6.52 151.37 156.82 0.34 0.47
L5/T11 6.45 97.91 66.71 0.35 0.11
L5/T7 5.28 85.74 73.83 0.33 0.00
L3/T1 5.98 83.45 86.89 0.33 -0.07
L3/T4 4.63 48.63 59.10 0.30 -0.17
L3/T9 6.63 124.14 155.18 0.27 -0.86
L3/T5 5.87 81.27 85.86 0.34 0.72
L3/T8 5.50 74.50 84.45 0.29 0.74
L3/T3 6.17 110.52 142.84 0.30 0.83
L3/T11 6.62 84.23 71.14 0.29 0.25
L3/T7 5.81 82.81 91.22 0.32 0.36
L3/T6 5.53 90.23 121.94 0.27 0.13
L8/T9 4.86 97.79 86.76 0.28 0.84
L8/T1 5.05 83.36 57.86 0.33 -0.30
L8/T4 5.22 99.83 79.23 0.28 -1.05
L8/T5 5.55 102.91 75.62 0.34 -0.13
L8/T8 6.79 156.75 127.89 0.26 -0.19
L8/T3 4.71 94.20 85.41 0.32 0.61
L8/T11 5.89 90.58 52.17 0.32 0.13
L8/T7 5.78 116.00 90.07 0.33 0.76
L8/T6 6.09 153.83 144.65 0.29 -0.54

Mean 5.19 117.39 85.29 0.31 0.03

Min 2.91 16.91 -9.21 0.20 -1.10
Max 6.79 253.53 162.77 0.37 0.99
GYD=grain yield; MPH=mid-parent heterosis; HPH=high-parent heterosis; GD=genetic distance; SCA=specific combining ability;
Min=minimum; Max=maximum.

284
Appendix 12 F1 mean grain yield (t ha-1), specific combining ability, mid- and high-
parent heterosis and genetic distance under low nitrogen conditions

Hybrid GYD MPH HPH GD SCA

L2/T10 2.50 1.81 37.61 0.35 -0.05

L2/T10 2.68 -4.93 34.05 0.35 0.07
L2/T4 3.06 -1.09 20.05 0.36 -0.11
L2/T9 4.45 12.84 59.92 0.33 0.73
L2/T5 3.76 -5.49 2.74 0.35 0.13
L2/T8 3.02 46.20 106.45 0.34 -0.39
L2/T3 3.66 -10.26 -2.44 0.36 0.18
L2/T11 2.85 14.00 31.63 0.36 -0.30
L2/T7 2.60 -10.25 5.15 0.33 0.26
L2/T6 2.69 -41.13 -49.88 0.32 0.04
L7/T10 2.69 7.22 47.57 0.36 0.09
L7/T1 2.97 132.01 110.26 0.34 -0.01
L7/T9 3.19 95.24 48.38 0.28 -0.50
L7/T7 2.73 54.27 3.01 0.24 0.10
L7/T6 1.90 -12.29 -47.34 0.31 -0.15
L6/T9 0.94 57.33 38.38 0.33 0.43
L6/T1 2.74 -59.76 -69.63 0.34 -0.13
L6/T4 2.47 52.94 36.89 0.35 -0.07
L6/T5 3.83 148.86 155.50 0.35 -0.06
L6/T8 2.14 19.85 -2.95 0.34 0.07
L6/T3 3.55 -8.61 -31.01 0.35 0.06
L6/T7 1.74 62.79 27.66 0.36 0.03
L6/T12 2.13 -48.53 -66.42 0.33 -0.28
L6/T6 2.02 22.20 11.91 0.34 0.66
L1/T2 3.12 273.74 47.49 0.32 -0.55
L1/T10 2.89 207.31 138.11 0.37 -0.41
L1/T12 2.87 189.25 125.97 0.33 -0.45
L1/T1 1.60 119.17 51.11 0.23 -0.28
L1/T4 3.97 17.57 -20.06 0.27 0.26
L1/T9 2.83 142.85 55.71 0.33 0.49
L1/T5 3.12 48.01 -8.81 0.31 0.10
L1/T8 1.76 180.81 107.80 0.31 -0.15
L1/T3 3.06 -8.03 -43.34 0.33 0.05
L1/T11 3.29 75.08 10.21 0.3 0.45
L1/T7 2.41 91.95 21.47 0.36 0.23
L1/T6 1.20 -18.33 -53.49 0.33 -0.56
L4/T4 3.55 15.00 -12.28 0.32 0.06
L4/T9 3.44 104.50 39.14 0.29 -0.16
L4/T5 1.95 70.97 10.85 0.34 0.74

285
Appendix 12 F1 mean grain yield (t ha-1), specific combining ability, mid- and high-parent
heterosis and genetic distance under low nitrogen conditions

Hybrid GYD MPH HPH GD SCA

L4/T8 1.89 61.25 30.07 0.31 -0.41

L4/T3 2.73 -6.15 -39.15 0.29 -0.12
L4/T11 3.10 47.56 -1.80 0.29 -0.16
L4/T7 2.85 170.81 126.34 0.32 0.34
L4/T6 3.84 -6.40 -44.89 0.29 -0.31
L5/T1 4.53 235.21 180.15 0.36 -0.07
L5/T4 2.26 66.63 126.62 0.34 -0.19
L5/T9 3.01 -24.52 -11.35 0.29 0.15
L5/T5 4.05 -8.05 -3.00 0.2 0.16
L5/T8 2.50 63.88 135.34 0.33 -0.09
L5/T3 2.57 -23.54 -19.35 0.34 0.41
L5/T11 4.51 -17.48 -7.68 0.35 -0.08
L5/T7 1.45 46.72 66.49 0.33 -0.12
L3/T1 2.72 -66.25 -71.92 0.33 0.21
L3/T4 3.27 12.38 35.99 0.3 -0.53
L3/T9 4.54 21.25 28.15 0.27 -0.18
L3/T5 3.68 6.21 37.45 0.34 0.20
L3/T8 3.83 52.87 46.45 0.29 0.67
L3/T3 5.35 69.78 145.61 0.3 0.41
L3/T11 4.27 29.09 23.67 0.29 0.08
L3/T7 2.71 90.32 92.37 0.32 -0.14
L3/T6 2.39 53.71 57.39 0.27 0.47
L8/T9 2.28 -32.50 -47.75 0.28 0.30
L8/T1 3.80 13.51 68.26 0.33 -0.64
L8/T4 2.39 -41.13 -26.32 0.28 -0.11
L8/T5 2.20 14.02 62.95 0.34 -0.09
L8/T8 3.16 -33.63 -6.17 0.26 0.18
L8/T3 1.76 -28.59 -5.60 0.32 -0.01
L8/T11 1.95 -18.45 2.07 0.32 0.43
L8/T7 2.19 -52.81 -36.86 0.33 -0.10
L8/T6 2.74 -47.17 -28.16 0.29 -0.12

Mean 2.90 33.09 25.73 0.31 0.01

Min 0.94 -66.25 -71.92 0.20 -0.64
Max 5.35 273.74 155.50 0.37 0.74
GYD=grain yield; MPH=mid-parent heterosis; HPH=high-parent heterosis; GD=genetic distance; SCA=specific combining ability;
Min=minimum; Max=maximum.

286
Appendix 13 Mean performance of three-way hybrids for grain yield and other agronomic traits under managed drought
in the 2011 winter season

ENTRY GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA

SC7//L5 2.89 63.3 2.8 178.3 78.0 0.43 10.6 19.8 0.76 5.4 0.5 1.5 2.4
SC8//L10 2.89 63.9 1.2 174.7 101.0 0.53 11.1 5.1 0.65 16.0 0.5 2.7 3.3
SC3//L2 2.88 66.4 1.6 167.7 82.7 0.45 15.8 22.0 0.65 30.8 0.6 3.5 3.6
SC2//L5 2.87 64.9 1.5 170.5 77.4 0.46 11.7 13.1 0.75 6.3 0.6 1.0 2.7
SC10//L6 2.86 66.4 1.0 177.1 88.9 0.52 -1.5 12.1 0.69 16.3 0.5 2.7 3.0
SC3//L1 2.63 65.2 5.4 176.3 81.3 0.41 8.0 12.2 0.56 21.9 0.5 3.0 3.0
SC2//L9 2.61 67.9 1.6 180.3 96.7 0.51 11.7 10.8 0.69 26.0 0.6 3.0 3.8
SC10//L5 2.59 63.7 0.8 178.4 90.4 0.47 10.4 29.4 0.76 4.9 0.6 3.0 3.2
CZH0616 2.57 64.7 4.0 174.8 85.8 0.50 6.3 9.9 0.69 11.4 0.5 3.0 2.9
023WH31 2.54 69.4 1.8 185.9 100.9 0.55 11.8 11.4 0.67 23.3 0.5 2.8 3.2
SC635 2.54 62.9 6.6 133.1 93.0 0.59 10.8 24.5 0.61 34.9 0.6 3.5 3.6
CZH0837 2.53 65.1 1.9 197.6 89.8 0.48 13.4 12.4 0.69 14.4 0.6 2.9 3.5
SC4//L10 2.51 63.0 1.5 170.3 99.4 0.55 33.0 -1.1 0.73 22.5 0.6 3.4 3.5
SC11//L2 2.50 66.6 1.7 182.7 91.3 0.44 9.4 13.9 0.67 24.0 0.5 3.1 2.9
SC3//L10 2.44 61.8 1.0 165.5 90.8 0.49 13.3 10.3 0.83 14.1 0.6 2.7 3.4
SC2//L8 2.42 64.7 -1.1 185.3 90.1 0.53 34.4 13.3 0.79 33.4 0.6 3.2 3.6
SC7//L10 2.42 62.3 3.1 168.6 66.5 0.37 17.2 16.0 0.74 6.8 0.5 2.3 3.3
SC1//L5 2.42 66.2 1.4 166.1 82.6 0.46 14.7 32.0 0.71 3.9 0.5 1.8 2.4
SC8//L8 2.34 70.6 1.2 169.1 104.6 0.56 -2.3 5.8 0.57 32.9 0.5 2.8 4.0
SC3//L9 2.33 67.3 1.1 167.1 103.6 0.55 21.7 22.5 0.54 19.3 0.6 2.8 3.1
013WH29 2.31 66.3 1.5 172.6 93.4 0.51 18.3 0.7 0.66 13.2 0.5 2.4 3.3
SC7//L7 2.28 65.7 3.7 170.6 81.9 0.48 2.7 11.1 0.62 32.9 0.5 2.3 3.5
SC1//L2 2.25 67.8 3.5 171.5 91.3 0.50 3.7 26.5 0.62 9.2 0.5 3.3 3.3
SC3//L8 2.25 62.7 0.3 165.4 81.9 0.48 6.7 25.2 0.53 180.9 0.6 2.7 3.8
SC7//L9 2.24 67.6 1.6 183.8 99.3 0.52 10.3 8.6 0.54 21.4 0.5 2.3 3.2
SC5//L10 2.23 66.3 1.7 188.2 105.7 0.53 28.9 9.6 0.62 13.3 0.6 3.6 3.6
SC1//L6 2.23 67.7 1.3 165.5 91.8 0.50 4.8 2.5 0.61 24.2 0.6 2.9 3.4
SC5//L8 2.22 66.4 1.3 173.1 99.7 0.54 8.7 15.6 0.65 34.2 0.5 3.1 4.0
SC2//L4 2.21 65.7 1.9 180.9 95.2 0.52 22.5 14.0 0.56 27.3 0.6 3.1 3.4
SC2//L10 2.19 62.2 0.9 166.1 92.4 0.54 50.7 9.7 0.65 14.4 0.6 2.6 3.3
SC11//L5 2.18 63.8 0.9 170.7 94.5 0.47 12.6 28.7 0.65 19.5 0.6 1.6 2.7
SC3//L11 2.14 64.0 0.7 156.6 91.8 0.47 23.1 33.6 0.70 26.5 0.6 2.7 2.9
SC4//L11 2.14 65.4 2.5 174.4 100.7 0.55 -0.4 14.5 0.63 19.6 0.5 3.3 4.1
SC3//L7 2.12 65.4 0.1 161.3 83.6 0.50 14.7 24.5 0.77 2.3 0.6 2.8 2.8
SC10//L8 2.11 64.0 1.8 163.6 95.6 0.53 12.1 6.0 0.72 23.8 0.5 3.6 4.0

287
Appendix 13 Mean performance of three-way hybrids for grain yield and other agronomic traits under managed drought in the
2011 winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA
SC10//L7 2.08 66.9 1.7 161.4 83.7 0.49 19.2 25.4 0.69 13.0 0.5 2.7 3.2
SC11/L8 2.08 62.8 1.3 177.1 95.3 0.51 15.0 9.0 0.82 21.4 0.5 2.2 3.2
SC11//L6 2.05 65.9 1.1 167.3 85.9 0.49 -1.9 10.1 0.68 9.6 0.5 2.6 3.5
SC4//L12 2.04 63.0 1.8 167.2 100.8 0.59 31.8 11.6 0.71 45.6 0.6 4.1 3.6
SC1//L8 2.01 64.6 0.7 178.8 92.6 0.48 20.9 9.7 0.75 32.3 0.6 2.8 3.3
SC3//L3 1.99 63.6 2.7 188.0 92.6 0.48 20.6 14.8 0.57 27.8 0.6 3.0 3.5
SC6//L12 1.99 62.7 1.8 182.6 100.2 0.48 17.0 20.3 0.66 22.5 0.6 3.6 3.8
SC8//L9 1.99 68.5 3.9 178.2 103.5 0.67 2.2 1.8 0.50 26.8 0.5 2.6 4.1
SC10//L2 1.98 68.0 3.3 179.2 95.6 0.56 9.8 6.4 0.53 37.6 0.5 3.4 3.7
SC10//L1 1.96 66.5 5.7 187.3 98.9 0.55 12.2 6.9 0.56 40.2 0.5 3.0 4.1
SC10//L3 1.94 67.7 5.8 180.5 96.8 0.52 10.3 5.5 0.58 17.6 0.5 3.4 3.4
SC2//L6 1.94 64.9 0.1 157.8 79.9 0.48 7.6 24.3 0.65 6.9 0.6 2.6 3.8
SC9/L12 1.94 64.8 1.1 174.0 98.9 0.53 17.7 20.2 0.66 27.8 0.6 4.0 3.4
SC2//L2 1.90 66.0 1.1 171.9 94.6 0.54 17.8 19.3 0.53 48.5 0.6 3.8 4.4
SC6//L10 1.89 63.9 3.1 169.6 97.2 0.56 26.3 18.5 0.59 34.0 0.6 3.5 3.9
SC9//L9 1.86 69.3 3.8 169.0 113.7 0.60 16.4 10.6 0.54 18.1 0.5 2.9 3.5
SC2//L11 1.85 66.0 1.7 169.7 92.8 0.51 12.9 11.8 0.69 42.7 0.6 2.7 3.9
SC513 1.82 65.6 2.8 189.8 82.1 0.43 41.9 8.0 0.58 28.2 0.5 1.9 3.4
SC2//L7 1.79 66.1 2.0 169.8 88.3 0.48 29.3 29.3 0.61 21.3 0.6 3.5 3.1
SC1//L3 1.79 67.6 3.0 175.2 93.8 0.50 16.9 13.6 0.53 37.5 0.5 3.1 3.0
SC11//L7 1.78 65.4 2.2 164.6 90.5 0.53 0.7 32.3 0.68 7.9 0.6 1.7 3.1
SC8//L5 1.75 63.4 1.9 171.6 98.0 0.53 5.7 10.5 0.57 32.1 0.5 2.8 3.8
SC8//L12 1.75 66.2 0.0 172.1 89.3 0.52 8.1 14.7 0.63 36.6 0.6 3.3 4.5
SC5//L1 1.74 68.4 5.8 177.1 107.3 0.57 14.2 7.1 0.34 16.9 0.5 3.7 3.2
SC9//L10 1.71 63.4 3.3 181.5 100.2 0.54 16.1 13.3 0.61 10.4 0.5 3.3 3.4
SC6//L11 1.67 67.5 3.3 186.3 107.1 0.58 7.8 11.8 0.52 43.0 0.5 3.5 4.6
SC2//L4 1.64 66.4 2.7 176.5 95.5 0.51 17.8 12.1 0.50 41.9 0.6 3.8 3.2
SC7//L6 1.64 66.8 2.0 175.7 83.0 0.46 0.0 12.1 0.58 31.7 0.6 2.7 3.9
SC3//L6 1.63 66.0 0.8 175.6 78.4 0.47 10.5 16.9 0.57 12.9 0.6 2.9 3.3
SC7//L2 1.60 66.2 5.1 188.6 99.3 0.50 14.3 9.7 0.47 18.1 0.5 2.9 3.4
SC3//L4 1.60 64.5 1.4 184.3 97.5 0.51 19.3 26.0 0.56 29.3 0.7 3.5 3.2
SC1//L4 1.59 66.6 4.1 191.1 96.2 0.50 1.6 10.1 0.52 31.9 0.6 3.3 3.2
SC5//L3 1.55 67.7 3.5 185.4 104.3 0.54 16.6 -0.3 0.42 29.1 0.5 3.1 3.5
SC5//L7 1.52 67.7 2.1 170.7 95.1 0.52 12.0 20.6 0.53 45.7 0.6 3.5 4.3
SC9//L11 1.51 67.3 3.6 160.9 102.2 0.70 9.9 7.7 0.48 25.5 0.6 2.8 3.5
SC2//L1 1.48 66.7 6.8 185.6 94.9 0.50 19.8 7.4 0.44 39.0 0.6 3.4 4.1
013WH01 1.45 68.5 5.2 198.5 109.2 0.54 7.9 4.6 0.33 40.3 0.6 2.5 3.2

288
Appendix 13 Mean performance of three-way hybrids for grain yield and other agronomic traits under managed drought in the
2011 winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA
SC1//L1 1.44 66.9 5.2 176.7 83.0 0.47 7.7 14.3 0.41 24.0 0.5 3.0 3.4
SC1//L7 1.41 68.1 2.0 168.3 88.5 0.48 8.1 26.7 0.51 12.5 0.6 3.0 3.5
SC5//L6 1.34 69.2 1.1 174.1 100.2 0.54 2.8 6.9 0.42 30.2 0.6 3.3 4.0
SC4//L9 1.27 67.6 3.0 165.9 108.2 0.60 1.7 8.6 0.42 40.6 0.5 2.9 4.2
SC7//L3 1.26 65.9 4.7 184.5 95.1 0.51 8.5 1.6 0.42 18.4 0.5 2.9 3.7
SC8/L11 1.22 67.6 3.6 179.6 102.3 0.54 6.4 14.6 0.36 31.1 0.6 2.7 3.5
SC6/L9 1.19 68.2 2.9 181.4 102.5 0.55 7.9 12.3 0.49 21.5 0.5 3.3 4.2
SC7//L4 1.18 65.7 5.9 195.1 103.0 0.52 9.8 5.6 0.39 34.1 0.5 2.5 3.5
SC5//L2 1.15 69.6 6.3 187.5 98.1 0.51 8.7 9.0 0.41 27.0 0.5 4.1 4.1
SC5//L11 0.83 68.9 2.6 168.5 117.7 0.65 0.2 7.3 0.37 26.8 0.6 3.3 4.1
SC10//L4 0.83 68.1 7.4 190.5 107.3 0.56 6.8 11.9 0.26 48.0 0.5 3.4 4.2
SC11//L4 0.80 66.1 3.4 178.6 102.2 0.51 22.3 16.8 0.51 22.9 0.5 3.0 3.3
013WH63 0.68 66.9 5.1 190.2 97.0 0.53 17.2 6.7 0.40 40.6 0.6 3.9 4.3
Mean 1.95 66.0 2.6 175.5 94.5 0.52 13.1 13.7 0.59 26.6 0.6 3.0 3.5
LSD
1.03 2.0 2.4 17.8 13.1 0.10 20.3 13.0 0.16 46.7 0.1 0.9 0.9
(0.05)
MSE 0.53 2.1 4.4 80.5 130.6 0.00 104.5 128.9 0.02 1651.1 0.0 0.2 0.4
Min 0.68 61.8 -1.1 133.1 66.5 0.37 -2.3 -1.1 0.26 2.3 0.5 1.0 2.4
Max 2.89 70.6 7.4 198.5 117.7 0.70 50.7 33.6 0.83 180.9 0.7 4.1 4.6
GYD=grain yield (t ha-1); AD=anthesis days; ASI=anthesis silking interval (days); PH=plant height (cm); EH=ear height (cm); EPO=ear position (0-1); RL=root lodging (%); SL=stem
lodging (%); EPP=ears per plant (#); ER=ear rot (%); SEN=senescence (1-10); TEX=texture (1-5); EA=ear aspect (1-5); LSD=least significant difference; MSE=mean square error;
Min=minimum; Max=maximum.

289
Appendix 14 Mean performance of three-way hybrids for grain yield and other agronomic traits under optimum
conditions in the 2011 winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER TEX EA
SC9//L10 4.79 69.9 2.1 162.9 90.9 0.57 34.6 17.8 0.66 15.4 3.6 4.1
SC10//L5 4.65 71.2 2.9 174.0 90.7 0.54 -5.2 16.5 0.94 7.8 3.1 3.7
SC4//L10 4.64 70.8 1.4 171.2 103.1 0.60 12.0 9.6 0.75 8.6 3.7 3.9
SC6//L10 4.64 71.2 2.0 176.5 98.3 0.56 30.4 7.6 0.83 8.6 3.8 3.8
SC8//L10 4.63 71.6 1.8 179.6 98.5 0.55 2.2 6.9 0.79 11.0 3.6 4.1
SC3//L8 4.63 69.7 4.0 157.6 82.2 0.52 0.9 20.2 0.84 2.8 3.3 3.8
SC11//L5 4.60 69.8 4.0 156.8 89.7 0.58 2.5 22.3 0.91 3.3 2.7 3.5
SC6//L9 4.58 73.6 0.9 187.0 104.2 0.56 -9.7 9.8 0.69 12.1 3.7 3.3
SC9//L9 4.52 75.1 0.7 177.7 99.6 0.57 -2.3 43.4 1.25 6.5 4.8 2.6
SC1//L6 4.50 71.6 2.1 172.6 89.1 0.53 9.4 11.2 0.79 3.2 3.3 3.6
SC10//L6 4.50 71.5 1.5 166.9 92.5 0.55 -6.9 9.2 0.75 6.1 3.7 3.9
SC8//L9 4.46 74.6 0.3 183.4 105.5 0.57 -2.8 2.3 0.80 6.0 3.4 3.5
SC10//L7 4.42 72.1 5.5 163.1 90.2 0.55 0.6 35.4 0.88 7.6 3.2 3.4
SC2//L9 4.40 72.7 1.4 181.8 95.9 0.55 -11.5 15.8 0.77 11.8 3.3 3.8
SC1//L2 4.36 73.8 1.1 174.9 88.3 0.51 -5.4 15.1 0.66 18.8 3.7 3.7
SC8//L5 4.36 70.9 2.4 174.8 94.8 0.54 6.4 6.7 0.80 8.9 2.9 3.2
013WH29 4.33 71.2 2.5 162.1 96.6 0.61 -5.2 12.8 0.80 19.6 3.3 3.6
SC10//L8 4.33 71.2 1.8 172.5 100.1 0.58 -11.1 15.5 0.79 9.8 3.4 3.9
SC8//L8 4.27 74.3 0.7 181.6 101.7 0.56 11.2 10.7 0.74 11.5 3.4 3.7
SC11//L8 4.22 70.4 1.4 171.3 103.0 0.60 7.8 10.4 0.86 7.3 2.9 3.6
SC5//L6 4.11 74.0 -0.1 182.6 86.8 0.49 -4.1 13.0 0.82 3.2 3.5 3.9
CZH0616 4.01 70.6 3.7 152.9 86.5 0.58 -6.4 22.8 0.89 13.5 3.1 3.7
SC10//L1 4.00 71.5 5.6 176.6 94.2 0.54 7.0 18.4 0.84 8.4 3.5 4.1
SC11//L6 3.99 72.8 -0.1 165.9 91.3 0.55 3.5 18.3 0.84 10.0 3.0 3.3
SC3//L10 3.99 68.6 5.7 180.3 94.6 0.53 5.1 11.6 0.85 5.8 3.1 3.7
SC5//L11 3.99 72.8 1.6 184.1 116.4 0.63 -7.4 15.3 0.69 15.4 3.3 4.4
SC2//L10 3.92 69.9 3.7 160.7 92.6 0.58 30.3 3.2 0.80 17.6 3.4 4.3
SC5//L10 3.87 72.2 0.9 177.3 108.8 0.62 14.2 9.1 0.70 11.0 2.9 3.4
SC6//L11 3.86 72.7 1.5 182.7 102.6 0.56 22.7 14.2 0.72 11.4 3.3 3.6
SC8//L11 3.83 73.6 -0.1 186.1 109.1 0.59 -3.8 24.3 0.75 14.7 3.2 3.9
SC7//L9 3.81 73.2 1.0 180.8 102.1 0.57 14.8 22.1 0.82 9.8 2.8 3.9
SC2//L11 3.81 70.2 3.8 167.6 94.3 0.57 8.0 21.3 0.77 11.5 3.6 3.7
SC7//L10 3.80 70.4 2.6 165.3 85.0 0.53 15.5 15.8 0.81 9.8 3.1 3.5
SC7//L6 3.78 71.9 1.4 176.3 85.4 0.49 -8.5 20.4 0.70 7.8 3.2 3.5

290
Appendix 14 Mean performance of three-way hybrids for grain yield and other agronomic traits under optimum
conditions in the 2011 winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER TEX EA
SC8//L12 3.74 72.3 1.1 171.0 90.9 0.54 0.6 7.3 0.85 15.5 3.7 4.3
023WH31 3.74 72.6 1.4 175.1 96.9 0.57 -3.3 8.2 0.62 7.0 3.1 3.9
SC2//L6 3.74 71.4 2.9 161.8 88.2 0.56 -4.4 22.7 0.75 -1.1 3.7 3.8
SC7//L5 3.72 69.7 3.5 163.6 79.9 0.51 -10.0 22.0 0.87 17.2 2.6 4.0
SC3//L9 3.60 72.6 2.3 165.7 92.3 0.58 -10.1 4.2 0.82 11.2 3.3 4.1
SC2//L4 3.55 72.6 -0.6 176.6 89.4 0.53 0.2 14.2 0.83 7.5 3.5 4.3
SC1//L8 3.51 70.6 2.7 162.2 104.2 0.66 -2.0 16.7 0.76 12.3 3.4 4.1
SC10//L2 3.50 74.2 1.4 179.5 93.8 0.53 -2.9 7.0 0.78 7.6 3.9 3.6
SC3//L11 3.50 68.5 6.9 159.9 82.2 0.53 -1.7 11.6 0.82 11.4 3.2 3.3
SC635 3.48 71.8 -0.5 153.1 82.4 0.55 57.8 13.0 0.51 10.0 3.6 4.0
SC10//L3 3.43 73.0 2.9 185.7 95.3 0.53 -3.7 23.2 0.73 27.1 3.7 4.3
SC9//L11 3.41 72.2 1.0 179.1 101.7 0.58 8.1 28.2 0.82 15.7 3.3 4.4
SC10//L4 3.33 75.2 -0.1 189.9 112.1 0.61 12.9 20.7 0.67 5.1 3.6 3.9
SC513 3.30 71.2 0.7 178.9 87.2 0.50 -2.3 22.8 0.81 15.7 3.1 3.6
SC5//L8 3.28 71.7 1.3 176.8 99.5 0.56 -6.6 11.5 0.77 15.1 3.3 4.4
SC5//L2 3.28 73.3 0.8 179.6 89.8 0.51 -5.8 22.9 0.58 17.6 4.0 4.1
SC3//L2 3.22 72.5 5.7 171.1 81.5 0.48 -1.9 11.5 0.77 13.0 4.0 3.4
SC2//L2 3.19 72.6 1.6 166.8 83.5 0.51 -2.4 18.8 0.83 20.5 3.6 4.3
SC4//L9 3.18 74.4 0.0 173.6 96.2 0.54 12.8 12.6 0.65 17.3 3.5 3.2
SC5//L11 3.18 73.6 1.4 187.4 105.9 0.58 2.9 12.6 0.71 15.9 3.8 4.5
SC4//L12 3.13 69.6 2.2 173.2 93.2 0.54 -3.0 13.9 0.76 22.1 4.0 4.6
SC7//L3 3.12 71.3 5.3 176.3 85.2 0.49 14.8 10.6 0.63 22.3 3.4 4.2
SC1//L3 3.12 73.0 2.8 177.1 85.9 0.50 14.3 22.5 0.83 14.6 3.7 3.7
SC7//L2 3.12 72.0 3.7 176.0 87.4 0.51 3.1 22.4 0.82 8.2 3.5 3.6
SC1//L1 3.07 71.0 6.1 182.3 91.8 0.50 -4.2 13.9 0.83 15.4 4.1 4.3
SC11//L7 3.07 70.9 4.7 162.1 81.7 0.51 7.6 20.6 0.80 8.9 2.9 3.4
SC6//L12 3.04 70.5 3.1 167.7 87.5 0.52 -2.8 25.1 0.81 20.9 4.1 4.6
SC3//L4 3.03 71.3 3.2 169.7 80.9 0.50 -3.5 7.1 0.74 8.6 3.6 4.6
SC11//L2 3.03 72.6 1.2 177.7 103.5 0.60 23.7 14.0 0.76 5.0 3.7 3.9
SC11//L4 3.03 72.9 1.3 186.4 106.0 0.59 54.0 22.3 0.63 9.8 3.7 3.9
SC3//L3 3.00 69.8 2.9 177.0 91.2 0.52 -2.8 18.2 0.82 11.8 3.8 3.9
SC5//L7 3.00 73.3 1.4 173.4 100.7 0.59 21.6 18.1 0.76 2.3 3.4 4.7
SC2//L8 2.92 70.5 2.6 165.1 94.2 0.56 -3.5 22.2 0.80 9.1 3.4 4.4
SC5//L1 2.90 72.5 4.9 190.7 93.7 0.48 -4.1 15.0 0.77 16.0 3.9 4.6
CZH0837 2.89 70.9 3.8 176.6 91.4 0.53 -5.2 23.9 0.76 22.6 3.5 4.1
SC7//L7 2.88 71.2 4.4 168.3 87.5 0.53 15.3 22.1 0.82 15.4 3.1 4.3

291
Appendix 14 Mean performance of three-way hybrids for grain yield and other agronomic traits under optimum
conditions in the 2011 winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER TEX EA
SC2//L5 2.73 71.0 1.6 153.9 78.3 0.54 5.0 21.9 0.88 15.3 3.2 4.6
SC2//L1 2.71 70.6 6.3 179.5 88.4 0.52 4.7 11.4 0.75 24.5 3.6 4.0
SC4//L11 2.71 71.7 2.1 170.9 107.3 0.63 1.3 17.8 0.73 21.8 4.0 4.3
SC1//L7 2.70 71.3 3.8 165.3 83.6 0.52 -3.5 25.1 0.76 12.8 3.4 4.3
SC2//L7 2.61 70.7 4.2 159.3 74.4 0.50 15.3 23.6 0.79 18.7 3.3 4.0
SC9//L12 2.61 72.0 1.0 165.2 80.5 0.49 14.0 16.3 0.79 25.2 3.8 4.4
013WH63 2.56 72.4 -0.2 171.0 93.0 0.54 24.2 18.9 0.60 5.8 3.8 4.6
SC3//L6 2.54 70.8 3.3 164.3 83.3 0.52 2.2 12.3 0.77 8.4 3.8 4.0
SC3//L7 2.43 70.2 3.3 149.5 69.3 0.48 1.1 25.9 0.81 2.9 3.2 3.9
SC2//L4 2.37 71.6 1.8 174.2 83.6 0.49 -0.2 21.9 0.85 18.6 3.9 4.0
SC7//L4 2.36 73.2 0.3 185.2 104.5 0.58 38.7 20.3 0.65 9.7 4.0 4.2
SC3//L1 2.24 69.7 7.0 171.0 88.0 0.52 -3.7 20.4 0.84 21.6 3.9 4.6
SC1//L5 2.20 71.3 3.8 158.3 81.3 0.52 -5.6 19.4 0.90 10.2 3.0 3.3
SC1//L4 1.99 72.7 0.0 178.1 100.3 0.57 26.4 16.6 0.66 18.3 4.2 4.6
013WH01 1.95 74.7 0.5 183.8 106.6 0.59 31.1 10.6 0.65 7.6 3.6 4.1
Mean 3.51 71.8 2.4 172.6 93.0 0.55 5.4 16.6 0.78 12.2 3.5 3.9
LSD (0.05) 1.33 1.6 2.6 13.0 11.6 0.06 29.4 17.5 0.21 12.8 0.7 0.8
MSE 0.90 2.0 1.7 128.8 101.5 0.00 218.8 154.1 0.02 41.3 0.3 0.4
Min 1.95 68.5 -0.6 149.5 69.3 0.48 -11.5 2.3 0.51 -1.1 2.6 2.6
Max 4.79 75.2 7.0 190.7 116.4 0.66 57.8 43.4 1.25 27.1 4.8 4.7
GYD=grain yield (t ha-1); AD=anthesis days; ASI=anthesis silking interval (days); PH=plant height (cm); EH=ear height (cm); EPO=ear position (0-1); RL=root lodging (%); SL=stem
lodging (%); EPP=ears per plant (#); ER=ear rot (%); SEN=senescence (1-10); TEX=texture (1-5); EA=ear aspect (1-5); LSD=least significant difference; MSE=mean square error;
Min=minimum; Max=maximum.

292
Appendix 15 Mean performance of three-way hybrids for grain yield and other agronomic traits in combined analysis in
the 2011 winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA
SC8//L10 3.76 68.5 1.3 178.3 99.8 0.54 6.6 5.8 0.70 14.8 0.5 3.4 3.7
SC10//L6 3.68 69.5 1.1 169.5 90.7 0.54 -4.2 10.9 0.71 13.8 0.5 3.5 3.5
SC10//L5 3.62 68.2 1.3 175.1 90.6 0.51 2.6 24.2 0.83 5.6 0.6 3.1 3.4
SC4//L10 3.58 67.7 1.4 170.9 101.3 0.58 22.5 3.2 0.74 19.1 0.6 3.7 3.7
SC2//L9 3.50 70.7 1.5 181.5 96.3 0.54 0.1 12.8 0.72 22.5 0.6 3.2 3.8
SC3//L8 3.44 66.9 1.2 159.6 82.0 0.51 3.8 23.2 0.66 136.4 0.6 3.1 3.8
SC11//L5 3.39 67.4 1.6 160.3 92.1 0.53 7.5 26.1 0.75 15.5 0.6 2.4 3.1
SC1//L6 3.36 70.1 1.5 170.8 90.4 0.52 7.1 6.0 0.68 19.0 0.6 3.2 3.5
013WH29 3.32 69.2 1.8 164.7 95.0 0.57 6.6 5.5 0.72 14.8 0.5 3.1 3.5
SC3//L1 3.31 67.1 2.9 167.3 78.9 0.48 0.3 20.6 0.80 8.4 0.5 2.3 3.2
SC1//L2 3.31 71.4 2.9 174.1 89.8 0.51 -0.9 21.9 0.64 11.6 0.5 3.6 3.5
SC8//L8 3.30 72.8 1.0 178.5 103.2 0.56 4.5 7.7 0.63 27.5 0.5 3.2 3.9
CZH0616 3.29 68.2 3.9 158.4 86.2 0.55 0.0 15.0 0.77 11.9 0.5 3.1 3.3
SC6//L10 3.26 68.3 2.8 174.7 97.7 0.56 28.4 14.1 0.69 27.7 0.6 3.7 3.8
SC10//L7 3.25 70.0 2.7 162.7 87.0 0.52 9.9 29.4 0.77 11.6 0.5 3.1 3.3
SC9//L10 3.25 67.3 3.0 167.5 95.5 0.56 25.3 15.1 0.63 11.6 0.5 3.5 3.7
SC8//L9 3.22 72.2 3.0 182.1 104.5 0.61 -0.3 2.0 0.62 21.6 0.5 3.2 3.8
SC10//L8 3.22 68.3 1.8 170.3 97.9 0.56 0.5 9.8 0.75 20.3 0.5 3.4 4.0
SC3//L10 3.22 65.9 2.2 176.6 92.7 0.51 9.2 10.8 0.84 12.0 0.6 3.0 3.5
SC9//L9 3.19 72.8 3.0 175.5 106.7 0.58 7.0 23.8 0.82 15.2 0.5 4.3 3.1
SC11//L8 3.15 67.4 1.3 172.8 99.2 0.56 11.4 9.6 0.83 17.9 0.5 2.7 3.4
023WH31 3.14 71.4 1.7 177.8 98.9 0.56 4.2 10.1 0.65 19.2 0.5 3.1 3.6
SC7//L10 3.11 67.2 3.0 166.1 75.8 0.46 16.3 15.9 0.77 7.6 0.5 2.9 3.4
SC8//L5 3.06 67.9 2.0 174.0 96.4 0.54 6.0 9.0 0.66 26.3 0.5 2.9 3.5
SC2//L10 3.06 66.8 1.6 162.1 92.5 0.56 40.5 7.1 0.71 15.2 0.6 3.2 3.8
SC3//L2 3.05 70.0 2.6 170.3 82.1 0.47 6.9 17.8 0.70 26.3 0.6 3.9 3.5
SC5//L10 3.05 69.8 1.5 180.0 107.2 0.59 21.6 9.4 0.65 12.7 0.6 3.1 3.5
SC7//L9 3.03 70.9 1.4 181.5 100.7 0.55 12.5 14.0 0.65 18.5 0.5 2.7 3.6
SC11//L6 3.02 70.0 0.8 166.3 88.6 0.53 0.8 13.4 0.75 9.7 0.5 2.9 3.4
SC635 3.01 68.2 4.8 148.1 87.7 0.57 34.3 19.9 0.57 28.7 0.6 3.6 3.8
SC10//L1 2.98 69.5 5.7 179.3 96.6 0.54 9.6 11.5 0.67 32.2 0.5 3.4 4.1
SC3//L9 2.96 70.5 1.4 166.0 97.9 0.57 5.8 15.2 0.65 17.3 0.6 3.1 3.6
SC6//L9 2.89 71.4 2.4 185.6 103.3 0.56 -0.9 11.3 0.57 19.1 0.5 3.6 3.7
SC2//L6 2.84 68.8 0.8 160.8 84.1 0.53 1.6 23.7 0.69 4.9 0.6 3.4 3.8

293
Appendix 15 Mean performance of three-way hybrids for grain yield and other agronomic traits in combined analysis in the 2011
winter season

ENTRY GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA

SC2//L11 2.83 68.5 2.2 168.1 93.5 0.55 10.5 15.6 0.72 34.9 0.6 3.4 3.8
SC3//L11 2.82 66.7 2.2 159.1 87.0 0.51 10.7 24.8 0.75 22.7 0.6 3.1 3.1
SC2//L6 2.80 68.6 1.5 158.0 77.9 0.51 8.4 16.6 0.80 8.6 0.6 2.6 3.6
SC6//L11 2.77 70.6 2.9 183.6 104.9 0.57 15.3 12.7 0.60 35.1 0.5 3.3 4.1
SC11//L2 2.76 70.2 1.6 178.9 97.4 0.54 16.6 13.9 0.71 19.3 0.5 3.6 3.4
SC1//L8 2.76 68.2 1.2 166.4 98.4 0.59 9.5 12.5 0.76 27.3 0.6 3.2 3.7
SC5//L8 2.75 69.6 1.3 175.8 99.6 0.55 1.0 13.9 0.69 29.4 0.5 3.3 4.2
SC8//L12 2.75 69.9 0.3 171.3 90.1 0.53 4.4 11.7 0.72 31.4 0.6 3.6 4.4
SC10//L2 2.74 71.7 2.8 179.4 94.7 0.54 3.5 6.6 0.63 30.1 0.5 3.8 3.7
SC5//L6 2.73 72.1 0.8 180.5 93.5 0.51 -0.6 9.3 0.58 23.5 0.6 3.5 4.0
CZH0837 2.71 68.5 2.4 181.8 90.6 0.51 4.1 17.0 0.72 16.5 0.6 3.4 3.8
SC7//L6 2.71 69.9 1.9 176.2 84.2 0.48 -4.2 15.4 0.62 25.7 0.6 3.1 3.7
SC10//L3 2.69 70.9 5.1 184.4 96.1 0.52 3.3 12.5 0.64 20.0 0.5 3.6 3.9
SC2//L8 2.67 68.2 -0.2 170.2 92.2 0.55 15.5 16.9 0.79 27.3 0.6 3.4 4.0
SC2//L4 2.59 70.1 1.9 176.6 92.5 0.52 9.0 13.0 0.63 33.3 0.6 3.6 3.8
SC4//L12 2.58 66.9 1.9 171.7 97.0 0.56 14.4 12.5 0.73 39.7 0.6 4.0 4.1
SC7//L7 2.58 69.0 3.8 168.9 84.7 0.51 9.0 15.5 0.70 28.5 0.5 2.9 3.9
SC513 2.56 68.9 2.3 181.6 84.7 0.47 19.8 13.9 0.67 25.1 0.5 2.8 3.5
SC2//L2 2.55 69.9 1.2 168.0 89.1 0.52 7.7 19.1 0.65 41.5 0.6 3.7 4.3
SC8//L11 2.53 71.2 2.7 184.5 105.7 0.57 1.3 18.5 0.51 27.0 0.6 3.1 3.7
SC6//L12 2.51 67.4 2.1 171.4 93.8 0.51 7.1 22.2 0.72 22.1 0.6 3.9 4.2
SC3//L3 2.50 67.3 2.8 179.7 91.9 0.50 8.9 16.1 0.67 23.8 0.6 3.6 3.7
SC9//L11 2.46 70.2 2.9 174.6 102.0 0.63 9.0 15.9 0.61 23.0 0.6 3.1 3.9
SC1//L3 2.46 70.8 3.0 176.6 89.9 0.50 15.6 17.2 0.65 31.8 0.5 3.6 3.4
SC3//L1 2.44 67.9 5.8 172.3 84.6 0.48 2.2 15.5 0.67 21.9 0.5 3.6 3.8
SC11//L7 2.43 68.7 2.8 162.7 86.1 0.52 4.1 27.6 0.73 8.2 0.6 2.6 3.2
SC4//L11 2.42 69.2 2.4 171.8 104.0 0.60 0.5 15.8 0.67 20.1 0.5 3.8 4.2
SC5//L11 2.41 71.2 2.4 180.2 117.0 0.64 -3.6 10.5 0.50 23.9 0.6 3.3 4.2
SC5//L3 2.36 71.3 3.0 186.9 105.1 0.56 9.8 4.9 0.54 25.8 0.5 3.6 4.0
SC7//L2 2.36 69.7 4.7 179.2 93.3 0.50 8.7 14.8 0.61 15.6 0.5 3.3 3.5
SC5//L1 2.32 70.9 5.6 187.3 100.5 0.52 5.0 10.2 0.51 16.7 0.5 3.8 3.9
SC3//L4 2.31 68.6 1.9 173.3 89.2 0.50 7.9 18.4 0.64 24.2 0.7 3.6 3.9
SC1//L5 2.31 69.3 2.0 160.2 82.0 0.50 4.5 27.0 0.78 5.5 0.5 2.7 2.9
SC2//L4 2.29 69.3 1.9 175.9 89.4 0.50 11.1 17.1 0.68 25.1 0.6 3.7 3.7
SC3//L7 2.27 68.3 0.9 152.5 76.4 0.48 7.9 25.1 0.79 2.4 0.6 3.1 3.3
SC9//L12 2.27 69.1 1.1 167.4 89.7 0.50 15.8 18.6 0.71 27.1 0.6 3.9 3.9

294
 Appendix 15 Mean performance of three-way hybrids for grain yield and other agronomic traits in combined analysis in the 2011
winter season
ENTRY GYD AD ASI PH EH EPO RL SL EPP ER SEN TEX EA
SC5//L7 2.26 71.0 1.9 172.8 97.9 0.56 16.8 19.6 0.62 34.9 0.6 3.4 4.5
SC1//L1 2.26 69.4 5.4 180.9 87.4 0.49 1.7 14.2 0.58 21.8 0.5 3.9 3.8
SC4//L9 2.22 71.7 2.3 171.7 102.2 0.57 7.2 10.2 0.52 34.8 0.5 3.4 3.7
SC5//L2 2.22 71.8 4.9 181.6 93.9 0.51 1.4 14.5 0.48 24.6 0.5 4.1 4.1
SC2//L7 2.20 68.8 2.6 161.9 81.4 0.49 22.3 27.0 0.68 20.6 0.6 3.3 3.5
SC7//L3 2.19 69.1 4.8 178.4 90.2 0.50 11.7 5.2 0.51 19.4 0.5 3.2 4.0
SC2//L1 2.09 69.0 6.6 181.0 91.6 0.51 12.3 9.0 0.56 35.4 0.6 3.6 4.1
SC3//L6 2.08 68.9 1.5 167.1 80.9 0.50 6.4 15.0 0.65 11.8 0.6 3.5 3.7
SC10//L4 2.08 72.4 5.5 190.1 109.7 0.59 9.9 15.5 0.43 37.3 0.5 3.6 4.0
SC1//L7 2.05 70.0 2.5 166.0 86.1 0.50 2.3 26.0 0.61 12.6 0.6 3.3 3.9
SC11//L4 1.91 70.2 2.9 184.4 104.1 0.56 38.1 19.0 0.56 19.6 0.5 3.5 3.6
SC1//L4 1.79 70.2 3.1 181.3 98.2 0.54 14.0 12.7 0.58 28.5 0.6 3.9 3.9
SC7//L4 1.77 70.2 4.5 187.7 103.7 0.56 24.2 11.4 0.50 28.0 0.5 3.6 3.9
013WH01 1.70 72.2 4.0 187.5 107.9 0.57 19.5 7.0 0.46 32.1 0.6 3.3 3.6
013WH63 1.62 70.2 3.8 175.8 95.0 0.54 20.7 11.6 0.48 31.9 0.6 3.9 4.4
Mean 2.73 69.5 2.5 173.3 93.7 0.53 9.3 14.9 0.66 23.0 0.6 3.4 3.7
LSD (0.05) 0.84 1.3 1.9 10.7 8.7 0.05 17.9 10.5 0.13 35.1 0.1 0.5 0.6
MSE 0.72 2.0 3.8 116.7 116.1 0.00 161.6 139.0 0.02 1248.7 0.0 0.3 0.4
Min 1.62 65.9 -0.2 148.1 75.8 0.46 -4.2 2.0 0.43 2.4 0.5 2.3 2.9
Max 3.76 72.8 6.6 190.1 117.0 0.64 40.5 29.4 0.84 136.4 0.7 4.3 4.5
GYD=grain yield (t ha-1); AD=anthesis days; ASI=anthesis silking interval (days); PH=plant height (cm); EH=ear height (cm); EPO=ear position (0-1); RL=root lodging (%); SL=stem
lodging (%); EPP=ears per plant (#); ER=ear rot (%); SEN=senescence (1-10); TEX=texture (1-5); EA=ear aspect (1-5); LSD=least significant difference; MSE=mean square error;
Min=minimum; Max=maximum.

295