V . m. l. tammes and r. a.

whitehead

with foreign countries are therefore possible (Whitehead, 1963, 1965).

Artificial pollination, either selfing or crossing, can be done with fresh or stored
pollen. The inflorescence is emasculated by cutting the branches above the female
flowers and removing by hand the few remaining male flowers. The emasculated
bunch is then enclosed in a cotton or canvas bag which is tied at the base of the bunch
stalk over a non absorbent cotton plug which will prevent insects crawling in. The
inflorescence must be regularly visited and may be pollinated as soon as the female
flowers start secreting nectar and stigmas open. Plastic windows in the pollination
bags facilitate observation.
Pollination can be done daily until the white stigmas turn brown and become post
receptive, but good fruit setting is often obtained with only one or two pollinations.
Pollen application may be by simply rubbing the fresh anthers onto the stigmas;
stored pollen is applied with a fine brush or more conveniently, from a polyethylene
puffer bottle.

a. Selfing

The progeny of selfed trees may show weaknesses. The first author observed that
lower germination and less vigorous seedlings were obtained after self-pollination in
Celebes. Gangolly et al., (1957) also mentioned reduced vigour in selfed lines, with
the exception of two types from New Guinea and Cochin China, and Satyabalan
et al., (1960) find a distinctly lower production in trees from selfed pollination
compared with those resulting from cross-pollination. Other deleterious effects
of selfing include the production of progeny which bear nuts with no copra (Ann.
Rep. Coconut Exp. Station Kasaragod (1959-1960)).
It is not known whether the weakness of some dwarf types is due to inbreeding,
but Swaminathan et al. (1961) found that irregular meiotic division is more prevalent
in dwarfs than in tails, and sterile pollen grains are more frequent in the former.
Though for general seed production it is important to avoid inbreeding the recom­
bination of inbred lines of strong hybrid vigour combinations is likely to lead to
significant improvements.

b. Crossing

Crosses between tall coconuts and dwarfs are possible and occur in nature. These
hybrids are early flowering, with nuts intermediate in size to those of the two parents ;
male and female flowering cycles overlap to only a small extent. The first cross, F1;
is very homogeneous and does not show the undesirable weaknesses of the dwarf
varieties. Some natural hybrids on the isle of Halmaheira proved to be as vigorous
as the tall coconut, but they fruited earlier (Tammes, 1955). The hybrid can therefore
be used in circumstances where early fruit bearing is of primary importance, for in­
stance in new settlements. Such hybrids can be produced on a large scale by regularly

184
 coconut

cutting away all male flowers from dwarf varieties growing near talis. A garden was
planted for this purpose in North Celebes. Unfortunately, the second generation (Fa)
segregates into all sorts of types, varying from pure dwarf varieties to tall coconuts,
although the hybrid type predominates. Attempts are being made in India and Ceylon
to obtain stable forms combining the good properties of the tall and the dwarf coco­
nut.
When talis are crossed with homozygous dwarfs, prepotency may be more clear and
will be observed at an earlier data owing to the precocity of the hybrids. The method
has been recommended by Harland (1958) and by Zuniga (1963).

BREEDING FOR SPECIAL CHARACTERS

On the island of Ternate there is a particular variety known as the Takome-coconut.

Trees of this variety bear large numbers of very small nuts because no young fruits
are dropped at fruit setting. Many of the progeny also had this characteristic and
special test plots were planted in the 'Mapanget' experimental garden in Celebes and
at Djailolo (Halmaheira). The variety might be important for the export trade, where
the usual nut-size is often too large for consumers.
In recent times attention has been drawn to breeding for resistance to diseases
like Cadang-Cadang in the Philippines, Lethal Yellowing in Jamaica, Maladie de
Kaïncopé in West Africa and Root-Wilt in India. The main difficulty in work of this
kind is that we know so little about the cause and spread of the diseases, inoculation
techniques have not yet been worked out and symptoms often occur in mature trees
only, so that testing is quite difficult and time consuming (c.f. Whitehead, 1968).
Nuts with unusual endosperm characters, e.g. with a jelly like endosperm, or with
a soft endosperm (like boiled rice), often fetch fancy prices at local markets as a delica­
cy in drinks or ice cream. They are known as macapuno coconuts in the Philippines
and kopjor or lilin in Java. Such nuts are only obtained from certain trees and not all
nuts on these trees show the desirable character. In addition, these nuts do not germina­
te because the embryo, though fully developed, has no support in the soft endosperm
and cannot grow through the germ pore. The character is genetically determined
(Zuniga, 1959) and seems to rest on a recessive lethal factor. The expression of this
character may be influenced by the double fertilization (i.e. fusion of pollen nuclei
with the oosphere and with the primary endosperm nucleus) in the same way as is
known for endosperm characters in maize.
It is possible that isolation and cultivation of embryos could be used under special
circumstances and successful embryo culture has already been reported (Abraham et
al., 1962).

MASS SELECTION AND COMMERCIAL SEED PRODUCTION

Many farmers practice mass selection by choosing nuts from heavy yielding trees.

185
P. M. L. TAMMES AND R. A. WHITEHEAD

The nuts to be harvested in the coming year can be seen on the trees and heavy yielders
can be distinguished by anyone having an eye for it.
Selection for practical purposes can be carried out by raising primary seed gardens
f r o m n u t s w h i c h h a v e b e e n o b t a i n e d b y artificially i n t e r c r o s s i n g h i g h yielders, o r -
from prepotent trees. There are such gardens in Celebes, India and Ceylon. Only
high quality trees are retained to produce the seed nuts for secondary gardens, the oth­
ers are removed. Care should be taken to use seed nuts of different origins, to prevent
inbreeding. A suitable size for a secondary seed garden is 2 ha, this will provide suffi­
cient seedlings for the planting of a million palms. Again only the best trees should
be retained. A number of small secondary seed gardens is preferable to a single large
garden as this reduces transport costs. The planting material from these seed gardens
can best be delivered as seedlings about six months old, after selection at the nursery
stage.
In several parts of Java there was a need for good seedlings and large numbers were
sold by the extension service during the years before the war. In certain areas the num­
ber increased to more than 100,000 per year. These nuts were obtained from select­
ed high yielding mother trees in a plantation. Production was based partly on harvest
records and partly on estimates, which are sufficient for these purposes. Furthermore,
the Agricultural Advisory Service in East Java started laying out special seed gardens,
which were planted partly with nuts of the above mentioned selected mother trees,
and partly with nuts from a collection of high yielding trees from the General Agri­
cultural Experiment Station, Buitenzorg. The Buitenzorg palms were raised in 1910
and 1911 from nuts of good trees from the whole Archipelago and this plantation was
selectively thinned in 1930 after yield records had been collected for many years (Eist,
1930).
As coconuts are mainly pollinated by insects coming from trees close by (Tammes,
1937a), an isolation of about 30 metres can be regarded as sufficient for seed gardens;
in Celebes isolation belts of two rows of tall bamboo with a row of trees between them
were used.
The mass selection method also includes the discarding of weak or deficient seed­
lings and nuts which germinate slowly or late. Such seedlings do not grow well, at
least in the beginning, and many of them die in the first year after being planted in the
field. It is, therefore, advisable to start with twice as many nuts as are required to be
planted in the field. This will provide an ample choice of healthy and vigorous seedlings
and some replacement for poor dooers and deaths in the field. Harland (1957) gives
figures on the effect of seedling selection on the yield of the grown-up trees.

Conclusions

1. Cross-pollination and heterogeneity in the tall coconut and lack of methods for
vegetative propagation make it necessary to work along lines of controlled pollination,
coupled with progeny tests.

186
 COCONUT

2. Prepotent trees, with good progeny or good crossing combinations can be used
for breeding tall coconuts.
3. Self-pollination may cause detrimental effects in the tall coconut and should be
avoided, but it may be an object for further research to test inbred lines for high
yield combinations.
4. The occurrence of haploid seedlings in twins may be useful to study crossing
possibility of such haploid trees with the aim of excluding recessive lethal or weakness
factors in one generation.
5. The discarding of weak or slow seedlings is recommended.
6. Dwarf coconuts are often weak, though there are some notable exceptions. They
are inbred forms due to natural self-pollination. Hybrids between tails and dwarfs
are much better. The Fx is homogeneous but the F2 segregates.
Hybridization is being attempted to combine the precocity of the dwarfs with high
yield, larger nuts and vigour of the talis ; some promising results have been obtained.
7. There is a possibility that homozygous dwarfs can be used as mother trees to
test the prepotency of talis. The precocity of the hybrids can shorten the time needed
for testing.
8. Breeding for particular characters e.g. disease resistance, is useful for certain
areas or when special forms or characters are wanted.
9. Coconut breeding is a long term work and progeny testing is the only reliable
method currently available.

References

ABEYWARDENA, V., 1962. The measurement of bienniality in coconut. Ceylon Cocon. Q. 13:112-125.
ABRAHAM, A. and THOMAS, J. K., 1962. A note on the in vitro culture of coconut embryos. Indian
Cocon. J. 15: 84-87.
BERRY, E. W., 1926. Cocos and Phymatocarpon in the Pliocene of New Zealand, Am. J. Sei. 69: 181—
184.
BEUMÉE, J. G. B., 1928. De bloemen van Cocos nucifera L. Trop. Natuur 17: 7-12.
CHARLES, A. E., 1961. Selection and breeding of the coconut palm. Trop. Agric., Trin. 38: 283-296.
CHILD, R., 1964. Coconuts. Longmans, London. 216 pp.
DWYER, R. E. R., 1938. Coconut improvement by seed selection and plant breeding. New Guinea
Agric. Gaz. 4: 24-102.
ELST, P. v.d, 1930. Enige gegevens ten dienste van de klapperselectie. Landbouw Buitenz. 6: 31.
FRÉMOND, Y., ZILLER, R. et DE NUCÉ DE LAMOTHE, M., 1966. Le cocotier. G. P. Maisonneuve and
Larose, Paris. 267 pp.
GANGOLLY, S. R., SATYABALAN, K. and PANDALAI, K. M., 1957. Coconut breeding, a review. Indian
Cocon. J. 10: 32-48.
HARLAND, S. G., 1957. The improvement of the coconut palm by breeding and selection. Circ. Paper.
Cocon. Res. Inst. Ceylon.
JACK, H. W., 1929. Variation in coconuts. Malayan Agric. J. 17: 31-38.
JACK, H. W., 1930. Improvement of the coconut crop by selection. Malayan Agric. J. 18: 30-39.
JACK, H. W. and SANDS, W. N., 1929. Observations on the dwarf coconuts in Malaya. Malayan
Agric. J. 17: 140-165.

187
P. M. L. TAMMES AND R. A. WHITEHEAD

LIYANAGE, D. V. and ABEYWARDENA, V., 1957. Correlations between seed nut, seedling and adult
palm characters in coconut. Bull. 16, Cocon. Res. Inst. Ceylon.
MENON, K. P. V. and PANDALAI, K. M., 1958. The coconut palm, a monograph. Indian Centr. Coco­
nut Cte: 102-110.
NINAN, C. A., PILLAI, R. V. and JOSEPH, J., 1960. Cytogenetic studies in the genus Cocos. Indian
Cocon. J. 13: 129-133.
PIGGOTT, C. J., 1964. Coconut growing. Oxford Univ. Press, London. 109 pp.
RAO, M. B. S., 1955. The dwarf coconut. Indian Cocon. J. 8 : 106-112.
REYNE, A., 1948. De cocospalm. In: Landbouw in de indische Archipel. Vol. 11° p. 459.
SATYABALAN, K. and LAKSMANACHAR, M. S., 1960. Coconut breeding: effects of some breeding pro­
cedures. Indian Cocon. J. 13: 94-100.
SWAMINATHAN, M. S. and NAMBIAR, M. C., 1961. Cytology and origin of the dwarf coconut. Nature,
Lond., 192: 85-86.
TAMMES, P. M. L., 1936. Productiecijfers van klappers. Landbouw Buitenz. 12: 147-150.
TAMMES, P. M. L., 1937a. Over den bloei en de bestuiving van den klapper. (On the flower biology of the
coconut). Landbouw Buitenz. 13: 74-89.
TAMMES, P. M. L., 1937b. Over de factoren welke de vruchtdracht van den klapper bepalen. (On the
factors determining the fruit production of coconut trees). Landbouw Buitenz. 13: 260-269.
TAMMES, P. M. L., 1940. Over de ontwikkeling van de vrucht van den klapper en de factoren, welke
van invloed zijn op de hoeveelheid copra per noot. (On the development of the coconut fruit and
the factors which influence the copra content of the nuts). Landbouw Buitenz. 16: 385-393.
TAMMES, P. M. L., 1955. Review of coconut selection in Indonesia. Euphytica 4: 17-24.
WEBSTER, C. C., 1939. A note on a uniformity trial with oil palms. Trop. Agric., Trin. 16:15-19.
WHITEHEAD, R. A. and CHAPMAN, G. P., 1962. Twinning and haploidy in Cocos nucifera. Nature,
Lond. 195: 1228-1229.
WHITEHEAD, R. A., 1963. The processing of coconut pollen. Euphytica 12, No. 2: 167-177.
WHITEHEAD, R. A., 1965a. The flowering of Cocos nucifera Linn, in Jamaica. Trop. Agric., Trin. 42:
19-29.
WHITEHEAD, R. A., 1965b. Speed of germination, a characteristic of possible taxonomie significance
in Cocos nucifera Linn. Trop. Agric. Trin. 42: 369-372.
WHITEHEAD, R. A., 1966a. Survey and collection of coconut germ plasm in the Pacific Islands. H.M.S.O.
London, 78 pp.
WHITEHEAD, R. A., 1966b. Some notes on dwarf coconut palms in Jamaica. Trop. Agric., Trin. 43:
277-294.
WHITEHEAD, R. A., 1968. Selecting and breeding coconut palms resistant to Lethal Yellowing disease.
Euphytica 17: 81-102.
ZILLER, R., 1962. La sélection du cocotier dans le monde. Oléagineux 17: 837-846.
ZuniGA, L. C., 1959. Studies on Makapuno bearing trees. I. Segregation of the nut endosperm of
artificially pollinated Makapuno-bearing and ordinary coconuts. Philip. J. Agric. 29: 51-67.
ZuniGA, L. C., 1963. Report on coconut improvement using local and foreign varieties and strains.
Proc. Symp. Cadang-Cadang. Manila.

188
COFFEE

Coffea arabica L. and Coff'ea canephora Pierre ex Froehner

A. CARVALHO1, F. P. FERWERDA2, J. A. FRAHM-LELIVELD3, D. M. MEDINA4, A. J. 1.

MENDES5 and L. C. MONACO6

Systematics

TAXONOMY AND GENETIC RELATIONSHIP

The genus Coffea has attracted the attention of many taxonomists in the past. The
most extensive and reliable taxonomy work on this genus was carried out by Chevalier
several years ago (Chevalier, 1947). According to this author, the valid coffee species
can be grouped into the following four sections: Eucoffea (24 spp.), Mascarocoffea
(18 spp.), Paracoffea (13 spp.) and Argocoffea (11 spp.). They are all native to the
tropical and subtropical regions of Africa and Asia. The Eucoffea section comprises
five subsections, as follows :
Section Subsection Species
C. arabica
C. congensis
Erythrocoffea C. canephora
C. eugenioides
C. liberica
C. klainii
Eucoffea Pachycoffea C. oyemensis
C. abeokutae
C. dewevrei
Nanocoffea 5 species
Melanocoffea 3 species
Mozambicoff'ea 7 species

Very little is known about the genetic affinities between the species of the genus
Coffea. Studies of the crossability and of the chromosomal homology have been li-

1),4),6)-6)- Instituto Agronômico, Campinas, Brazil.

2) Institute of Plant Breeding, Wageningen, The Netherlands.
3) Department of Tropical Crop Husbandry, Wageningen, The Netherlands.

189
A. CARVALHO, F. P. FERWERDA, ET AL.

mited to a small number of species existing in a few living collections. The species of
the subsection Erythrocoffea have been studied more extensively than any others.
C. arabica was shown to be more closely related to C. eugenioides than to C. cane-
phora or to C. congensis (Monaco and Carvalho, 1964). Hybrids between C. arabica
and canephora are highly sterile, not only owing to genetic differences, but also owing
to their triploid nature.
Very low chromosome pairing is observed in the triploid hybrids (Mendes, A. J. T.,
1958). Hexaploids or tetraploids (arabica X 4 n canephora) are more fertile. C.
canephora and C. congensis seem to be closely related species, as a fairly high seed
set is obtained in some combinations (Cramer, 1957).
Hybrids between C. arabica and C. eugenioides are also usually sterile because of
their triploid condition, but there is good evidence for the presence of several loci
which are dominant for the corresponding alleles of C. arabica. On the basis of both
the behaviour of these hybrids and the geographical distribution of C. eugenioides, it
has been suggested (Monaco and Carvalho, 1964) that this species might have partici­
pated in the origin of C. arabica. Narasimhaswamy (1962) claimed that C. eugenioides
and C. liberica may be the probable ancestors of the Arabica coffee. On the other
hand, it was considered (Cramer, 1957) that C. congensis presents many character­
istics which indicate that it may also have contributed to the formation of the poly­
ploid species.
Crosses between C. dewevrei and C. liberica have produced fertile hybrids. Despite
some morphological differences between these two species, it should be recalled
that some of their variants seem to overlap. No clear-cut separation in their distribu­
tion seems to exist and they occur sympatrically in the region of the Congo. However,
they should be maintained as separate entities until the wild populations have been
better studied.
The place assigned to kapakata coffee is somewhat controversial. Chevalier (1947)
arranged it under the genus Psilanthopsis, but the fact that it is easily crossable with
C. canephora and other Coffea species (Lambers, 1935; Leliveld, 1938, 1940; Car­
valho and Monaco, 1959) suggests a closer relationship to the genus Coffea. This
made the latter authors propose its inclusion in the subsection Mozambicoffea. The
hybrids between C. kapakata and C. canephora are particularly vigorous and fairly
fertile.
In a recent survey several other species of coffee have been described (Leroy, 1962;
Portères, 1962) most of them native to Madagascar.

COMMERCIAL SPECIES

The species C. arabica, C. canephora and C. liberica are the only ones extensively
cultivated at present.
Coffea arabica L. is native to the south-western region of Ethiopia at altitudes be­
tween 1000 and 2000 m, lat. 5-8° North and long. 34-38°. East of Greenwich and,

190
 coffee

possibly, also to adjoining territories of south-eastern Sudan. It is the only species

cultivated in Latin America and to a lesser extent in Africa. Approximately 80 % of the
total world exportable coffee production belong to this species.
It should be pointed out that C. arabica is the only polyploid species so far describ­
ed in this genus. Various speculations have been made about its origin, but there is
still very little reliable evidence about the nature of the polyploidy involved. On the
basis of the pairing observed in haploids it has been suggested that C. arabica could be
an autotetraploid. On the other hand, the meiotic behaviour of some interspecific
hybrids and the mode of inheritance of duplicate factors were used by Monaco and
Carvalho to indicate that C. arabica is an allopolyploid or perhaps a segmental allo-
tetraploid. Its geographical distribution is characteristic of polyploids, as this falls
almost completely outside the range of distribution of the diploid species.
C. canephora has a very wide geographic distribution extending from the western
to the central tropical and subtropical regions of the African continent, from Guinea
and Liberia to the Sudan and the Uganda forest, with a high concentration of types
in the Republic of Congo Kinshasa. The commonest forms occur in regions with a
low altitude. The species has, however, a great range of adaptation. C. canephora,
or robusta coffee as it is commonly called, because of its mode of reproduction is
highly polymorphic, and it is very difficult to characterize its varieties. Several va­
rieties have been described for C. canephora, some of which should be considered
cultivars (Leon, 1962).
Coffea liberica occurs more frequently in the south of Guinea, Ivory Coast, Liberia,
Ghana, Gabon, Congo and in the north of Angola. Its forms usually constitute
characteristic elements of the tall and dense forest. This species is also self-sterile and
highly polymorphic. It is cultivated to a small extent in Liberia, Surinam and a few
other areas.
All other coffee species are cultivated to a rather limited extent or are found only
in coffee collections maintained in a few coffee research centres.
When grown undisturbed, the Arabica coffee plant may attain a maximum height
of 4.5 to 5 m. A greater height may be attained by other species such as C. dewevrei
and C. liberica. Certain techniques usually applied in the plantations, such as several
types of pruning, force the plant to develop into a much smaller shrub.
Arabica and Robusta coffee usually take three to four years to reach the fruiting
stage, while liberica and dewevrei may require four to five years. The coffee tree can
live to a rather great age, cases of one-hundred-year-old Arabicas being known. The
economic life span is, however, not much longer than 25 years or even less.
Vegetative reproduction is possible by grafting and softwood cuttings, but it must
be kept in mind that coffee has both orthotropic and plagiotropic branches and that
only the orthotropic branches, when used as scions and cuttings, give rise to normal
upright-growing plants. Brief details of the methodology of vegetative reproduction
have been given by Fernie (1958) and Robinson (1964). In some species of coffee the

191
A. CARVALHO, F. P. FERWERDA ET AL.

juvenile stage can be shortened by grafting young seedlings upon shoots of full-grown
trees. Thus, a breeding cycle can be shortened by one or two years.
The variability, both morphological and physiological, existing among coffee spe­
cies is remarkable, offering excellent opportunities for the breeding of new types.
The establishment of living collections of Coffea species and of related genera which
are as complete as possible is urgently needed to enable a better insight to be obtained
into the variability within each species and to allow a study to be made of the existing
interspecific relationships. This aspect is of great importance not only for the study
of the evolution within the genus Coffea, but also as a source for the genetic variation
of yet unknown favourable characteristics of economic importance for future breeding
work. Of primary importance would be the establishment of an extensive collection of
C. arabica variants in Southern Ethiopia.

THE MODE OF REPRODUCTION AND ITS CONSEQUENCES FOR BREEDING

As regards the mating system, the genus Coffea includes two contrasting categories.
Coffea arabica, the most important representative of the tetraploid coffee species, is
predominantly self-fertilizing, while the diploid coffee species are distinct cross-ferti­
lizers, as far as can be judged from studies in this respect.
This essential difference in sexual reproduction implies that arabica coffee requires
a breeding system which is quite different from that of C. canephora and the other
allogamous coffee species. The breeding of arabica coffee and of C. canephora will
be discussed in the following sections.
As an introduction to breeding proper some basic data on cytology, gametogenesis
and the development of seed and fruit will be given, which apply to both the Arabica
group and the Canephora group.

Cytology, gametogenesis and development of seed and fruit

(J. A. Frahm-Leliveld, D. M. Medina and A. J. T. Mendes)

The information available on cytology, gametogenesis and seed and fruit develop­
ment in the genus Coffea has mainly been derived from research carried out as an
aid to breeding.
With the exception of the old investigations by von Faber(1912) nearly all publica­
tions date from 1930 onwards.
The main centres of cytological and embryogenetic research are: 1. Brazil;
2. South-East Asia; 3. Central Africa. The research activities in Central Africa com­
prise the French investigations made in Ivory Coast and Equatorial Africa in addition
to Belgian research in Congo which has been continued at Louvain of recent years.
Sybenga (1960) should be credited with a clear and comprehensive survey of litera­
ture on genetic and cytological research till 1960.

192
 coffee

Cytology

The basic chromosome number for the genus Coffea is x = 11 and two groups
of species are distinguished: one diploid with 2n = 22 and another tetraploid with
2n = 44 chromosomes.
A complete list of the chromosome numbers of Cofl'ea species and varieties has been
presented by Sybenga (1960).
The diploid group includes the species C. canephora, C. dewevrei, C. congensis,
C. eugenioides, C. racemosa and C. stenophylla, all of them being self-incompatible.
Recently twelve less known wild coffee species from Malagasy have also been reported
to have 2n = 22 (Leroy and Plu, 1966).
The only economically important representative of the tetraploid group is C. arabica
which is self-fertile and comprises several varieties and cultivars. The type of inheri­
tance of the tetraploid C. arabica is disomic and in meiosis only bivalents are observ­
ed. This justifies the conclusion that C. arabica is an allotetraploid. Di - haploids
(2x = 22), hexaploids (6x = 66) and octoploids (8x = 88) occasionally occur in the
progenies of normal 44-chromosome plants. Artificial and natural pollinations among
these polyploids resulted in plants with 55, 66 and 88 chromosomes. Aneuploids
(2n = 43, 45, 46, 53, 54, 58 etc.) have been encountered in the progenies of normal
44-chromosome plants, di-haploids and polyploids (Mendes, A. J. T., 1947, 1958;
Monge, 1962).

CHROMOSOME RELATIONSHIPS IN INTERSPECIFIC HYBRIDS

According to Bouharmont (1959), the somatic chromosomes of C. arabica (2n = 44)

and the diploid (2n = 22) species such as C. canephora, C. liberica, C. racemosa and
others show a very similar morphology. A mean idiogram has been constructed for
the genus, based upon the observations of 10 species. The only three species that
show an idiogram somewhat different are C. abeokutae, C. lebruniana and C. hors-
fieldiana.
The difficulties of crossing C. arabica with diploid species must mainly be caused by
differences in chromosome number.
Triploid (2n = 33) hybrids between C. arabica and C. canephora, C. racemosa and
C. kapakata have been obtained and studied cytologically (Krug and Mendes, 1940;
Medina, 1963; Monaco and Medina, 1965). The chromosome associations at dia-
kinesis and metaphase I showed that there are more trivalents in the arabica X cane­
phora hybrid than in the other two hybrids; the number of bivalents, however, is smal­
ler in the former. The chromosome distribution during anaphase, the formation of
abnormal tetrads and the number of chromosomes of the pollen grains are very similar
in these three hybrids. Most of the pollen grains formed have 13-17 chromosomes in
the arabica X racemosa and arabica X kapakata hybrids and 12-16 in the arabica X
canephora hybrid. All these hybrids are highly sterile. Very seldom do the tripoids set

193
A. CARVALHO, F. P. FERWERDA ET AL.

seed when back-crossed to C. arabica either through open or controlled pollination.

Plants obtained from open pollinated seeds of C. arabica X C. canephora hybrids,
have indicated that some triploids give rise to seedlings having mostly a chromosome
number around 44, while in other hybrids this number is mostly around 55 (Mendes,
A. J. T., 1951). Numerous aneuploids have been found among these progenies.
Leliveld (1940) studying the chromosome pairing relationships in a number of
hybrids between C. arabica and the diploid species C. laurentii, C. liberica and conu-
ga arrived to similar conclusions. The limited affinity between the genome of C. ara­
bica and those of the diploid species is reflected in the high proportion of univalents
that is almost invariably observed.
The artificially induced doubling of the chromosome number, of a C. arabica x
C. canephora hybrid (2n = 33) has resulted in a change from sterility to almost com­
plete fertility in the resulting hexaploid. The partial sterility of this hexaploid may be
caused by irregularities still occurring in meiosis, which are supposed to be due to
certain similarities among the chromosomes of C. arabica (Mendes, A. J. T., 1947).
One spontaneous tetraploid (2n = 44) hybrid which probably originated from C.
arabica x C. dewevrei has also shown irregularities in meiosis but it produces a good
deal of fertile pollen and gives high yields.

MICROSPOROGENESIS

The process of development of the pollen grains follows the general pattern of the
dicotyledons. About 36 hours after a blossom inducing rainshower meiosis has been
completed and the young pollen grains are arranged in tetrads. At anthesis they have
attained their definite size and are provided with an exine (Dublin, 1957, 1958;
Leliveld, 1938,1939; Mendes, A. J. T., 1950).
Microsporogenesis in interspecific hybrids as a rule proceeds abnormally owing to
meiotic disturbances caused by the lack of homology between the chromosomes of
different genomes, as has already been mentioned before. In most cases they cause a
high percentage of defective pollen grains.

MACROSPOROGENESIS

Macrosporogenesis and the development of embryo and endosperm proceed simi­

larly in C. arabica and C. canephora. Brazilian research workers focussed their inter­
est mainly in C. arabica. The diploid species were the main subject of research in S. E.
Asia and Africa (Fagerlind, 1937, 1939; Heyn, 1938; Leliveld, 1939; Lebrun, 1941;
Dublin, 1957a, b, 1958; Bouharmont, 1959; Moens, 1965a).
The ovule of C. arabica consists of an embryosac mother cell covered by a single
layered nucellus and one integument; the former disappears as the ovule grows
(Graner, 1936, 1938). The macrospore mother cell gives rise to four macrospores of
which the chalazal one becomes functional. After two successive divisions a seven

194
 COFFEE

celled, eight nucleate embryosac develops. Usually, in C. arabica, the embryosac is

ready for fertilization at anthesis. In C. canephora the embryosac reaches the eight cel­
led stage in the course of the first day of flowering (Leliveld, 1938; Mendes, C. H. T.,
1950) and in C. dewevrei its formation is still a little more delayed (Dublin, 1958;
Medina, 1960).

EMBRYO AND ENDOSPERM DEVELOPMENT

The need of a systematic embryogenetic investigation was particularly felt in the

middle thirties when the phenomena of premature fruit dropping, unilateral seed
abortion in Robusta coffee and arrested endosperm development in various inter­
specific hybrids required an explanation.
In Indonesia these phenomena were first investigated by Leliveld (1938) for C.
canephora; independently Mendes, A. J. T. (1941) in Brazil studied the same processes
for C. arabica. The development runs largely parallel in the two species and the gener­
al picture can be characterized as follows. After the pollen tubes penetrated into
the embryosac the double fertilization is effectuated immediately in C. arabica. In
C. canephora this fusion is somewhat slower in taking place ranging from one day to
a week (Leliveld, 1938; Moens, 1965) after pollination.
The fertilized embryosac increases in volume and compresses the inner integument
cells but remains inactive in other respects. The outer cells of the integument multiply
actively, mainly by tangential divisions, giving rise to a transitorial perisperm which
pushes back the swollen embryosac and eventually attains the size of a normal seed
occupying the entire locule for some time (Mendes, A. J. T., 1941 ; Leliveld, 1938).
This mode of development initially led Houk (1936) and Krug (1937) to misinter­
pret the nutritive tissue of the coffee bean as being perispermic in origin.
Inside the embryosac the synergids and slightly later also the antipodals degenerate.
The zygote stays near the micropyle in a resting stage while the primary endosperm
nucleus starts dividing approximately four weeks after successful pollination. The
endosperm which originally consists of free nuclei develops into a cellular structure
after a series of nuclear divisions occurring in rapid succession. As the number of cells
increases the endosperm takes a disc shape.
The first division of the zygote takes place 60-70 days after anthesis. The milky
endosperm now rapidly increases in volume and gradually replaces the hyaline gree­
nish perisperm. By the time the endosperm entirely fills the cavity earlier occupied by
the perisperm, the embryo is well differentiated into a hypocotyl with radicle and two
small cotyledons.
In the ripe seed, the remains of the perisperm form the 'silver skin" which envelops
the now hard endosperm. The parchment layer enveloping the seed is the endocarp
(see fig. 1).
These investigations showed that the coffee bean contains real endosperm, a fact
corroborated by the other investigators (Fagerlind, 1939). The nature of the endo-

195
A. CARVALHO, F. P. FERWERDA ET AL.

Fig. 1 Cross section of a normal berry (A) and a peaberry (B) a exocarp, b mesocarp (pulp),
c endocarp (parchment), d silver skin, e endosperm. Adapted from Leliveld, 1938.

sperm has also been demonstrated genetically (Krug and Carvalho, 1939) making
use of the recessive 'cera' (yellow endosperm) mutant which clearly shows xenia when
pollinated by normal green endosperm plants.
The physiological aspects of the growth of the coffee berry in connection with the
embryogenetic differentiation have been extensively reviewed by Leon and Fournier,
(1962) and by Wormer (1964,1966).

ABNORMAL DEVELOPMENT OF FRUIT AND SEED

Peaberries

When one ovule aborts early in the fruit development the seed of the other locule
develops freely and assumes a pea like shape (peaberry, sp. caracolillo). Although
such seeds are not considered to be defective in a commercial respect, they cause a
decrease in yield owing to the absence of the other seed.
Peaberry formation is generally the result of lack of fertilization caused either by
inviability of the ovule itself or by inadequate pollination. The latter is the case in the
self-incompatible diploid species when windless cloudy weather hampers the tree to
tree pollen transport so that ineffective self-pollination prevails (Ferwerda, 1948;
Sybenga, 1960).
In C. arabica peaberry formation must be related almost exclusively to ovule invia­
bility. In di-haploid arabica plants (2x = 22) where inviable gametes are produced in
high proportion very few, practically always one seeded fruits are formed even when
the ovules are fertilized by normal pollen. The high frequency of peaberries obtained
from irradiated arabica plants indicates that ovule disturbances are most likely to
be responsible for this anomaly (Monge, 1962).

196
 COFFEE

5)

Fig. 2 General appearance and cross sections,

of endosperms showing stages of reduction (A,
• B, C) in beans of liberica-arabica hybrids as
compared with the endosperm in a normal bean
(D). Endosperm reduction may proceed much
further than is indicated in A. From Ferwerda,
1948.

Empty loculi

When endosperm development is arrested an entirely or partly empty locule results.

In such cases fruit development is apparently normal and the fruits cannot be distin­
guished externally from those containing normal seeds. If the endosperm arrest com­
prises both loculi an entirely seedless fruit results (Mendes, A. J. T., 1946). Depending
on the time at which the interruption of the endosperm development occurs and on
the nature of the abnormality involved, the size of the endosperm mass may vary
from a pellet only a few millimeters in diameter up to an almost completely developed
bean. Even the smallest malformed endosperms present the same colour and texture
as the fully developed ones (fig. 2). In some cases they contain an embryo, sometimes
they are embryoless (Leliveld, 1938).
Defective embryos appear in all coffee species and varieties but are most frequently
found in interspecific hybrids and autopolyploids.
A special case of endosperm abnormality has been found in the Mundo Novo
cultivar of C. arabica. In this case endosperm development in a certain percentage of
the ovules is arrested in an early stage occasionally resulting in a disc shaped body
about three millimeters in diameter. This anomaly is genetically conditioned and rests
on one recessive gene (Mendes and Medina, 1955).

197
A. CARVALHO, F. P. FERWERDA ET AL.

Fig. 3 Open flowers of C. arabica.

The breeding of arabica coffee

(A. Carvalho and L. C. Monaco).

BIOLOGY OF REPRODUCTION

Periodicity of flowering

In regions with a pronounced and markedly contrasting dry and rainy season, as
found in the state of S. Paulo, Brazil, main flowering occurs once or twice, and seldom
more frequently, soon after the end of the dry season. A large number of smaller
'flower showers' is characteristic of regions with a more evenly distributed rainfall.

Inflorescence, flower andfruit characteristics

The coffee inflorescence consists of very short lateral axes with short pedicellate
flowers disposed in axillar glomerules, with generally three to five in each glomerule.
There are two pairs of bracteoles in each flower set (fig. 3). The calyx is very rudimen­
tary, being five-denticulate, and the corolla is white, the five petals being united in a
tube to form a salver-shaped corolla. The stamens are epipetalous and the anthers are

198
 COFFEE

Fig. 4 A. Selected tree of C. arabica cv. Mundo Novo in full bearing. B. Fruiting branches with
ripe or near ripe cherries.

bilocular, opening lengthwise. The pollen grains are numerous and globose in shape.
The pistil is represented by an inferior ovary, a terminalstyle and two stigmatic branches.
The ovary is normally bilocular, each locule bearing a single anatropous ovule. The
coffee fruit is a drupe normally containing two seeds (fig. 4). Ripe fruits have a fleshy
and thick pericarp. The endocarp is thick and in the ripe fruit constitutes the so-called
'seed parchment'. The coffee seed is elliptical, plane-convex, possessing a longitudinal
furrow on the plane surface. The seed coat is represented by the so-called 'silver skin'.
The endosperm is formed by cells with very thick walls, the hemicellulose functioning
as food storage. The small embryo located at the bottom of the seed is represented by
a hypocotyl and two adherent cordiform cotyledons (Dedeca, 1958).
The flower bud under normal conditions generally opens on sunny days early in the
morning and pollen-shedding starts shortly afterwards ; the stigma is receptive at the
opening of the bud and remains receptive for three to four days, depending upon the
weather conditions.

The flower morphology is such as to permit natural self-pollination, and all arabica
varieties are self-fertile. Pollen tubes take approximately 24 hours to reach the ovary.
Arabica pollen can be kept alive for as long as 49 days at 5 °C and at reduced humidity,

199
A. CARVALHO, F. P. FERWERDA ET AL.

while C. canephora pollen can be maintained alive under these conditions for only
seven days.

Mating system

As already mentioned, arabica coffee is preponderantly autogamous. This fact

has been ascertained by research workers in various centres of arabica cultivation.
The structure of arabica flower is such as to allow self-pollination to take place even
inside the closed bud when anthesis is retarded on cloudy days.
Apart from self-pollination some spontaneous outcrossing may occur. The rate of
natural cross-pollination varies with the weather conditions, which have an influence
on the action of the agents of pollination, viz. insects, wind and gravity (Carvalho and
Krug, 1949; Krug, 1935; Nogueira, Carvalho and Antunes, 1959).
The incidence of natural cross-pollination in the mutant purpurascens has led earlier
research workers (Taschdjian, 1932) to conclude that C. arabica is allogamous.
Extensive studies made at Campinas, Brazil, with the cera marker gene (yellow endo­
sperm) gave more accurate information about the rate of outcrossing in this mutant.
It was found for the years 1945/1947 that 7.3-9.0% of the seeds originate from out­
crossing, a figure which is largely corroborated by information compiled over the years
1954—1965. Slightly different percentages of outcrossing were observed in some other
mutants (Carvalho and Monaco, 1962).
Roughly half (4.8 to 5.3%) of the above 9% of outcrosses are caused by gravity,
wind and insects being responsible for 2 to 5 % and 0 to 2 %, respectively.
Seeds collected in 1964 and 1965 from the cera mutant in a region with a climate
markedly differing from that at Campinas revealed a much higher rate of spontaneous
crossing, 20.5 % of the seeds being of hybrid origin. From Puerto Rico Dhaliwal (1965)
reported for the cera mutant 8.6 % outcrosses.
Results of studies on the rate of outcrossing are few and far between for other coffee
regions, particularly for Africa. Of particular significance would be the analysis of
the results collected in Ethiopia, which is the centre of variability for C. arabica. We
should expect the rate of outcrossing in the centre of origin to be higher, as has been
shown for other crops (Rick, 1950). Meyer (1965) indicated that studies carried out
at the Jimma Experiment Station in Ethiopia gave 40-60 % of outcrossing for C. arabica,
which seems excessively high, however.
This rate of natural cross-pollination found in C. arabica is sufficient to substantiate
a certain degree of heterogeneity in the coffee plantations, a finding which is of value
for selection programmes.

Crossing techniques

Emasculation of the arabica flower buds must be done one or two days before anthe­
sis (fig. 5). In order to emasculate a large number of flowers in a short period of time

200
 coffee

Fig. Flower buds of C. arabica the candle

a pair of scissors is used which has bsen especially adapted to this kind of work. A
small nick is made on each blade and the aperture so formed is never entirely closed.
The scissors are applied at the base of the corolla tube and will cut only this tube, leav­
ing the style untouched (fig. 6). With a slow upward movement one pulls off the
corolla and with it the inserted stamens are removed without touching the sensitive
stigma. The whole branch is then covered with a paper bag. Branches of the plant
which will furnish the pollen are brought to the laboratory a day before anthesis
and kept in a container with water and protected with paper bags to prevent any
contamination. The flowers are transported to the field in Petri dishes. With the
help of a forceps, which is dipped into alcohol before each pollination, the anthers
are rubbed on the stigma of the emasculated buds (fig. 7). One flower may be used for
every five stigmata to be pollinated. After pollination, the bag is closed again and left
on the branch for 10-15 days. Weekly inspections are necessary to remove new flower
buds which will develop on the same leaf axils (Carvalho, 1958; Krug, 1935).

For artificial selfing, branches with fully developed but still unopen flowers are
protected with paper bags for about 10-15 days. Any small flower buds or young
fruits derived from previous flowerings must be removed before the branches are cover­
ed. The fruit-set percentage in C. arabica is about 50 %, while in unbagged branches

201
a. carvalho, f. p. ferwerda et al.

Fig. 6 Emasculation of C. arabica flowers.

this percentage can be higher, up to 70 %, depending on the climatic conditions pre­

vailing during the days of pollination.

GENETIC ANALYSIS

Coffea arabica is the only coffee species which has been submitted to a rather ex­
haustive genetic analysis. About forty genes have already been described, while a few
others are being analysed. The results already obtained will prove increasingly useful
in colfee breeding programmes.

Arabica variety as a standard

It would be superfluous to emphasize the need for establishing a standard type for
genetic analysis in Coffea arabica. Whereas, it might be difficult to choose a proper
standard type for several other plant species, the choice was relatively easy in the case
of C. arabica. The variety arabica (= C. arabica L. var. typica Cramer) has been select­
ed from the very beginning as a standard, not only because it was used by Linnaeus as

202
 COFFEE

Fig. 7 Artificial cross-pollination of C. arabica.

the prototype of the species, but also because its morphological characteristics were
well known and because it was of economic importance.
It would be desirable to use the arabica variety also as a standard in any other ex­
perimental stations where the genetic analysis of C. arabica would be carried out. A
uniform standard would facilitate both the characterization of dominance relation­
ships and the uniform use of genetic symbols.
Besides being used as a standard in routine genetic analysis, the arabica variety has
also been included as a check in progeny trials.

203
A. CARVALHO, F. P. FERWERDA ET AL.

Procedures for genetic analysis

Selfed and crossed seeds are sown in a nursery, the illumination of which is main­
tained at 50% by means of artificial shade. Approximately one year later it is already
possible to score the seedlings for the purpose of studying some of the genetic factors.
Progenies segregating for genes controlling the branching system, leaf shape or colour
need to be transplanted to areas with larger spacings (40 x 30 cm) and are classified
twelve to fifteen months after germination. Mutants with changed flower, fruit and
seed characteristics require a still larger spacing (80 x 50 cm), which will allow them
to be maintained for three to four years until the classification has been completed.
Plants selected for the continuation of the genetic analysis are transplanted to the field
with a spacing of 2.0 x 2.5 m, where they are observed for variable periods of time.
A detailed genetic analysis in the case of coffee takes about sixteen or more years, to
complete.

Mutants of C. arabica

Coffea arabica is a relatively stable species. So far, only 40 mutants have been stu­
died out of hundreds of thousands of seedlings and adult plants examined in nurseries
and in plantations. In spite of having twice as many chromosomes (2n = 44) than the
other species of the genus, almost all observed mutants are conditioned by single fac­
tor differences or by interactions between them.
For a detailed description of the various mutants reference can be made to Carvalho
(1958); Krug and Carvalho (1951). In the present chapter only a few mutants of
some importance to breeding will be mentioned.

Mutants affecting plant habit

Maragogipe (symb. Mg) having a strong pleiotropic effect on the production of tall
and vigorous plants, large leaves, flowers, fruit and seeds.
Caturra (symb. Ct) with a reduced plant size, short internodes, wide, large, dark
green leaves and a high yielding capacity. Säo Bernardo (symb. SB), with a reduced
plant size, short internodes and elliptic leaves, as in arabica.

Mutants affecting seed characters

Laurina (symb. Ir), having a strong pleiotropic effect on the shape of the plant, the
branching system, the size and shape of the leaves and of the seeds, which are narrow
and pointed at one end or at both ends.
Mokka (symb. mo), having very small elliptic leaves with large and prominent do-
matia. The internodes are short and the fruits small and roundish and containing round

204
 COFFEE

seeds. The double recessive Irlrmomo characterizes the mokka variety described by
Cramer (1957).
Cera (symb. ce). The factor cera is responsible for the colour of the endosperm.
Some of the famous coffee of Yemen possibly belongs to the cera type, according to
Krug (1959).
Discoid (symb. di) causes abortion of the endosperm, which is reduced to a small disc.
There is evidence that this type of endosperm abortion is controlled by at least two
complementary factors.

Ploidy variation

C. arabica, being a tetraploid, has been shown to be capable of supporting ploidy

variation without any great effect on plant development. Haploids occur at a frequen­
cy of about one in every 10,000 seedlings. They are almost completely sterile. Pro­
genies from duplicated haploid plants are normal. Since they are isogenic, they havé
been used to study the effects of X-rays on coffee seeds and to check in progeny trials
the evaluation of the environmental effect on quantitatively inherited characteristics.
The haploids seem to be more resistant to low temperatures, to long periods of drought
and to the attack of Leucoptera sp. They have been described as the monosperma va­
riety, although such a classification is no longer accepted.
Polyploid seedlings with 2n — 66 or 2n = 88 chromosomes, which have been ob­
served in most varieties of C. arabica present broader and thicker leaves. Their sterility
is high and the fruits are usually seedless. They have been found in nurseries and in
coffee plantations somewhat more frequently than haploids. They have been describ­
ed as the bullata variety.

Mutants obtained by X-rays

Seeds of I4 and I5 generations of arabica and bourbon varieties obtained by succes­

sive selfings and also seeds of isogenic lines of bourbon have been treated with X-rays.
It has been observed that seeds treated with more than 23,000 r start germination but
the young seedlings do not develop. At higher doses the rootmeristem is apparently
completely destroyed, since no root growth is evident after germination. The LD 50
for coffee seeds seems to be around 15,000 r. Most of the mutants obtained are partial­
ly sterile and characterized by leaf abnormalities with regard to shape and chlorophyll
distribution. The branching habit is also affected. Some plants are polyploids. No
visible chimeras have so far been found among the abnormal plants. As the mutations
involve whole plants, it was suggested that the growing point of the coffee plant is one-
celled (Moh, 1961). The M2's of some of these treated plants segregate in one or more
types of angustifolia. Angustifolia mutants have also been obtained in Turrialba as
a result of the direct treatment of seedlings of the arabica variety. It has been shown
that this variety is more sensitive to radiation than is bourbon (Cevallos, 1963).

205
A. CARVALHO, F. P. FERWERDA ET AL.

Somatic mutations

The somatic mutations reported for C. arabica involve increases and reductions in
the chromosome number and other known genetic factors. Bullata plants with 88
chromosomes have been found to give branches with 44 chromosomes and haploid
plants with 22 chromosomes have also been described, which spontaneously give
branches with 44 chromosomes. In some dwarf coffee plants (ttnana) the na allele is
unstable in the somatic tissue from which vigorous branches or sectors of ttNana
constitution originate. In other plants the mutation nana (Antunes and Carvalho,
1954) to Nana seems to constitute a sectorial chimera affecting only part of the cell
layers which will give rise to the gametes.

Mutants from Ethiopia

The arabica material which has lately been collected in Ethiopia was found to con­
tain several of the genetic factors previously encountered in the South American va­
rieties and, in addition, other new alleles not yet described.
The rather intensive variability encountered among the coffee seedlings received in
Campinas from Ethiopia (Carvalho, 1959; Carvalho, Monaco, Scaranari, 1962;
Lejeune, 1958; Monaco, 1964; Monaco, 1966; Sylvain, 1953) may be the result of a
non-randomized procedure of seed collection. Seeds have very often been intentional­
ly harvested from the more or less conspicuous variants occurring in the Ethiopian
forests. On the other hand, it seems reasonable to assume that a higher rate of natural
cross-pollination, apparently occurring in the native C. arabica, is also responsible
for this marked genetic heterogeneity.
The occurrence of several new genetic factors in the Ethiopian coffee indicates that
further exploration for coffee mutants in Ethiopia is highly desirable for the purpose
of throwing more light on the genetics of Coffea arabica and also of providing more
basic material for its improvement.

GENETIC VARIABILITY

A survey carried out of the primitive coffee population in Brazil and other Latin
American countries revealed that they represent a very restricted gene pool for the
arabica species. Most of the coffee orchards in Latin America have been planted with
seeds derived from a few trees of Coffea arabica var. arabica which had been
brought from Java to the Botanical Garden of Amsterdam and later to the one in
Paris.
Owing to its particular origin and also to its mode of reproduction, little genetic
variability is to be found among the many millions of coffee trees belonging to this
species grown in Latin American countries. However, later on, introductions were
made, at least in S. Paulo, Brazil, of other commercial varieties such as bourbon

206
 COFFEE

(C. arabica L. var. bourbon), Sumatra (C. arabica 'Sumatra') and others. The bour­
bon population is also represented by a large progeny derived from a few plants acci­
dentally brought to Brazil in 1864, the same having occurred in the case of the Su­
matra cultivar, which was introduced a little later in 1896. By virtue of the origin of
these coffee populations a fairly small degree of genetic variability has arisen, a fact
which apparently contradicts the relative success obtained by the pedigree method of
selection which has been used for C. arabica in various areas. Mutations may have
played an important role in the evolution of the cultivars of this species, since several
major genes have been incorporated into the genotype of these commercial varieties.
However, the mutations occur at a low frequency and, consequently, they cannot en­
tirely account for the diversity of the favourable genotypes already isolated from the
original coffee populations brought to this continent. On the other hand, the occur­
rence of an approximate rate of 10% of cross-pollination in this species at Campinas
would not only ensure the maintenance of a certain degree of heterogeneity, but would
also create new combinations as a result of hybridization between the different in­
troductions. The Mundo Novo cultivar, for instance, is derived from selections
carried out within a naturally segregating population, probably involving the Bour­
bon Vermelho and Sumatra cultivars. The original coffee population introduced into
the Americas may have also brought in some genes in a heterozygous condition.
Cramer (1957), in describing several mutants of C. arabica found in Java, has
also reached the conclusion that the natural genetic variability in this species was com­
paratively small, owing to the restricted provenance of the arabica material in Indo­
nesia.
Natural variability may be expected to be larger in the coffee areas of East Africa,
particularly in Kenya. Since Kenya is close to Ethiopia, the exchange of seeds could be
easier and more frequent. In spite of the fact that yield is highly influenced by envi-

Table 1 Contributions made to the yield by the 'best' and the 'worst' fractions of a population of
individually recorded Arabica trees (calculated from Gilbert's (1938) data).

percentage contribution to yield ratio

total yield
size of
fraction best fraction worst fraction (best : worst)

1934 1936 1934 1936 1934 1936

(%) (%) (%) (%) (%)

10 18.9 20.0 2.1 2.9 9.0 6.9
25 41.0 42.2 10.4 10.2 3.9 4.1
50 69.5 70.3 30.5 29.7 2.3 2.4

207
a. carvalho, f. p. ferwerda et al.

ronmental factors, the figures collected by Gilbert(1938-1939)andcompiledintable 1

provide a good illustration of the incidence of this characteristic.

Table 2 Contributions made to the yield by the 'best' and the 'worst' fractions of a population of
900 individually recorded C. arabica 'Bourbon Vermelho' trees from Campinas.

percentage contribution to
total yield, 1933 1950 yield ratl°
size of
fraction best fraction worst fraction (best : worst)
(%) (%) (%)

10 15.3 5.2 3.0

25 34.7 16.1 2.1
50 62.0 38.0 1.6

It is not possible in this case to discriminate between genotypic and environmental

variability. It can, nevertheless be assumed that at least part of the observed variability
may be ascribed to genotypic differences, which fact is of course of importance from
a breeder's point of view.
Data collected from a population of the Bourbon Vermelho cultivar in Campinas,
the yield of which was checked for 18 consecutive years, are given in table 2. The
differences noted may be due to the smaller degree of variability of the above cultivar.
In this particular case it was not possible to separate the genetic and environmental
components of the variation, and a clear-cut analysis of the main type of variation
observed would require a further progeny test.

Variability in isogenic lines

Isogenic lines provide a useful means for the evaluation of the effect of the environ­
ment on certain economically important characteristics controlled by a polygenic
system. Two isogenic lines and the hybrid between them were used in Campinas to
evaluate the effect of the environment on several important characteristics (Monaco
andCarvalho, 1963). It was possible to confirm that variability in yield is enhanced by
the homozygous condition of the lines. The yield was found to be more affected by
environmental factors than by plant height.

Variability encountered in coffee progenies

If progenies of selected plants are grown in lines and individual records are taken
for several years, a large range of variability can be observed in the total yield of the

208
 COFFEE

trees in each progeny. Similar behaviour is observed when F, populations are examin­
ed in the field. The components of the yield variation within the progenies were stu­
died in a series of 12 progenies derived from other trees of the highest, of the lowest
and of the medium yield classes all belonging to four selected I, progenies (Carvalho,
Monaco and Antunes, 1959). From each selected Ix plant a total of five new I2 pro­
genies was taken for yield evaluation. The data indicated that all I2 progenies have a
very similar production, which proves that the variability noted in the progenies of
'Bourbon Vermelho' studied was due to environmental causes. In the case, at least,
of the latter cultivar, these data indicate that it is advisable to take the average total
yield of the entire progeny into account in a selection programme. The effect of the
environment on the yield can also explain the absence of correlation frequently noted
between the yield of the mother trees and that of their progenies (Carvalho, 1952).

Variance analysis in C. arabica

Pronounced annual variability in the case of coffee is found in regions, where the
plant is grown without shade and where no pruning is carried out. The subsequent
yields of a coffee plant do not constitute independent variables, since a high yield is
usually followed by a reduced crop in the next year. It has been found that the first
four or five crops are not so variable as the subsequent ones. For the analysis of va­
riance the total yield of six or eight consecutive years or the accumulated yields of
two successive years can be used. This procedure reduces the range of variation ob­
served.
The available information obtained on the analysis of several yield trials has indicat­
ed that the interaction between variety and locality within the same year is low. This
fact corroborates the results obtained in the yield of selected progenies, which present
an appreciably wide range of adaptation. The ability of certain coffee progenies to
thrive well in regions ecologically different from the one in which they were selected
is of interest as it allows these selections to be disseminated in new areas without the
need for any prior local evaluation.

BREEDING METHOD

Several breeding methods are being used for the purpose of isolating those genetic
combinations which are the best adapted to each coffee area. In view of the pronounc­
ed autogamous nature of Arabica coffee, the most common system of breeding has, in
the majority of Arabica-growing countries, been some kind of line selection. In ad­
dition, artificial crossing between selected parent types is being applied in order to
bring about a combination of certain desirable characteristics.
The starting point of any breeding programme is the choice of mother trees from
the best cultivars available. Tests for the purpose of discovering the most adapted
cultivars are being made on a large scale in many coffee regions e.g. in Brazil, Puerto

209
A. CARVALHO, F. P. FERWERDA ET AL.

Rico, Colombia, El Salvador, Costa Rica, Tanzania, and other countries (Carvalho,
Scaranari, Antunes and Monaco, 1961 ; Dhaliwal, 1965; Krug, 1958). After the choice
of the mother trees, the nest step is the testing of their progenies. In connection with
these two stages of the work, the following important question immediately arises:
'How long must a mother tree be observed before a fair judgement can be made?' The
same question applies to the progenies.

Selection of mother trees

Selection of outstanding mother trees can either be carried out in commercial

coffee plantations of the most adapted cultivars or in experimental plots established
for this purpose. Harvesting of these selections can be carried out for a certain number
of years in order to select the highest yielding ones and then to study their progenies,
as has already been done in Campinas (Carvalho, 1952), in Tanzania (Fernie, 1965)
and elsewhere (Gilbert, 1939). In the selection of these mother trees, their vegetative
vigour and the coffee quality should be taken into account. It has been found that
the general vigour of the coffee tree can be correlated with its yield (Carvalho, Scara­
nari, Antunes and Monaco, 1961 ; Dhaliwal, 1965; Fernie, 1965).
Since a great deal of variation in the coffee yields is due to environmental conditions,
a very poor or even negative correlation has been found between the yield of the
mother plant and that of its progeny. For this reason it is proposed (Carvalho, 1952)
that the yield records of the mother trees should not cover too long a period before
the final selections are made. This selection should be based on only one year's yield
record and also on the general growth vigour, which is usually an indication that a
reasonable crop will be obtained in the succeeding year. The potentialities of the moth­
er trees are better evaluated by studying their progenies in a replicated trial. Vegetative
multiplication of the mother trees to test their yield can be applied but this is perhaps
unnecessary, since the analysis of their selfed progenies is much easier and more effi­
cient. Vegetative multiplication of the selected mother trees can be applied for their
maintenance. It has been suggested that under conditions prevailing in S. Paulo, the
selection of mother trees on private farms, in the absence of controlled production,
must be covered out in high-yielding years, since the best plants produce more in these
years. Selection during small crop years will, on the other hand, give poor progenies.
Apart from considerations of yield and vegetative vigour, a further test has to be
made in order to avoid the selection of mother trees with a high amount of empty
fruits (Leliveld, 1938), a characteristic which is genetically controlled. About 100
fruits are placed in a container with water and the number of fruits which float is
recorded. Plants with more than 10% 'floats' can be eliminated (Carvalho a.o.,
1952).
An analysis of the seed to avoid the selection of mother trees with a high percentage
of pea berries can also be made. This defect, although varying greatly from year to

210
 COFFEE

year has a fairly high heritability. Other characteristics of the seeds, such as size,
weight and density, in addition to cup quality, can also be determined.
After these observations have been made and all trees have been eliminated that do
not fully meet the requirements, it is best to proceed to the much more important
stage of testing the progenies.

Pedigree method of selection

The success of this selection method in the case of C. arabica depends on the degree
of genetic variability encountered in the original population. Positive results in improv­
ing C. arabica have been obtained with this method, particularly when the coffee po­
pulations have not been previously selected. Outstanding progenies of several cultivars
have been isolated by this method in various coffee regions of the world (Castillo,
1965; Gutierrez, 1965; Interamerican Inst, of Agr. Sciences, 1964, 1965; Salvadorian
Inst, of Coffee investigation, 1963).
For the study of the progenies, answers must be found to the following questions :
1. What is to be preferred?
(a) testing a large number of progenies, each represented by a relatively small
number of individuals, or
(b) testing a small number of progenies resulting from a rigorous selection, each
consisting of a large number of individuals?
2. How long should the observation be continued?
The number of plants per progeny has varied in different trials from 12-24. Owing
to the intensive effect of environmental conditions on the yield of the mother trees,
it is preferable to have a larger number of progenies with 12-20 plants each than to
have larger progenies from fewer selected mother trees.
It has been observed that, after four years of consecutive crops, the lowest yielding
progenies do not need to be harvested any more. It is, however, possible, although
this does not occur frequently, for late-producing progenies to improve their total
production even after 20 years of consecutive harvests. The selection of precocious
progenies is by no means a disadvantage, because coffee orchards should in any case
be replaced after periods of 15-20 years by higher yielding coffee progenies.
In order to be able to evaluate better the progenies of selected mother trees, it is
highly desirable to test them simultaneously at various localities. An average result
can then be obtained which will enable the selection of progenies well adapted to an
array of different ecological conditions. It has been found that some of the excellent
progenies so far isolated and widely cultivated bear this desirable characteristic of
broad ecological adaptability.

Multiplication plots

Plants of selected progenies may be artificially selfed in order to furnish basic seed

211
A. CARVALHO, F. P. FERWERDA ET AL.

stock for the establishment of seed orchards. The location of these plots needs to be
carefully studied, particularly if the selected strains are homozygous for certain reces­
sive alleles such as xanthocarpa. In the latter case a certain degree of isolation is ne­
cessary in order to prevent cross-pollination with other varieties or selections bearing
red fruits.
However, seed multiplication of several progenies of the same cultivar may be se­
cured in adjoining plots, as a certain amount of natural crossing among these selected
progenies may even be considered a favourable occurrence, as it will help to diversify
somewhat the genetic constitution of the seed to be furnished to the farmers. These
selections are morphologically very similar and the few resulting natural hybrids
will not affect the uniformity of the plantations.
It is also advisable not to plant only one selection on a given farm, but always to
use several lines in order to secure a good average yield under varying environmental
conditions.

Hybridization

A great deal of attention has recently been paid to artificial hybridization as a breed­
ing method in coffee. Crosses are made among selected plants of the best coffee
progenies in order to obtain, in advanced generations, recombinations carrying an
adequate concentration of genes for yield purposes. Complex crosses involving three or
even six plants of different cultivars have been carried out for this purpose in Campi­
nas. Crosses have also been made in order to combine specific characteristics such as
resistance to disease or to the attack of insects with that of a high yield.
Heterosis in terms of yield has not frequently been found in C. arabica after the
crossing of selected plants, although such cases have been reported in Costa Rica
(Leon, 1965) and Tanzania (Fernie, 1965). However, the commercial production of
hybrid seeds is not economically feasible at the present time. One might consider
increasing attractive heterotic F! progenies by means of vegetative propagation.
The identification of the components of yield would give valuable information for
the planning of the hybridization programme in the case of coffee. It has been found
that progenies with similar yields differ with regard to the characteristics responsible
for yield, namely the number of flowers, the percentage of fruit set and the seed density
(Carvalho and Monaco, 1965). It is, therefore, possible to base the selection of given
plants to be used in crosses on such differences in the yield components. For instance,
crosses can be made between plants of similar yields but differing in the number of
flowers, the percentage of fruit set and the seed set or seed weight. The recombinations
may have a better yield than both parents have, because they carry different sets of
genes favourably affecting yield.
A promising new combination which is still under observation was obtained in
Campinas in the F4 of a cross between Caturra and Mundo Novo cultivars. It is
expected that this new cultivar, owing to its small height, will be useful in regions where

212
 COFFEE

the coffee tree normally reaches a considerable size and also in mountainous areas,
where the harvesting of traditional coffee cultivars is difficult and expensive. Other
new combinations are being developed in Tanzania (Fernie, 1965) and Colombia
(Castillo, 1965).
Interspecific hybridization involving the tetraploid C. arabica and other diploid
species also offers possibilities for securing new forms of Coff'ea. The triploid hybrids
are sterile and their chromosome number must be duplicated for it to become fertile
(Monaco and Medina, 1965).

Backcross method of breeding

Backcrossingis being rather extensively used for transferring certain characteristics

of C. arabica, such as the size of the beans or the size of the plant, which are control­
led by single genes.
The backcross method is also being used to transfer certain characteristics of the
tetraploid C. canephora and C. dewevrei to C. arabica by successive crosses of the hy­
brid with selected plants of C. arabica. It has been found that after four backcrosses
the plants still show a variable degree of sterility.

Breeding for disease and pest resistance

A great deal of excellent research work is still being carried out on the resistance to
the leaf rust caused by various physiological races of Hemileia vastatrix. Arabica lines
with resistance to some races of this fungus were first developed in India, where fun­
damental research was carried out on the differentiation of physiological races of
H. vastatrix. In spite of the great difficulties involved in the selection of types of coffee
resistant to H. vastatrix using all available sources of resistance, highly valuable work
has already been done in India, Congo (Kinshasa), Indonesia, Kenya, Tanzania and
Uganda (d'Oliveira, 1958). Special mention should be made of the research carried out
at Balehonnur Coffee Station in India on the breeding and selection of C. arabica
and C. canephora with resistance to leaf rust. Excellent research work was more re­
cently carried out in Portugal, in the Centro de Investigaçào das Ferrugens do Cafeei-
ro, on the same subject. A series of alleles have been detected in C. arabica (d'Oliveira
and Rodrigues, 1960) which confer resistance to most of the known physiological
races. Hybridization between resistant coffee plants is being pursued in order to
concentrate the different genes for resistance and, possibly, to develop coffee strains
resistant to all known races of the fungus (Coffee Rusts Research Center, Oeiras,
Portugal, 1965).
Resistance to Hemileia coffeicola is also being sought among the many coffee in­
troductions analysed in Portugal. Up to now no resistant arabica plants have been
discovered, whereas other coffee species, such as C. racemosa, are highly resistant
(d'Oliveira, 1958).

213
A. CARVALHO, F. P. FERWERDA ET AL.

Attempts are being made in Angola and Tanzania to select types which are resistant
to coffee berry disease (Colletotrichum coffeanum) (Fernie, 1965) and in Puerto Rico to
select plants with resistance to the root disease caused by Fusarium oxysporum f.
coffeae (Dhaliwal, 1965). One single Arabica plant was found in Colombia which was
resistant to the disease 'Ilaga macana' caused by Ceratocistisfimbriata. This resistance
is transmitted to the progeny, although there is no information about its mode of
inheritance. Resistance to this latter disease (Echandi and Fernandez, 1962) may be
correlated with a high content of chlorogenic acid in the plant.
Screening for resistance to the coffee leaf miner (Leucoptera coffeella) has been at­
tempted (Dhaliwal, 1965). All Arabica types are severely attacked in some regions,
particularly during the dry months of the year. The species C. stenophylla seems to be
immune and C. eugenioides and C. kapakata are highly resistant. At Campinas the lat­
ter two species and also C. racemosa, C. salvatrix, C. dewevrei and C. liberica have
been found to be highly resistant.

Breeding for other characteristics

More attention is now being given to obtaining a better insight both into the variabi­
lity of the components of both green and roasted coffee beans and into cup quality.
Coffee oil. The oil content has been shown to vary in different mutants and culti-
vars of C. arabica (Pinto and Carvalho, 1961). Mucronata and Sao Bernardo mutants
were found to have the highest oil content (17.8 and 17.1%, respectively), while the
xanthocarpa has the lowest one (10.2%) (Tango and Carvalho, 1963). The oil content
is also significantly different among lines of selected cultivars of C. arabica so that
the selection of groups of progenies with low and high oil contents is possible.
Caffeine content. Variations were found in the amounts of caffeine in green coffee
beans, particularly among the mutants of C. arabica. It has been shown that the
laurina mutant (Irlr) accounts for about one half (0.62%) of the caffeine content of
arabica (1.29%) (Tango and Teixeira, 1961). The Maragogipe (Mg) and the mokka
(mo) alleles, on the other hand, seem to increase the caffeine content (Carvalho
a.o. 1965). The information gathered about the variability of caffeine content un­
derlies the value of this material in any breeding programme aiming at an increase or
a reduction of the caffeine content.
Soluble solids of the beverage. The expanding instant coffee industry highlights the
need for a better knowledge of the soluble solids content of the coffee bean and its
quality. The soluble solids content has been found to be independent of seed size.
In the species C. arabica, the mokka mutant, with small seeds, and the Maragogipe,
with very large ones, both have a high soluble solids content (Toledo a.o., 1963).
Data indicating the variability of the soluble solids content found in selected progenies
of the main cultivars of C. arabica (Toledo a.o., 1963) are also available.
Flavour evaluation. Coffee quality must be evaluated by a panel of selected indivi­
duals and reference samples with known hard and soft cup qualities are used for

214
 COFFEE

checking purposes. The testers award points for each coffee sample, the quality
being represented by consecutive numbers. A statistical analysis may be performed,
and comparisons can be made with the reference samples (Garruti and Conagin, 1961).
Using this procedure, most of the selected progenies developed by the Institute Agro-
nômico, Campinas, have been evaluated in order to determine the variability in
their cup quality components.
It has been verified in the case of the C. arabica cultivars selected in Tanzania, that
cup quality seems to be primarily dependent on the shape, texture and colour of the
beans (Fernie, 1965) and these are considered to be inherent characteristics.
Niacin. Very few data have been gathered on the niacin content of selected proge­
nies (Carvalho, 1962), although some data have already been reported which indicate
a reasonable variation among cultivars of C. arabica in respect of this complex char­
acteristic.

Selection for resistance to low temperature

Several investigations have been carried out in order to test the resistance of select­
ed cultivars of Coffea arabica to low temperature. It has been verified in Campinas
that a temperature of 4°C below zero kills the seedlings of most cultivars, although
a few seedlings were not harmed by this temperature, even when they were submitted
to it for a few hours.
Indications were found to show that low temperature resistance is a heritable charac­
teristic.
Resistance to low temperature was also investigated in the coffee collection main­
tained in Florida, USA (Söderhölm and Gaskins, 1961). Whilst a few C. arabica
cultivars were shown to be more resistant, the great majority were affected by the low
temperature.

Attainments and problems for future research.

Coffea arabica, owing to its almost autogamous nature, offers many more possibili­
ties in connection with the genetic analysis of its characteristics and with breeding
work aimed at the development of subtle traits such as flavour and other organoleptic
properties, suitable for making evaporated coffee, bean size and shape, resistance to
diseases and pests than does an allogamous coffee species like C. canephora (Ferwerda,
1958).
Genetic analysis of C. arabica has been undertaken in Campinas arabica being
used as a standard. It is desirable that this same standard should be used wherever
genetic analyses of this species are being carried out. It has been verified that the in­
heritance of most of the main characteristics of its cultivars is simple, the differences
being controlled by one or two independent genetic factors or their interactions. How­

215
A. CARVALHO, F. P. FERWERDA ET AL.

ever, the number of mutants so far analysed is still too small for the establishment of
linkage groups and for an adequate evaluation of the genetic constitution of C. arabica.
Ethiopia, the centre of variability of the species, needs to be searched for new genes
(Monaco, 1964; Sylvain, 1958). The use of mutagenic agents may be of interest for
producing new mutants not yet found in natural populations. It has been observed
that C. arabica has been shown to be resistant to the mutagenic effect of ionizing radia­
tions, probably owing to its polyploid nature.
The many yield trials on local or introduced cultivars under observation in the cof­
fee research centres are giving valuable information as to which are best adapted to
each particular coffee region. Progeny evaluations from mother trees derived from
previously selected or unselected populations of the main cultivars are being discover­
ed to constitute an effective breeding method. High-yielding coffee progenies with
disease resistance and with good cup quality are also being obtained by the pedigree
method of selection. Artificial hybridization between selected plants and the back-
cross method are being more frequently used to obtain genetic recombinations.
Interspecific hybridization, though a promising breeding method, is more remotely
applicable in breeding because of the sterility involved and also because of the transfer
to the hybrids of blocks of undesirable genes.
A knowledge of the chemical composition of green or roasted coffee beans and of
the beverage is considered to be of value in designing the new projects in which quanti­
tative changes of certain constituents are being'considered. Evaluation of the cup quali­
ty is considered to be of importance, and experiments to determine quality must be
carefully designed so that the results may be submitted to statistical analyses.
The economic results so far obtained on several breeding projects carried out during
the last few years have been very promising and it is hoped that these projects will
be further expanded and new ones added. Newly developed coffee selections producing
100-200% more than the arabica cultivar have been obtained and the importance
of this fact for the future of the world's coffee industry cannot be overemphasized.

Breeding of canephora coffee

(F. P. Ferwerda)

FLORAL BIOLOGY

Inflorescence and flower structure

The flowers of canephora coffee are borne on one year old lateral branches. The
other cultivated diploid coffee species present a similar picture except for excelsa
coffee (C. dewevrei var. excelsa), which regularly flowers on older wood or even on the
stem. The flowers are arranged in dense clusters in the axils of the opposite leaves.
The number of flowers in one cluster may amount to 80 on profusely flowering bran­

216
 COFFEE

ches of canephora coffee. Detailed descriptions of the structure and the development
of the inflorescence have been made by van der Meulen (1939) and Moens (1962).
The structure of the canephora flower is largely similar to that of C. arabica, which
has been described and pictured on page 198.

Flowering periodicity

The flowering of canephora coffee and also that of most of the other cultivated spe­
cies is characterized by a remarkable periodicity, which manifests itself especially in
areas with well defined wet and dry seasons. Approximately a week after the first
significant showers in the second part of the dry season the shrubs are in full blossom
which lasts two or three days. As a rule there is one main flowering period preceded or
followed by some less extensive flower bursts. In regions with an evenly distributed
rainfall there is no main flowering period, but several small flower showers. One va­
riety shows a far more marked flowering periodicity than another. In East Java typical
Robusta cultivars normally have a distinct flowering rhythm, whereas Uganda varieties
which more readily respond even to small amounts of precipitation are characterized
by regularly recurrent periods of flowering throughout the year. Periodicity of flower­
ing in relation to flower bud development and differentiation have been studied by
several investigators (de Haan, 1923; van der Meulen, 1939; Portères, 1946, 1947;
Moens, 1962 and 1965; Ramaiah, P. K. a.o., 1965). The work performed in Kenya has
been reviewed by Wormer (1965).
In a climate with well defined rainy and dry seasons the flower buds are initiated
during the second half of the rainy monsoon. At the beginning of the subsequent dry
season the still small, but fully differentiated flower buds enter a period of dormancy
from which they are awakened as soon as a rainshower surpassing a certain threshold
value has fallen. For further details on this interesting subject, which has only an
indirect bearing on breeding, reference can be made to the authors cited above.

Biology of flowering

After a rainfall sufficient to induce blossoming the flower buds swell visibly and
double their length and volume in two or three days' time. On the day before they
open the buds look like small white candles as is pictured for C. arabica in fig. 5 p. 201).
The flowers open just before sunrise and the anthers start dehiscing as soon as the
sun breaks through. The light powdery pollen is mainly conveyed by wind and thermic
air currents. The part played by insects in transferring pollen is negligible.
Experiments in Java (Ferwerda, 1936a; Snoep, 1940) with glass slides covered with
a sticky substance and placed at various distances from a flowering coffee garden,
demonstrated that appreciable quantities of wind-borne pollen can be detected at
a distance of 100 m, downwind. This pollen cloud may rise fairly high; a fair density
of pollen grains was observed on slides eight meters above the ground. Considerable

217
A. CARVALHO, F. P. FERWERDA ET AL.

quantities of pollen descend from higher branches to lower ones of the same tree.
This pollen fall is of no interest to fertilization in self-incompatible coffee species but
important in self-fertile species such as C. arabica.
The longevity of coffee pollen is limited. The germinative power of Robusta pollen
is lost after a few days when it is stored without special precautions. Its life span can
be prolonged up to a month or even longer by storing it in a dry atmosphere over
quicklime, calciumchloride or some other strong desiccant (Ferwerda, 1937a; Mendes,
C. H. T., 1950; Devreux a.o., 1959).
Stigmas remain receptive up to six days after anthesis (Ferwerda, 1948).

Compatibility relations

A common feature of the coffee varieties belonging to the species C. canephora

Pierre (Robusta, Uganda, Quillou) is that they are all pronouncedly self-incompatible.
Early experiments in Java (Lambers, 1929, 1932; Ferwerda, 1932) showed that prac­
tically no fruits are set on self-pollinated branches, whereas cross-pollination results
in fruit setting percentages up to 50 % or even more. This fact has later been confirmed
by research workers in Central Africa (Dublin, 1957; Devreux 1959; Butt 1966) and
in Brasil(Mendes, C.H.T., 1949). C.dewevrei (excelsa) is also distinctly self-incompatible
(Ferwerda, 1948; Dublin, 1960). Devreux a.o. (1959) found that in compatible polli­
nations the pollen tubes take 30-36 hours to reach the ovules, a figure well in accor­
dance with that found by other research workers. Incompatible pollen tubes never
travel beyond the papillary zone of the stigma and mostly end in a bladdery swelling
which often ruptures.

Little is known about the genetic background of incompatibility in Cojfea. The

observational data have been tentatively interpreted (Devreux a.o., 1959) as gameto-
phytic incompatibility controlled by an oppositional S-allele system. This phenomenon
deserves more detailed investigation. The nature of the incompatibility mechanism
implies important consequences for the breeding procedure and the design of seed
orchards. The latter subject is further discussed on page 228.

Development of seed andfruits

The developmental anatomy of seed and fruit have been dealt with on page 195.
After fertilization the ovary develops into a ripe cherry in approximately 300 days
- ranging from 290 to 330 days - under the conditions prevailing in East Java. Slightly
different values may be found elsewhere. The duration of the development is de­
termined by the female parent. No influence on the part of the pollen parent has been
observed (Ferwerda, 1937b).
Of the numerous flowers produced by a coffee tree only a comparative small pro­
portion, on average not more than 20 or 30% develops into mature cherries, the

218
 COFFEE

others being shed as ovaries or immature fruits. Fruit fall in Robusta coffee occurs in
two waves.
A large number of cherelles are shed in the first four months after flowering particu­
larly during the first two months. Then, after a short interlude, a second much reduced
fruitfall sets in and lasts till maturity (Ferwerda, 1948; Dublin, 1957, 1960; Devreux
a.o., 1959).
According to Leliveld (1938) the cherelles shed during the first period mostly con­
tain unfertilized embryo sacs. On the other hand, all young fruit shed in the second
period show well developed endosperms and embryos. Obviously the unfertilized or
inadequately fertilized ovaries are discarded by the plant as soon as possible after
flowering. During the second phase of fruit drop, success or failure in the struggle for
existence determines which of the fertilized ovaries will develop into a fully ripe fruit.
In many respects fruit fall in Canephora coffee shows similarities to that of fruit crops
in temperate regions.

Defective seedformation

On page 196 the main forms of seed defects have been described and pictured
(fig. 1 and 2). The first type: unilateral seed abortion leads to peaberry formation, the
second: arrested development of the endosperms gives rise to the formation of empty
beans. The two forms of seed defects are of great practical importance to a crop like
coffee where the bean constitutes the product that is wanted.
Peaberry formation in the Canephora group is connected with pollination. If pollen
transfer from tree to tree is impeded by fog or rain or by absence of wind during
flowering, self-pollination tends to prevail which ultimately leads to bad fruit setting
combined with a high percentage of peaberries.
In a series of test crosses with C. canephora and dewevrei (excelsa) fruit setting and
peaberry percentage were found to be inversely correlated. A low fruit setting percen­
tage usually coincides with a high percentage of peaberries (Ferwerda, 1948) and vice
versa.
Empty bean formation is likely to be attributed to disharmony between the parental
chromosome sets resulting in irregularities in gametogenesis and growth disturbances
of the endosperm (Leliveld, 1940; Ferwerda, 1948).
Both seed anomalies cause a loss of yield. In the case of peaberries a simple calcula­
tion shows that, within certain limits, an increase of 1 % in the proportion of pea­
berries means a decrease in yield of about 0.75%. If one considers that a peaberry
percentage of 40 is no exception one realizes the gravity of the loss this anomaly can
cause.
The loss of yield resulting from empty beans is more difficult to estimate because
here both the proportion of abnormal beans and the extent of endosperm reduction
should be taken into account. Experimental evidence indicates that losses up to 30 or
40 % may be incurred.

219
 A. CARVALHO, F. P. FERWERDA ET AL.

Fig. 8 Part of a branch of C. canephora with emasculated flowers (A). To effect emasculation, the
unopened corolla with adhering stamens is lifted (B) leaving the flower emasculated with the bifurca­
tions of the stigma still sticking together (C). B and C about 4 x enlarged.

Technique of controlled pollination

The methods used for C. canephora differ only slightly from those described for
C. arabica on page 201.
Flowers are emasculated the day before anthesis when they are in the so called
'candle stage'. The top of the bud is gently bent to the right and left and slightly
twisted so that the corrollar tube snaps off just above its insertion on the ovary. The
corolla with the inserted stamens can then easily be lifted so that only the pistil of

\ 220
 COFFEE

Fig. 9 A flowering Robusta-tree with some branches bagged for crossing.

each flower is left (fig. 8). The flowering branches thus emasculated are next wrapped
in tightly woven cotton sleeves supported by rattan hoops or a light iron frame, or
put in strong paper bags (fig. 9).
In spite of this rather drastic treatment the stigmas develop normally and the next
day they can receive the desired pollen which is applied by means of a fine brush. Then
the bag is closed again. After a week it can be removed since by then the stigmas will
have withered and are no longer receptive.
With pronouncedly self-incompatible species like C. canephora one might consider
omitting emasculation altogether, when simple test crosses are to be made.

BREEDING

Coffea canephora and the other allogamous coffee species present great difficulties
in breeding because of their pronounced, almost absolute self-incompatibility. On the

221
A. CARVALHO, F. P. FERWERDA ET AL.

other hand it is this pronounced tendency towards cross-fertilization which allows

the use of breeding methods especially designed for this category of plants and aiming
at the development of synthetic varieties or progenies similar to single crosses.
On account of the longevity of the trees and their unlimited possibilities of being
propagated vegetatively these methods can be applied with even greater success than
is possible in annual cross-fertilizers.
Coffee breeding is a time consuming business as a consequence of the perennial
nature of the crop. It takes one breeding cycle at least 20-22 years to revolve. Long-
term breeding programmes have been carried out or are still in progress in Indonesia,
Congo Kinshasa, the Central African Republic, Ivory Coast and on a smaller scale
in Malagasy (Madagascar), Uganda and India. For summarizing reviews of those
programmes reference is made to Ferwerda, 1948; 1958; Fressanges, 1954; Vallaeys,
1956; Thirion, 1962; Capot, 1962; Braudeau c.a., 1962; Dublin, 1967 and Thomas,
1947.
Natural variability is the basis of the breeding of any crop plant. This subject will
now be discussed for C. canephora.

Variability of the initial material

The original unselected or only superficially selected canephora material available

in the various areas of cultivation shows a wide variability in practically all properties
that can be observed visually viz. general habit, size and shape of leaf and fruit. It
also applies to such complex and polygenically based properties as yielding ability.
The scarce data on the yields of individual trees of only slightly selected canephora
populations demonstrate that there exist enormous contrasts in this respect as can be
seen from table 3.

Table 3 Contribution to yield made by the 'best' and the 'poorest' fractions of a population of
individually recorded canephora coffee trees. A adapted from Ferwerda (1932), B from Snoep (1937).

size percentage contribution to total yield ratio

of best fraction poorest fraction (best : poorest)
fraction A B A B A B
0/
/o

10 31 20 1.5 3
Ö
1

25 52 40 10 10 5.- 4.-
50 88 70 22 30 4.- 2.3

This variability may be attributed to two factors: genotypical differences and en-

222
 COFFEE

vironmental influences. It is difficult to estimate precisely to what extent each of these

causes has contributed to the total variability although the coefficient of variability
within clones i.e. within genotypically uniform material might offer a guide. The values
found for clones prove to approximate half those calculated from seedling populations
grown under similar conditions. This contrast indicates that a considerable proportion
of the variability observed within seedling populations must be ascribed to genotypic
differences and consequently offers a starting point for breeding. If one realizes further
how much infinite variability the other members of the canephora group contain e.g.
Coffea laurentii, C. bukobensis, C. ugandae, all forms capable of mutual crossing, there
is no reason for anxiety about the range of the basic breeding material. Moreover it is
possible to widen this basis by involving new material from Central Africa where the
coffee shrub originated. These areas have only partly been explored in this respect.

Choice of mother trees

The choice of the mother trees is the first step in starting a breeding programme.
After a most critical visual inspection a number of eligible mother trees are selected
from the initial material and put under observation. These candidates should combine
as many good properties as possible. Apart from yielding capacity such factors as
regularity of bearing, general habit, resistance to diseases and pests, shape and size
of the beans and conversion ratio (outturn) are taken into account. At a more advanc­
ed stage the organoleptical qualities should also be evaluated. Relative productivity
should be stressed more than absolute productivity, that is to say the yield of a mother
tree is best compared with that of a group of neighbouring trees of the same age grow­
ing under similar conditions and expressed as a percentage of the latter (yield index).
A yield index of 300, meaning that the tree yielded thrice the average of the plot where
it grew, is considered a fair criterion for admitting a mother tree to the breeding
programme. If one is to form a good idea of the productivity of a tree it is necessary
to harvest it separately for at least four successive years. A period of that length is
required owing to the marked tendency to biennial or triennial bearing of Robusta
coffee. That the minimum number of years required is four can be inferred from the
fact that the coefficient of correlation between the average yield over n years and
the multi-annual average no longer increases when n>4; in that case r approximates
0.87 (Ferwerda, unpublished data). Similar correlation values are mentioned by
Dublin (1967) who found r-values of over 0.90 between the accumulated yield over
4 years and that over a period of 5 or 6 years. This justifies the conclusion that longer
records of yield will not change the judgement on yielding capacity of trees. It is safe
to make a choice among the prospective mother trees after a comparatively short
period of observation, say 2 or 3 years, and to multiply the provisionally selected in­
dividuals vegetatively. This has a twofold aim.
First it is a safety measure: if the mother tree falls victim to disease, storm, light­
ning or some other calamity, the clone is still there.

223
A. CARVALHO, F. P. FERWERDA ET AL.

Secondly, there is an opportunity of studying the rejuvenated replica of mother

trees under similar external conditions which is impossible in the case of mother trees
scattered over a large number of plantations.
Such a living herbarium provides much valuable additional information which the
mother trees themselves fail to reveal. In addition some preliminary impressions are
obtained with regard to the capacities of their clonal offspring. This may be of impor­
tance to a selection programme aiming at the obtention of clones.
On the strength of some years of critical observation several mother trees are dis­
carded and only a limited number selected for a progeny test.

Progeny testing

In breeding coffee two objectives may be aimed at:

a. the production of improved seedling strains,
b. the development of superior clones.
In the case of C. canephora both objectives are pursued. As to their relative impor­
tance b should be considered subordinate to a. Vegetative reproduction only allows
fixation of existing genotypes as a clone. New gene combinations can only be achieved
through true breeding.
In carrying out a breeding programme both objectives are simultaneously pursued
because they are inseparable. It is only for reasons of survey that breeding proper is
first discussed whereas clonal selection will be treated in a separate paragraph (page
229).
In the early experiments started in Java between 1915 and 1930 mainly open pollinat­
ed seed was used for progeny tests. In some experiments seed was taken from branches
that during the flowering season had been enclosed in bags of mosquito cloth. This
wide mesh fabric as we now know is not an efficient means of excluding cross-pollina­
tion. Most of the progenies so obtained did not fulfill the expectations. This method
aiming at obtaining pure lines soon fell into the background and emphasis was shifted
to the progenies obtained from open-pollinated seed.
Progeny testing should be done in comparative trials, whenever possible replicated
and with one or more well established strains or clones as checks. We still are very
scantily informed about the proper size of the plots and the desirable number of re­
plications.
In these comparative trials the progenies are submitted to a critical judgement
not only of their potential yielding capacity but also of other valuable properties.
Four or five years are required in a capriciously bearing crop like robusta coffee in
order to get a sufficiently founded idea about its yielding capacity.
The chances of hitting upon really outstanding families appear to be small. Among
the several hundreds of open-pollinated progenies tested at the Bangelan experiment
station in Eastern Java only 1-2% were really outstanding. At the Yangambi (Congo)
station (Vallaeys, 1956) the score was rather better: 5 out of 70 progenies originating

224
 COFFEE

from critically selected mother trees exceeded their standard by 25-30 %.

Once it was established that mother trees occur whose open-pollinated (illegitimate)
progeny goes far beyond the average the next move was to explore the possibility of
raising such valuable seedling families in quantity and planting them on a commercial
scale.
The following solution seemed acceptable. Assuming that the illegitimate progeny
of mother tree 187 proves to be outstanding, then seed of approximately the same in­
trinsic value may be obtained from clone 187. With this purpose in mind clonal plots
for the production of seed were planted. In the beginning large monoclonal orchards
were laid out for this purpose in Java but owing to the pronouncedly self-incompatible
nature of robusta coffee - not yet fully realized at that time - these large monoclonals
were a failure.
After the exclusively allogamous nature of the canephora-group had been recognized
in the early thirties and its consequences realized the monoclonal seed orchards were
replaced by others consisting of narrow strips - often single tree rows - of the various
valuable clones. The seed formed there was the result of interclonal pollination and,
consequently, not entirely legitimate. Hence the name semi-legitimate or prope-legiti-
mate seed under which it was brought into the trade in pre-war Indonesia. Despite its
plural parentage this seed proved to have a considerable less variable composition
than the illegitimate seed of the mother trees that was used for the first tests. As a rule
repeated tests made by means of seed from clonal seed orchards have largely corroborat­
ed the first findings. These prope-legitimate families proved to be really valuable and
practical planters used them profitably for a number of years. However they do not
yet represent the best attainable final product : at most they may be considered as an
intermediary product.
In further widening the range of breeding work breeders have, in accordance with
the self-incompatible nature of C. canephora, directed their activities towards the pro­
duction of crosses between outstanding mother trees of the first breeding cycle in the
hope that progenies would be obtained in which the good qualities of both parents
were combined. At first the test-crosses were entirely empirical and chiefly guided by
'breeders' intuition' but later on they were made according to a definite system. By
doing so breeders hit upon particularly interesting Fx combinations which were su­
perior to the prope-legitimate progenies in various respects. Their more diversified
genetical structure contained the possibility of further widening the scope of work.
It is one of the strong points of the perennial crops allowing vegetative reproduction
that such outstanding Fx combinations are comparatively easily reproduced. One has
only to lay out a seed-orchard in which the clones involved in the prominent crosses
are planted in alternate rows. A strict condition to be fulfilled is that the two clones
apart from 'nicking' well should also coincide as to flowering time and rhythm in
order to ensure successful interpollination. Furthermore good care should be taken
that no alien pollen can penetrate. Relevant measures are discussed on page. 228.
The breeding system of which the development has been described above is based

225
A. CARVALHO, F. P. FERWERDA ET AL.

© TSL © © © TSi © © ©

O O O O

m X y

O O O O O O O O O O O O
O •© O O © O O O
© O © O O © O
EZ O O © O
O O O O © O ö
© O O O
O ©:O O O O O
O O © O O O © O
O oio O O O O O o o:o

A+ A +A
A+ A+ A
O " O O O O O A+ A+ A
A+ A +A
A+ A + A

Fig. 10 Schematic representation of the canephora coffee breeding procedure in Java and elsewhere.
I. Initial material O selected, / rejected. II. Testing of clones. III. Crossing. IV Progeny testing and
selection of new mother trees (©). VA. Testing of secundary clones. VB. Biclonal seed garden.

mainly on experience and experimental evidence compiled by the breeding stations in

Java and the Yangambi station in Congo Kinshasa. In a somewhat revised form it
has also been adopted by the O.R.S.T.O.M. institutions operating in francophone
Africa (Braudeau a.o., 1962). It has been diagrammatically represented in fig. 10.
In spite of its unmistakable advantages this breeding method has one serious
drawback: as a rule the number of test-crosses required for stage III is very large. As
long as a moderate number of parent trees (10 or less) are involved the number of
possible combinations will not exceed 45. With fairly extensive parental material as
is quite common in a well conceived breeding programme the number of possible
combinations may easily amount to several hundreds. Owing to the short flowering
period of canephora coffee it is not always possible to execute a complete series of
crosses in one season. Consequently the crossing programme has to be spread over
several years and the results become known proportionally later. This, indeed, is a
serious bottleneck and ways will have to be found to circumvent it.

226
 COFFEE

$ ® ® & ® " ® ® ® ® ^ A

O () O B

•••• •\ • ••• ;•
• • • • • \ «V• • • :•••••
• • • • • \ • •V» •
• • • •• I • •
:•••••
'•••••
TL
• • • • • ; • • • •V >••••

HE

• -f o m v-^-A Fig. 11 Schematic representation

-y"A • + O • V
• V-^-A « + 0 of the improved breeding method.
+ 0 • V-<J>A • 12
A • + O • V^- IA and B as in I and 11 of fig. 10.
V-^-A • + O •
O • V-^-A • + II. Testing of top cross progenies.
B I III. Crossing of outstanding com­
biners, IVA. Progeny testing.
IVB. Multiclonal seed garden that
2 is gradually thinned. V. Large scale
planting of progeny obtained in
IVB. (From Ferwerda, 1954).

At the beginning of stage III one may, guided by breeders' intuition, restrict oneself
to a limited number of combinations that seem attractive. This procedure can lead,
and in fact has led, to remarkable results. But at a further stage there is no alternative
to more systematic work. Both its scope and the time required for it can be considera­
bly confined by adopting the method used in producing hybrid maize and alfalfa, and
to proceed in steps as indicated in fig. 11.

First, the mother trees are evaluated for general combining ability by means of a
top-cross or a polycross test, or simply by testing their illegitimate progenies. Only
the restricted number of individuals endowed with a good general combining ability
can be considered for diallel crosses in order to ascertain specific combining ability
(stage IVA of the diagram fig. 11). With this method the number of combinations
remains within reasonable bounds and no longer offers difficulties.
Once it has been established which are the favourable combinations the next
stage of the breeding programme follows: the lay-out of seed orchards for large scale
production of the desirable cross combinations (fig. 10, VB, fig .11, IVB).
For this purpose it is possible to choose between biclonal and multiclonal seed
orchards. The former yields seed of the constitution A x B while from the latter a
seed mixture is obtained comprising all possible cross combinations between the
clones present. The seed of a biclonal seed garden may be referred to as an Fx family
whereas that of multiclonal orchards resembles a synthetic variety.
The allogamous and anemophilous nature of the canephora-group necessitates

227
A. CARVALHO, F. P. FERWERDA ET AL.

several precautions in the lay-out of seed orchards. Proper care should be taken to
exclude unwanted pollen as effectively as possible, either by laying the seed plots in
isolation or by surrounding them by a shelter belt of identical material of at least six
tree rows (± 20 m) deep. In accordance with the observations on pollen dispersal by
wind mentioned on page 217, a distance of at least 100 m is required to give a seed
plot without guard rows an adequate protection against contamination by intruder
pollen.
It is important to start the lay-out of seed gardens as early as possible. If the choice
of clones considered for inclusion is postponed till progeny testing (stage IVA of the
scheme, fig. 11) has yielded its definite results it will be another four years before such
an orchard comes into bearing. This waiting period may be bridged by artificially
performing the desired cross-combinations on a large scale. This can be done by dust­
ing a large number of flowering branches in the middle of a monoclonal block of clone
A with pollen of clone B by means of a powder duster. In the middle of an isodiametric-
al block of sufficient size the chance of contamination by airborne pollen is negligible.
Although the correct application of this 'versatile seed orchard method' (Lambers,
1934) yields good results it requires too much labour to find general acceptance and
therefore never went beyond the 'stop gap' stage. It is far more attractive to use the
following method which gradually aims at the definitive composition of the seed or­
chard. If clones A to K should be included in the systematic test-cross programme
(stage III of the scheme, fig. 11) a large seed orchard comprising all these clones is
laid out right from the start of this programme (IVB, fig. 11). They are interplanted
according to a definite system by which random interpollination is secured and yet
the possibility remains of easily tracing back each clone.
As soon as the test crosses have been evaluated and one knows which families
excel, all clones that do not come up to the standards are removed and only those with
good combining ability are kept.
The gaps resulting from the cutting of the unwanted trees can be easily filled by
regrafting the stumps with scions of the selected clones. So in a few years a close stand
is regained.
It is an open question whether the reduction should be carrried to the point where
only a minimum of two clones remains or whether a slightly larger number, say five or
six, should be retained. In so doing it is possible to compensate for small differences in
flowering time and then assure a good seed set. Another point is that, if incompatibility
in C. canephora is gametophytic and governed by an oppositional S-allele system, as
postulated by Devreux a.o. (1959), then groups of inter-incompatible individuals
are less likely to occur within the progeny of a multiclonal seed orchard than within
one of biclonal parentage.
It is clear that for the production of the best possible combinations one depends on
biclonal seed gardens. Additional research work will help to make sure beforehand
that no disturbing signs of incompatibility within the Fj family need be feared. If
they should occur they will show up at the test of the Fj families and manifest them­

228
 COFFEE

selves in poor fruit setting and fruit bearing which in itself will justify removal of
such a family.

Testing for local adaptability

Outstanding performance of a new variety on the trial fields of an experiment sta­

tion by no means implies that its behaviour will be similar under other climatic or
edaphic conditions. The need to assess local adaptability is generally recognized and
consequently local trials scattered over the main coffee producing districts have been
organized by the experiment stations.
Especially in Indonesia during the years 1934-1942 and in Congo Kinshasa this
subject received major emphasis. In Indonesia much valuable material was lost during
the war and the turbulent post-war years. The evidence obtained from the experiments
that could be carried out uninterruptedly or resumed after the war clearly indicates
that some seedling strains and, even more pronouncedly, clones, are very specific
as to their requirements for climate and soil. In contrast there are others that display
a much wider range of adaptability, and perform very well not only in various dis­
tricts of Java but also in Congo Kinshasa (Gude, 1956, 1957; 's Jacob, 1938; Ong,
1957; Yallaeys, 1956). On the strength of information obtained from local experiments
and the experience of planters it is possible to draw up an advisory report serving to
indicate the right place for planting the new material.
Some of the results achieved by the above breeding system are discussed on page 234.

Selection of clones

In the breeding procedure outlined above clonal reproduction fulfilled the indis­
pensable yet subordinate role of handmaid to breeding proper inasmuch as it made
possible the recreation of certain valuable cross combinations.
As a side-line to this breeding system a selection procedure was developed having
as its main objective the development of clones suitable for composing commercial
plantations - just as in fruit cultivation.
The vegetative propagation of most coffee species does not present any particular
difficulties. The oldest method applied is that of grafting. More recently efficient meth­
ods of rooting soft-wood cuttings were developed by which the use of rootstocks,
always considered rather cumbersome, is rendered superfluous but which, on the other
hand, also exclude the specific advantages that certain rootstocks may offer.
The methodology of vegetative propagation has been briefly described on page 191.
On superficial consideration the practice of selecting clones seems to be easier and
more efficient than generative selection since every attractive genotype can be easily
fixed as a clone which is an exact replica of the mother tree from which it is derived.
In such a strongly heterozygous crop as Coffea canephora it is only after a long series
of years that the fixation of a favourable combination of genes might be generatively

229
A. CARVALHO, F . P. FERWERDA ET AL.

approached. Properties such as bean size and cup quality which presumably are govern­
ed by a complex of genes are certainly easier to fix vegetatively than by sexual repro­
duction. The advantage of vegetative propagation is also very conspicuous in the case
of interspecific hybrids where multiplication by seed results in a very heterogeneous
and largely worthless progeny.
Nevertheless, in conducting a breeding programme directed towards obtaining
clones one soon realizes that the problem is not as simple as might be concluded at
first sight.
A good habit, a high yielding ability and a satisfactory performance of a mother
tree in other respects by no means offer a guarantee of obtaining a superior clone.
The reason why there is so little correspondence between the mother tree and the
clone derived from it is still incompletely understood. Probably only a few mother
trees respond well to the process of vegetative reproduction and the forced symbiosis
with an alien rootstock. If this conclusion - based on experience gained with clones
obtained by means of grafting - should prove to be right then vegetative propagation
by means of cuttings is likely to show a closer correlation between the performance of
a mother tree and its clonal offspring. However the required evidence on this point
is still lacking. For the time being it is essential to test large numbers of clones to
find a limited number of really outstanding vegetative descendants. Planted in
a way that ensures cross-fertilization these mixtures of clones equal or sometimes
even exceed the productivity of the best seedling families. In addition clones have the
advantage of being completely uniform as regards outward appearance of the trees,
production rhythm, size and shape of the beans and organoleptic properties.
Testing of clones should be done in narrow plots, often comprising a single tree
row only, in order to ensure cross-pollination. Randomization and a sufficient number
of replications are essential. Records should cover a period of five years at least.
The time required for developing and testing a clone can be put at six to ten years,
which means a considerable gain in time in comparison with that needed for develop­
ing a seedling family.

When applying clonal material for practical purposes one strict rule dictated by
bitter experience should be observed. Having regard to the pronounced allogamous
nature of Coffea canephora monoclonal planting should be avoided otherwise self-
incompatibility will exclude virtually all fruit setting particularly in the central part
of plantations of this type. If clonal material is to show to full advantage it should be
planted in mixtures that have been judiciously composed. The prospective members
of such a mixture should flower simultaneously, be inter-compatible and produce
beans of similar shape and size.
Inferences about compatibility relations can be drawn from the fruit setting percen­
tage obtained in diallel test crosses involving all the prospective members. A diagram
illustrating the compatibility relations in such a series of test crosses is given in table 4.
An abnormally low setting percentage, particularly when it is observed for some

230
 coffee

Table 4 Diagram illustrating compatibility relationships between six clones of C. canephora.

+ = compatible, - = incompatible, ± = uncertain. Pollinations not made are left blank.
B and D are inter-incompatible.

clones A B C D E F

A - + ± +
B + - ±
C + + - +
D ± - +
E + + + -

F + + + -

years in succession is indicative of cross-incompatibility. The partners of such a cross

should not be included in a clonal mixture.
Cross-incompatibility, as far as the present experience goes, seems to be a rather
rare occurrence. From the extensive test-cross series carried out in Java (Lambers,
1933; Ferwerda, 1936b) only two cases of non reciprocal cross-incompatibility have
been reported. These data are at variance with those published by Mendes (1949)
who found 35 out of 71 combinations incompatible.
Mendes herself ascribes the discrepancies to a higher degree of genotypic uniformity
among the C. canephora material which was used in her experiments in Brasil. Al­
though cross-incompatibility was not considered a matter of serious concern by the
Dutch research workers in Indonesia, the possibility of its occurrence was nevertheless
taken into account especially in mixtures where only a limited number of clones was
involved. In multiclonal mixtures cross-incompatibilities are less likely to have any
detrimental effect.
The beneficial effect of interplanting clones has been demonstrated by replicated
trials in which a number of clones, single or in mixtures, were compared. When inter-
planted these clones were found to yield approximately 50 % more than in monoclonal
blocks. In reality the contrast is even more striking because the yields of the monoclonal
blocks were flattered as a consequence of border effects resulting from cross-pollina­
tion by the adjacent plots (Ferwerda, 1941).
Despite the fact that clones have been obtained that can easily compete with the
best seedling strains this material has found little acceptance for commercial planting-
even in Java where the longest experience with clones has been gained. Clonal ma­
terial has there been mainly used, on a restricted number of estates, as an efficient
means of improving seedling plantations by topworking the potentially low yielding
individuals or by restoring damaged trees.
Particularly if scions are taken from lateral branches growing obliquely upwards,
grafts are obtained that grow faster, fill gaps earlier and reach the fruit bearing stage

231
A. CARVALHO, F. P. FERWERDA ET AL.

sooner than those made with orthotropic scions. Healthy but poor yielding individuals
can thus be converted into prolific shrubs that soon regain the fruit-bearing stage
thanks to the fact that branch grafts of the proper type skip the juvenile stage.
The merits of the different types of graftwood in Coffea have been investigated by
research workers in Java during the years preceding World War II. (Meyer, 1939;
Lambers, 1939). Problems and possibilities of branch grafting have been clearly and
comprehensively reviewed by Coolhaas (1953).
One fact that should be borne well in mind is that good performance of a clone
propagated as orthotropic graft does not offer any guarantee as to its suitability when
used as branch graft. Obtaining good branch-, or whip graft clones requires a separate
selection programme aiming at these specific properties. Selective topworking carried
out systematically and thoroughly cannot but result in a considerable improvement.
The progress attained is, however, difficult to express in exact figures as untreated
check plots are lacking in most of the cases.
General appearance and fruit bearing habits of selectively topworked gardens im­
prove so strikingly that there is no reason to question the effectiveness of this method.
Inferences about the progress to be expected may be made from the frequency dis­
tribution of individual tree yields of a population and from the yielding capacity
of the clones to be used. Estimates are in the neighbourhood of 30-35 % (Coolhaas,
1941, 1953).
In the preceding pages it has been tacitly ignored that a graft is a dual being, the
product of the artificial union of two different individuals. In this enforced compan­
ionship the rootstock is as important a partner as the scion. Our knowledge with
regard to the mutual influence of rootstock and scion is limited. The basis to the study
of this complex of problems was laid by Cramer (1928) and the result of two systematic
rootstock experiments have been published by Schweizer and's Jacob (1938). The
evidence collected and published leads to the conclusion that Robusta clones should
preferably be grafted on to Robusta rootstocks. Robusta grafts placed on C. excelsa
rootstocks take and develop much less satisfactorily whereas C. dewevrei s.s. and
C. canephora are distinctly incongenial when combined by grafting. On the other hand
C. excelsa has proved to be very suitable as a rootstock to interspecific uganda-
congensis (Congusta) and liberica-arabica hybrids.
One or two robusta strains have been found which, apart from giving satisfaction
as a rootstock for many clones, also show a marked resistance to root nematodes
(Tylenehus sp. sp.) and probably offer possibilities for growing coffee on nematode
infested soils. In this respect grafted coffee possesses specific advantages in comparison
to clonal material raised from cuttings and consequently standing on their own roots.
For this reason grafting despite its being a rather elaborate procedure will maintain
its position side by side with the far simpler method of rooting softwood cuttings
mentioned on page 191.
The experience with Robusta clones obtained by the latter method of vegetative
propagation is still relatively limited. Dublin (1967) reports the results of a clonal

232
 COFFEE

selection programme conducted at the Boukoko experiment station (Central African

Republic) since 1956. On the strength of a very critical selection which not only
comprised the yielding capacity but also other properties such as bean size, cup quality
and resistance to diseases and pests, a number of 368 mother trees were chosen and
propagated into clones. These were planted in replicated trials for testing.
The data so far compiled comprise only six harvest years for the clones longest under
observation and only permit the provisional choice of 27 clones that look really out­
standing. A direct comparison between these prominent clones and the best clones of
Yangambi (Congo Kinshasa) and Java is not possible as these are not included in the
trials. In comparison to the seedling check plots containing mixed clonal seedlings the
Boukoko clones were distinctly superior.
Continued observation, especially in local test trials which have already been started
will in the near future make possible a definite judgment of this material.
As a consequence of their superficial root system cuttings keep a less firm hold to
the soil than seedlings or grafts. This is considered a drawback but continued obser­
vation will decide how seriously this disadvantage is to be taken.

Interspecific hybridization

Interspecific hybridization in Coffea is known to be possible. In fact it has repeated­

ly occurred where two species were growing in close proximity to each other and in­
tercrossed spontaneously.
Artificial crosses between species have only incidentally been made and a systematic
study of crossability within the genus Coffea is still lacking. Bouharmont's (1959)
survey of the chromosome affinities in Coffea leads to the conclusion that there is no
reason to exclude the possibility of making crosses between species with the same chro­
mosome number. This certainly offers prospects for the large group of diploid coffee
species.
Spontaneous crosses between diploid species are known (example: C. ugandae X
C. congensis) as well as those between diploids and tetraploids (C. liberica X C.
arabica; C. arabica x C. canephora). Remote species such as C. kapakata- by some
authors not even recognized as true coffee - have been artificially hybridized with
C. canephora and C. arabica (Lambers, 1935 ; Monaco and Medina, 1965).
On a restricted scale some of the spontaneous hybrids such as the Congusta (C.
ugandae x C. congensis) hybrids and the Hemileia tolerant liberica-arabica hybrids
have found acceptance in commercial planting in Indonesia. Particulars about the
origin of these hybrids are mentioned by Cramer (1934, 1948). As these hybrids never
breed true to type they can only be propagated vegetatively. Congusta clones have
been found to perform well in the high mountainous regions of Java at elevations up
to 900 m where robusta no longer thrives, giving yields never attained by robusta
under such conditions. They are very precocious and able to bear a reasonable crop at
the age of three years. Congusta is not decidedly self-incompatible although seed is

233
A. CARVALHO, F. P. FERWERDA ET AL.

more readily set from foreign pollination. Consequently planting of clonal mixtures is
preferable (Leupen, 1950; Berczy andGude, 1955; Ferwerda, 1941).
The Uganda x congensis cross has been artificially performed in Indonesia and
Congo Kinshasa (Leupen, 1950; Ineac, 1956).
A number of promising new clones have resulted from the series mentioned first.
Hybrids between C. canephora and C. arabica have been obtained in various coun­
tries either by artificial crossing or spontaneously. Nowhere did they go beyond the
stage of experimentation. A disadvantage generally encountered is the high percentage
of seeds with a defective endosperm (spongy beans). Fertile 66-chromosome arabica-
robusta amphidiploids have been reported by Brazilian research workers (Krug and
Carvalho, 1952; Mendes, 1947).
Interspecific hybridization deserves much more attention than it has received thus
far. Particularly when such properties as a low caffeine content are the main objective,
interspecific hybridization including wild species from Malagasy which reportedly
contain very little or almost no caffeine (Coste, 1965) might be attempted.

Progress obtained by breeding and desiderata for future work

Long term research programmes such as have been conducted in Indonesia and
Congo Kinshasa have led to the production of synthetic varieties and F, seedling
families of which the yielding capacity was practically twice that of the unselected
starting material. Similar gains are to be expected from the breeding programmes now
under way in the Central African Republic and Ivory Coast. Bean shape and size have
also been considerably improved and the first steps have been taken towards improv­
ing the organoleptic properties.
In continuation of breeding work the same striking gains as in the first breeding
cycles, cannot be expected. In the early phases of a breeding programme when the
breeding stock has only slightly departed from the wild or semi-wild initial material
progress in breeding is achieved by strides whereas in the following cycles only a step
by step improvement may be expected.
The breeding system described in the preceding pages inevitably leads to a narrow­
ing of the genetic bases and a growing uniformity in genetic constitution, so that at a
certain moment a plateau level will be reached. In order to transcend this plateau or,
better still, to avoid its becoming established, a continuous introduction of 'new
blood' will be required. In order to have such source material ready at hand it would
be useful to maintain a 'germ plasm bank' as diversified as possible. The wild and
semi-wild coffees available in endless variation in their natural habitat in Central
Africa would be invaluable for this purpose. They have only been partly explored and
hardly yet assessed as to their breeding potentialities.
The vegetative selection has provided clones which when interplanted with well
matched partners, are capable of yields equalling or surpassing those of the best seed­
ling strains and have the additional advantage of a greater uniformity in cropping

234
 COFFEE

season and shape and size of the bean. Improvement of organoleptic properties and
of suitability for making evaporated coffee (instant coffee) very probably can be
achieved easier by clonal selection than by breeding proper.
To all appearances clones obtained by rooting softwood cuttings will supersede
those from grafts - formerly the only type of vegetatively propagated material.
In regions where agriculture is at high level and where coffee is treated like a fruit
tree rather than like a field crop sophisticated methods such as selective top-working
and the use of certain rootstocks in order to obtain a distinct type of growth or resis­
tance against root nematodes may still offer practical opportunities.
In the course of half a century of coffee breeding satisfactory results have been
obtained, but considerably more can be achieved provided that the breeding program­
mes in progress can be continued uninterruptedly and breeders in various countries
join efforts by co-operating more closely than in the past. To this end meetings of
technical working parties like the one organized by F.A.O. and held in Rio de Janeiro
in October 1965 can be a great stimulus.

References

ANTUNES, H. (Filho) and CARVALHO, A., 1954. Melhoramento do cafeeiro. VII - Ocorrência de lojas
vazias em frutos de café 'Mundo Novo'. (Coffee breeding. VII - Empty fruits locules in the Mundo
Novo coffee). Bragantia, 13 (14): 165-179.
BÉRCZY, L. and GUDE, C. J., 1955. Praktijkenquête Conuga-koffie (Opinion poll on Conuga coffee).
Bergcultures, 24: 557-563.
BOUHARMONT, J., 1959. Recherches sur les affinités chromosomiques dans le genre Coffea. Publ.
INEAC Série Scientifique, 77: 94 pp.
BRAUDEAU, J., CAPOT, J., DUBLIN, P., and Foury, C., 1962. Les principes de sélection des caféiers
canephoroïdes et libérico- excelsoïdes. Bull. No. 5 Inst. Français du Café et du Cacao.
BUTT, D. J., 1966. The hybridization of Coffea canephora Pierre. A technique using waxed cartons.
Euphytica, 15: 87-89.
CAPOT, J., 1962. Le caféier Robusta dans la cuvette centrale Congolaise. Bull. d'Inf. INEAC, 11: 33-
40.
CARVALHO, A., 1952. Melhoramento do cafeeiro. VI - Estudo e interpretaçào, para fins de seleçào, de
produçôes individuais na variedade bourbon (Breeding of the coffee tree. VI - Investigation and
interpretation of individual tree yields of the bourbon variety for selection purposes). Bragantia,
12 (4-6): 179-200.
CARVALHO, A., 1958. Advances in coffee production technology. Recent advances in our knowledge
of coffee trees. II Genetics. Coff. Tea Ind. Flav. Field, 81: 30-36.
CARVALHO, A., 1959. Genética de Coffea. XXIV - Mutantes de Coffea arabica procedentes da Etiopia
(Genetics of Coffea. XXIV - Mutants of Coffea arabica from Ethiopia). Bragantia, 18 (25) : 353—
371.
CARVALHO, A., 1962. Variability of the niacin content in coffee. Nature, 194: 1896-1907.
CARVALHO, A. and KRUG, C. A., 1949. Agentes da polinizaçao da flor do cafeeiro (Coffea arabica L.)
(Pollinating agents of the coffee flower (Coffea arabica L.)). Bragantia, 9 (1-4): 11-24.
CARVALHO, A. and MONACO, L. C., 1959. Hîbridos entre Coffea e Psilanthopsis (Hybrids of Coffea
and Psilanthopsis). Bragantia, 18 (4): XXI-XXIX.
CARVALHO, A. and MONACO, L. C., 1962. Natural cross-pollination in Coffea arabica L. Proceedings

235
A. CARVALHO, F. P. FERWERDA ET AL.

XVIth International Horticultural Congress, Brussels, 4: 447-449.

CARVALHO, A. and MONACO, L. C, 1965. Melhoramento do caffeeiro. XXVI - Fatôres que afetam a
produçào (Coffee breeding XXVI - Factors influencing productivity). Rel. Seçào Genética do Inst,
agronômico de Campinas, 1965 (Mimeographed, non published).
CARVALHO, A., MONACO, L. C. and ANTUNES, H. (Filho), 1959. Melhoramento do cafeeiro. XV -
Variabilidade observada em progênies de café. (Coffee breeding XV - Variability observed in coffee
progenies). Bragantia, 18: (26) : 373-386.
CARVALHO, A., MONACO, L. C. and SCARANARI, H. J., 1962. Melhoramento do cafeeiro. XXIV -
Variaçâo na produtividade de cafeeiros importados, com referência especial material da Etiópia e
do Sudäo (Coffee breeding XXIV - Yield variability of imported coffee, with special reference to
the coffee from Ethiopia and Sudan). Bragantia, 21 (13): 215-239.
CARVALHO, A., SCARANARI, H. J., ANTUNES, H. (Filho) and MONACO, L. C., 1961. Melhoramento do
cafeeiro. XXII - Resultados obtidos no ensaio de seleçôes regionais de Campinas (Coffee breeding
XXII - Results from the regional coffee selection trial from Campinas). Bragantia, 20 (30): 711-
740.
CARVALHO, A., TANGO, J. S., and MONACO, L. C., 1965. Genetic control of the caffeine content of
Coffee. Nature 205: 314.
CEVALLOS, J. V., 1963. Respuesta de algunas variedades de café a la irradiacion gamma (Response of
some coffee varieties to gamma irradiation). Turrialba, 13 (1): 37-38.
CHEVALIER, A., 1947. Les caféiers du globe. III: 1-356. Paul Lechevalier, Paris, 1947.
COFFEE RUSTS RESEARCH CENTER - OEIRAS - PORTUGAL, 1965. Progress Report 1960-1965. First
Session Technical Working Party on Coffee Production and Protection - Rio de Janeiro, Brazil.
COOLHAAS, C., 1941. Wat kan men met tuinverenting in de koffie bereiken? (What can be attained
with selective topworking in coffee?). Bergcultures, 15: 62-71.
COOLHAAS, C., 1953. L'application des greffes de la branche dans la culture du caféier Robusta de
Java. Neth. Journ. Agric. Sei., 1: 130-136.
COSTE, R., 1965. The wild coffees of Madagascar. FAO Technical Working Party on Coffee Produc­
tion and Protection. Brazil 1965. Working Paper Ce/65/29.
CRAMER, P. J. S., 1928. Groeicijfers bij koffie (Growth figures in coffee). Arch. Koffiecultuur, 2: 4-52.
CRAMER, P. J. S., 1934. Early experiments on grafting coffee in Java. Emp. J. Exp. Agric., 2: 200-204.
CRAMER, P. J. S., 1948. Les caféiers hybrides du groupe Congusta. Bull. Agric. Congo B., 39: 29-48.
CRAMER, P. J. S., 1957. Review of literature on coffee research in Indonesia. Turrialba, C. Rica, Inst.
Interam. Ciências Agrfcolas, (Miscellaneous publication No. 15).
DEDECA, D. M., 1958. Advances in coffee production technology. Recent advances in our knowledge
of coffee tree. 4 - Anatomy. Coff. Tea Ind. Flav. Field., 81: 44-51.
DEVREUX, M. c.a., 1959. Recherches sur l'autostérilité di caféier robusta (Coffea canephora Pierre).
Publ. INEAC. Série scientifique, 78: 44 pp.
DHALIWAL, T. S., 1965. Progress report of the project coffee breeding in Puerto Rico 1954-1965.
University of Puerto Rico, Agric. exp. sta. Rio Piedras, Puerto Rico. Miscellaneous publication
No. 53.
DUBLIN, P., 1957a. Recherches sur la floraison et la fructification du caféier de la "Nana". Agronomie
Tropicale, 12: 173-203.
DUBLIN, P., 1957b. Biologie florale du Coffeadewevrei de Wild, race Excelsa-A. Chevalier. Ire partie:
Miscrosporogénèse et étude du gaméthopyte mâle. Agronomie Tropicale, 12: 576-594.
DUBLIN, P., 1958. Biologie florale du Coffea dewevrei de Wild, race Excelsa A. Chevalier. 2me partie:
Mégasporogénèse et étude du gamétophyte femelle. Agron. Trop., 13: 704-731.
DUBLIN, P., 1960. Biologie florale du Coffea dewevrei de Wild, race Excelsa A. Chevalier. 3e partie:
Floraison, pollinisation, fécondation. Agron. Trop., 15: 189-212.
DUBLIN, P., 1967. L'amélioration du caféier robusta en République Centrafricaine: dix années de
sélection clonale. Café Cacao Thé, 11 (2): 101-138.

236
 COFFEE

ECHANDI, E. and FERNANDEZ, C. E., 1962. Relación entre el contenido de âcido clorogénico y la
resistencia a la llaga macana o cancer de los cafetos causado por Ceratocystis fimbriata (Relation
between chlorogenic acid content and resistance to llaga macana disease of coffee trees caused by
Ceratocystis fimbriata). Turrialba, 12 (2): 87-90.
FABER, F. C. VON, 1912. Morphologische-physiologische Untersuchungen an Blüten von Coffea-
Arten. Ann. Jard. Buitenzorg, Java. 2 ser. 10 (= 25): 59-160, taf. 1-12.
FAGERLIND, F., 1937. Embryologische, zytologische und bestäubungsexperimentelle Studien in der
Familie Rubiaceae nebst Bemerkungen über einige Polyploiditätsprobleme. Acta Hort. Berg., 11
(1937): 195-470.
FAGERLIND, F., 1939. Perisperm oder Endosperm bei Coffea? Svensk Bot. Tidskrift, 1939, 33: 303-
309.
FERNIE, L. M., 1958. Advances in Coffee Production Technology. The supply of Better Planting
Material. 3- Asexual Propagation of Coffee. Coff. Tea Ind. Flav. Field, 81 (11): 64-67.
FERNIE, L. M., 1965. Coffee breeding in Tanganyika. (First session of the Technical Working Party on
coffee production and protection. FAO. Rio de Janeiro, October 1965 (typewritten, non published)).
FERWERDA, F. P., 1932. Enten versus zaailingen bij koffie (Grafts versus seedlings in coffee). Berg-
cultures, 6: 629-645.
FERWERDA, F. P., 1936a. Enkele waarnemingen over de bestuiving van koffie (Some observations on
the pollination of coffee). Arch. v.d. Koffiecult. Ned.-Ind., 10: 32^11.
FERWERDA, F. P., 1936b. Die Befruchtungsverhältnisse bei den in Niederl jndisch-[ndien angebauten
Kaffeearten. Der Züchter, 8: 92-101.
FERWERDA, F. P., 1937a. Kiemkracht en levensduur van koffiestuifmeel (Germinative power and
longevity of coffee pollen). Arch, koffiecult., 11: 135-150.
FERWERDA, F. P., 1937b. Gegevens over de ontwikkelingsduur der vrucht bij verschillende koffie-
soorten (Data on the duration of development of the fruit in various coffee varieties and species).
Bergcultures, 11: 198-203.
FERWERDA, F. P., 1941. Resultaten van enige mengproeven met koffiecloonen op Bangelan (Results of
some trials with clonal mixtures on Bangelan estate). Bergcultures, 15: 557-566.
FERWERDA, F. P., 1948. Coffee breeding in Java. Econ. Bot., 2:258-272.
FERWERDA, F. P., 1954. A tentative breeding method for Robusta and other allogamous coffee species.
Euphytica, 3: 12-19.
FERWERDA, F. P., 1958. The supply of better planting material. II Canephoras (Robustas). Coffee Tea
Industries (New York), 81: 58-63.
FRESSANGES, R., 1954. La sélection du caféier en Côte d'Ivoire. Contribution à l'étude du caféier en
Côte d'Ivoire. Bull, scientifique, Min. France d'Outre-Mer, 5: 223-231.
GARRUTI, R. S. and CONAGIN, A., 1961. Escala de valores para a avaliaçao da qualidade da bebida do
café (A scoring for quality evaluation of the coffee beverage). Bragantia, 20 (18): 557-562.
GILBERT, S. M., 1938, 1939. Variability in yield of Coffea arabica. East Afr. Agric. Jour., 4: 131-139,
249-259.
GRANER, E. A., 1936. Megasporogenesis in Coffea arabica L. Archivos Inst. Biol. Veget. Rio de
Janeiro, 3 (1): 3-69.
GRANER, E. A., 1938. Embryogenese de Coffea. I - Desenvolvimento do óvulo em 'C. arabica L.'.
(Embryogenesis of Coffea. I- Development of the ovule in C. arabica L.) Anais Prim. Reun. Sulam.
Bot., 3: 193-202.
GUDE, C. J., 1956. Some thoughts concerning the selection of Robusta in connection with the 1956
planting recommendation. Bergcultures, 25: 159-165. PBA Vol. XXVI No. 3599.
GUDE, C. J., 1957. Results of trials of Robusta coffee on the Gunung Gumitir estate and at the Sumber
Asin Experimental Station. Bergcultures, 26: 19-28. PBA Vol. XXVII, No. 3 Abst. No. 3336.
GUTIERREZ, G. Z., 1965. Various aspects of the genetical improvement programme carried out
by the Department of Coffee investigations of the Ministry of Agriculture and Animal husbandry

237
A. CARVALHO, F. P. FERWERDA ET AL.

of Costa Rica. First session of the Technical Working Party on coffee production and protection.
FAO. Rio de Janeiro, October 1965 (Typewritten, non published).
HAAN, H. R. M. DE, 1923. De bloembiologie van Robusta koffie (Floral biology of Robusta coffee).
Meded. Proefst. Malang, 40: 97 pp.
HEYN, 1938. Du nombre du chromosomes chez quelques végétaux en culture aux Indes Néerlandaises
(Coffea, Ceiba, Oryza, Derris et Palaquium). Ann. J. Bot. Buitenzorg, 48: 103-120.
HOUK, W. G., 1938. Endosperm and perisperm of coffee, with notes on the morphology of the ovule
and seed development. Am. J. Bot., 25: 56-61.
[NEAC, Rapport annuel 1956: 67-70.
Instituto Salvadoreno de Investigaciones del Café, 1963. Santa Tecla - El Salvador. Informe de La­
bores, 1963: 48-101. (Salvadorian Institute of Coffe investigations. Santa Tecla - El Salvador.
Annual progress report, 1963: 48-101.)
JACOB, J. C. 's, 1938. Results of seedling testgardens in Besuki. Arch. Koffiecultuur, 12: 57-95.
KRUG, C. A., 1935. Hybridization of coffee. J. Hered., 26: 325-330.
KRUG, C. A., 1937. Cytological observations in Coffea III. J. of Gen., 34: 399—414.
KRUG, C. A., 1958. Advances in coffee production technology. The supply of better planting material.
I - Arabicas Coff. Tea Ind. Flav. Field, 81: 52-57.
KRUG, C. A., 1959. World coffee survey. FAO. Rome, 1959, 292 pp.
KRUG, C. A. and CARVALHO, A., 1939. Genetical proof of the existance of coffee-endosperm. Nature,
Lond., 144: 515.
KRUG, C. A. and CARVALHO, A., 1951. The genetics of Coffea. Advanc. Genet., 4: 127-158.
KRUG, C. A. and CARVALHO, A., 1952. Melhoramento do cafeeiro V. Melhoramento por hibridaçâo
(Coffee breeding V. Improvement by hybridization). Bragantia, 12: 141-152.
KRUG, C. A. and MENDES, A. J. T., 1940. Cytological observations in Coffea IV. Journ. Gen. 39:
189-203.
LAMBERS, M. Hille Ris, 1929. Een en ander over koffieselectie (Something about coffee breeding).
Bergcultures, 3: 1924-1931.
LAMBERS, M. Hille Ris, 1932. Vijf jaar selectiewerk op Soember Asin (Five years of coffee breeding at
Sumber Asin). Arch. Koffiecult., 6: 1-42.
LAMBERS, M. Hille Ris, 1933. Nieuwe gegevens over de koffieselectie en de Soember Asin nummers
(New data on coffee breeding and the Sumber Asin strains). Bergcultures, 7; 587-598.
LAMBERS, M. Hille Ris, 1934. Over kunstmatige kruisbestuiving van koffie tuinen (Artificial pollina­
tion of coffee orchards). Bergcultures, 8: 880-885.
LAMBERS, M. Hille Ris, 1935. Kapakata koffie (Kapakata coffee). Arch. Koffiecultuur, 9: 105-115.
LAMBERS, M. Hille Ris, 1939. The branchgraft problem (Du.; Engl, summary). Arch. Koffiecultuur,
13: 70-86.
LEBRUN, J., 1941. Recherches morphologiques et systématiques sur les caféiers du Congo. Publ.
I.N.E.A.C. hors serie, 1941 : pp. 183.
LEJEUNE, J. B. H., 1958. Rapport au gouvernement Impérial d'Ethiopie sur la production caféière.
Rome, FAO, 1958, 49 pp (Rapport 797).
LELIVELD, J. A., 1938. Vruchtzetting bij koffie (Fruit setting in coffee). Arch. Koffiecult. Ned. Ind., 12:
127-161.
LELIVELD, J. A., 1939. Cytologische gegevens betreffende enige clonen van Coffea 'Robusta' (Cytolo­
gical data on a number of Coffea Robusta clones) (Dutch, Engl, summary). Arch. Koffiecult. Ned.
Ind., 13: 2-25.
LELIVELD, J. A. FRAHM, 1940. Nieuwe gezichtspunten voor het selectieonderzoek bij soortskruisingen
van koffie naar aanleiding van cytologische gegevens (New aspects for breeding interspecific hy­
brids in coffee based on cytological data). Bergcultures, 14: 370-386.
LELIVELD, J. A. FRAHM, 1940. Ontstaan en voorkomen van rondboon en voosboon bij koffie (Origin
and incidence of pea-beans and spongy beans in coffee). Bergcultures, 14: 1358-1362.

238
 COFFEE

LEON, J., 1962. Especies y cultivares (variedades) de café. (Species and cuitivars (varieties) of coffee).
Inst. Interam. Ciências Agricolas, Turrialba Costa Rica. Materiales de ensenanza de café y cacao,
no. 23, 1962, 69 pp.
LEON, J., 1965. Intravarietal hybridization in Coffea arabica work carried out at IAIAS, Turrialba,
Costa Rica. First Session Technical Working Party on Coffee Production and Protection. Brazil,
1965.
LEON, J. and FOURNIER, L. A., 1962. Crecimiento y desarollo del fruto de Coffea arabica L. (Growth
and development of the fruit of Coffea arabica L.). Turrialba 12: 65-72.
LEROY, J. F., 1962. Prospection des caféiers sauvages. Rapport préliminaire sur une mission scienti­
fique à Madagascar et aux Isles Mascaréignes. J. Agric. Trop, et de Bot. Appl., 9: 221-244.
LEROY, J. F. and PLU, A., 1966. Sur les nombres chromosomiques des Coffea malgaches. Café, Cacao,
Thé, 10: 216-218.
LEUPEN, F. F., 1950. Conuga coffee. Bergcultures, 19: 377-383.
MEDINA, D. M., 1950. Observaçôes citológicas em Coffea. XIV- Microsporogênese em Coffea arabica
L. var. rugosa KMC. (Cytological observations in Coffea. XIV - Microsporogenesis in Coffea
arabica L. var. rugosa KMC). Bragantia, 10 (2): 61-66.
MEDINA, D. M., 1960. Macrosporogênese, formaçâo e desenvolvimento do saco embrionârio, do
endosperma e do embriâo em Coffea dewevrei De Wild et Th. Dur. (Megasporogenesis, formation
and development of embryosac and embryo in Coffea dewevrei De Wild et Th. Dur.). Bragantia,
19 (48): 767-784.
MENDES, A. J. T., 1941. Cytological observations in Coffea. VI - Embryo and endosperm develop­
ment in Coffea arabica L. Amer. J. Bot., 28: 784-789.
MENDES, A. J. T., 1946. Partenogênese, partenocarpia e casos anormais de fertilizaçâo em Coffea.
(Parthenogenesis, parthenocarpy and cases of abnormal fertilization in Coffea). Bragantia, 6 (6):
265-273.
MENDES, A. J. T., 1947. A citologia e o melhoramento do cafeeiro. (Cytology and improvement of the
coffee plant). Bol. Sup. Serv. Café, 22 (242): 236-240.
MENDES, A. J. T., 1947. Observaçôes citológicas em Coffea XI. Metodos do tratamento pelo colchicina
(Cytological observations on Coffee XI. Methods of colchicine treatment). Bragantia, 7: 221-
230.
MENDES, A.J.T., 1950. Observaçôes citológicas em Coffea. XVI - Microsporogênese em Coffea arabica
L. (Cytological observations in Coffea. XVI - Microsporogenesis in Coffea arabica L.) Bragantia,
10(3): 79-87.
MENDES, A. J. T., 1958. Advances in coffee production technology. Recent advances in our knowledge
of coffee trees. 3 - Cytology. Coffee and Tea Industries, 81: 37-42.
MENDES, A. J. T. and MEDINA, D. M., 1955. Contrôle genéticodos 'frutos chôchos' no café 'Mundo
Novo'. (Monofactorial inheritance of 'empty-locule' in the Mundo Novo coffee). Bragantia, 14
(9): 87-99.
MENDES, C. H. T., 1949. Introduction to the study of self sterility in the genus Coffea. Bragantia, 9:
35-41.
MENDES, C. H. T., 1950. Observaçôes citológicas em Coffea. XVI Microsporogenese em C. canephora
Pierre ex Froehner (Cytological observations in Coffee. XVI Microsporogenesis in C. canephora).
Bragantia, 10: 97-104.
MENDES, C. H. T., 1950. Observaçôes citológicas em Coffea. XVII - O saco embrionârio em Coffea
canephora Pierre ex Froehner (Cytological observations in Coffea. XVII - The embryosac in Coffea
canephora Pierre ex Froehner) Bragantia, 10: (4): 105-111.
MEULEN, A. VAN DER, 1939. On the structure and the periodical development of the flower buds of
Coffee species (Du.; Engl. summ.). Proc. Roy. Neth. Acad, of Sei. 38 (2): 1-128.
MEYER, F. G., 1965. Notes on wild Coffea arabica from Southwestern Ethiopia, with some historical
considerations. Ec. Bot., 19 (2): 136-151.

239
A. CARVALHO, F. P. FERWERDA ET AL.

MEYER, W. H., 1939. The branchgraft problem (Du.; Engl, summ.) Arch. Koffiecultuur, 13: 51-69
and 85-86.
MOENS, P., 1962. Les bourgeons végétatifs et génératifs de Coffea canephora. (Étude morphologique et
morphogénétique). La Cellule, 63 (1962/63): 165-244.
MOENS, P., 1965a. Développement et embryogénèse chez Coffea canephora Pierre. La Cellule 65:
129-147.
MOENS, P., 1965b. La vascularisation de l'appareil floral des caféiers canephora et arabica. La Cel­
lule, 65: 247-293.
MOH, C. C , 1961. Does a coffee plant develop from initial cell in the shoot apex of an embryo? Radia­
tion Botany, I: 97-99.
MONACO, L. C., 1964. Report on a voyage to Ethiopia for collecting and preserving the germ plasm of
Coffea arabica L. Campinas, Inst, agronômico, 1964, 26 pp. (Mimeographed, in Port.).
MONACO, L. C., 1968. Consideration about the genetic variability of the coffee population from Ethio­
pia. (Report in preparation by FAO).
MONACO, L. C. and CARVALHO, A., 1963. Coffee breeding XXIII: New data on the variability of
isogenic lines in coffee. (Port.; Engl, summ.) Bragantia, 22: 117-123.
MONACO, L. C., and CARVALHO, A., 1964. Genética de Coffea. Hïbridos entre espécies (Genetics of
Coffea. Interspecific hybrids) Ciência e Cultura, S. Paulo 16 (2) 144.
MONACO, L. C. and MEDINA, D., 1965. Crossability between Coffea arabica and C. kapakata and cy-
tological analysis of the hybrid. Technical Working Party on Coffee Production and Protection.
FAO. Brazil, 1965.
MONGE, S. FERNANDO, 1962. Frecuencia de Café caracolillo en plantas provenientes de semillas irra-
diadas. (Frequency of peaberry in coffee plants originating from irradiated seeds). Turrialba, 12
(4): 209-210.
NARASIMHASWAMY, R. L., 1962. Some thoughts on the origin of Coffea arabica. Coffee 4 (12): 1-5
Turrialba, Costa Rica.
NOGUEIRA, P. (NETO), CARVALHO, A. and ANTUNES, H. (Filho), 1959. Efeito da exclusâo dosinsetos
polinizadores na produçâo do café bourbon (The effect of the exclusion of pollinating insects on the
yield of the Bourbon coffee). Bragantia, 18 (29): 441-468.
D'OLIVEIRA, B., 1958. Advances in coffee production technology. Selection of coffee types resistant to
the Hemileia leaf rust. Coff. and Tea Flav. Field, 81:112-120.
D'OLIVEIRA, B. and RODRIGUES, C. J. (Junior), 1960. I - A survey of the problem of coffee rusts. II -
Screening for resistance to Hemileia vastatrix of Coffea arabica. Junta de Exportaçao do Café,
Lisboa, Portugal, (Mimeographed) 46 pp.
ONG AN PANG, 1957. De aanleg van kleine koffietoetsproeven op ondernemingen (Small exploratory
estate trials with new coffee strains). Bergcultures, 26: 535-541. Trop. Abst., 13 (2): No. 1451.
PORTÈRES, R., 1946. L'atrophie florale non parasitaire chez les caféiers et les causes physiologiques.
Rev. Int. Bot. Appl. Agr. Trop., 26: 81-94.
PORTÈRES, R., 1947. Action de l'eau après une periode sèche sur le déclenchement de la floraison chez
Coffea arabica L. L'Agron. Trop., 1: 148-158.
PORTÈRES, R., 1962. Sur quelques caféiers sauvages de Madagascar. J. d'Agric. Trop, et de Bot. Appl.
9: 201-210.
RICK, C. M., 1950. Pollination relations of Lycopersicon esculentum in nature and foreign species.
Evolution, 4: 110-122.
RAMAIAH, P. K. a.o., 1965. Studies on the physiology of flowering and fruit growth in coffee (Coffea
arabica L.); ascorbic acid content in relation to the flower opening. Proc. Technical working party
on coffee production and protection. F.A.O., Rio de Janeiro, Oct. 1965, working paper Co 65/11
(Mimeographed).
ROBINSON, J. B. D., 1964. A handbook on arabica coffee in Tanganyika. Ed. Tanganyika Coffee
Board. 173 pp.

240
 COFFEE

SCHWEIZER, J. and JACOB, J. C. 's, 1938. Eerste resultaten van een tweetal onderstam proeven bij
Robusta enten op Kaliwining en in Djember (First results from two rootstock trials with Robusta
clones in the experimental gardens at Kaliwining and Djember). Bergcultures, 12: 1526-1532.
SNOEP, W., 1937. On variability in the production of individual trees in coffee. Bergcultures, II: 137—
144. P.B.A., 8(1937/38) Abs. 603.
SNOEP, W., 1940. Bepalingen omtrent de stuifmeelverspreiding door de lucht bij robusta-koffie (Ob­
servations on airborne pollen dispersal in robusta coffee). Arch. Koffiecultuur, 14: 275-281.
SÖDERHÖLM, P. K. and GASKINS, M. H., 1961. Evaluación de la resistencia al frio en el género Coffea.
(Evaluation of cold tolerance in the genus Coffea). Café 3 (9): 49-57. Turrialba, Costa Rica.
SYBENGA, J., 1960. Genetics and cytology of coffee. A literature review. Bibliographia Genetica, 19:
217-316. Also in: Turrialba, 10(3): 83-137.
SYLVAIN, P. C., 1958. Ethiopian coffee - Its significance to world coffee problems. Econ. Bot., 12:111-
139.
TANGO, J. S. and CARVALHO, A., 1963. Teor de óleo e de cafeina em variedades de café. (Oil and caffe­
ine content in the coffee bean). Bragantia, 22 (65): 793-798.
TANGO, J. S. and TEIXEIRA, C. G., 1961. Teor de cafeina em progênies de café (Caffeine content of
coffee progenies). Bol. Suptda Serv. Café. S. Paulo, 36: 6-10.
TASCHDJIAN, E., 1932. Beobachtungen über Variabilität, Dominanz und Vizinismus bei Coffea arabica
Zeitschrift für Züchtung, 17: 341-354.
THIRION, F., 1962. Vingt années d'amélioration de la culture du caféier Robusta à Yangambi. Bull.
INEAC, 1: 323-356.
THOMAS, A. S., 1947. The cultivation and selection of robusta coffee in Uganda. Emp. J. Exp. Agric.,
15: 65-81.
TOLEDO, O. Z. et al., 1963. Variabilidade em algumas das caracteristicas da bebida do café (Variability
in some of the coffee beverage characteristics). Bragantia 22 (31): 393-399.
VALLAEYS, G., 1956. L'amélioration du caféier Robusta. Bull. Inform. INEAC, 5: 27-37.
WORMER, T. M., 1964. The growth of the coffee berry. Ann. of Botany N.S., 28: 47-55.
WORMER, T. M., 1965. Some physiological problems of coffee cultivation in Kenya. Café (Lima,
Peru), 6: 1-19.
WORMER, T. M., 1966. Shape of bean in Coffea arabica in Kenya. Turrialba (Turrialba, Costa Rica),
16: 221-236.

241
DATE PALM

Phoenix dactylifera L.

J. H. M. OUDEJANS

CIBA (Pakistan) Limited, Dacca, East Pakistan

Introduction

The date palm (Phoenix dactylifera L.J is a tree, which, according to an Arab say­
ing, should grow with "its feet in running water and its head in the fire of the sky". Its
ability to produce fruits abundantly at extremely low humidities, at least if the supply
of groundwater suffices, makes the tree one of the chief producers of food in the desert
belt of North Africa and South-West Asia.

Systematics in connection with breeding work

TAXONOMY

The date palm (Phoenix dactylifera L.J belongs to the family of Palmae. Chevalier
(1952) divided the genus Phoenix into twelve species, all native to tropical and hot
subtropical parts of Africa and Asia. Apart from the main species P. dactylifera L.,
the names of P. atlantica Chev., P. canadensis Chabaud, P. reclinata Jacq. and P.
sylvestris Roxb. also deserve mention.

P. dactylifera distinguishes itself from P. sylvestris and P. canadensis by its ability

to produce offshoots or suckers and from the other species by its tall, columnar, relati­
vely thick trunk.

P. canadensis Chabaud, the Canary Island palm, is valued as an ornamental tree

along driveways and P. sylvestris is cultivated in India as a source of sugar.

P. atlantica Chev. is called 'spurious date' on account of its close resemblance to

P. dactylifera L. Its fruits are slightly less fleshy and hardly edible, although the variety
maroccana Chev. of that species produces fairly tasty fleshy fruits. According to Che­
valier (1952) and Munier (1962) the palm population of Marakech-Morocco, num­
bering about 15,000 trees, consists of selected individuals of the variety maroccana.
An important trait of P. atlantica is that this species has so far proved to be resistant to
the 'Bayoud' disease.

243
J. H. M. OUDEJANS

TROPIC OF
CANCER

EQUATOR

Fig. 1 Distribution of the date palm in North Africa and South-West Asia.

CYTOLOGICAL DATA

A chromosome number of n = 18, 2n = 36 was observed in six Phoenix species

and ten P. dactylifera cultivars investigated by Beal (1937). The chromosomes in all
species and varieties examined showed a marked similarity in size and shape. Accord­
ing to Beal (I.e.) this probably accounts for the interspecific fertility which has been
demonstrated through crosses made between several of the species.

DISTRIBUTION AND CENTRE OF ORIGIN

The principal cultivation area of the date palm in the Old World is outlined in fig. 1.
Moreover, the relatively recent commercial groves of South California in latitude
33° North should be mentioned, especially since much research work has been and
is being carried out in this area. The investigations are concerned with systematics,
cultivation and breeding problems (Nixon, 1966).
The most important countries producing dates are Iraq, Saudi Arabia, Iran, Egypt,
Algeria and West Pakistan. Although it is very difficult to assess precisely the number
of trees and, in particular, the individual yields in the vast desert areas, the world pro­
duction of dates in 1963 was estimated to be 1,440,000 metric tons and the number of
trees to be approximately 87,000,000.
Archeological and palaeontological research have provided some indications as to
the centre of origin of the date palm. The earliest findings on the genus Phoenix were
made in Eocene sediments in the Parisian basin and in young tertiary layers in Central
and South Europe. Kaul (1951) concluded that P. dactylifera was originally a European
rather than an Asian species on the basis of the fact that North Africa, the Middle

244
 date palm

East and South Asia were below sea level in the Tertiary Period. The cultivation of date
palms possibly started as early as the Neolithic Age (Munier, 1953). Chevalier (1952)
suggests that the date palm developed practically simultaneously in different places in
the belt between the Atlantic and the Indus. The traces dating back to the Third
Millenium B.C. found in West India and depictions of the Sumerian civilization and
the Ancient Egyptian Empire support his supposition. Furthermore the name 'date',
which is derived from the Hebrew word 'dacheb', points to a very ancient Near Eas­
tern origin.

Physiology of development

GERMINATION OF SEEDS

The narrow cylindrical seed, about 2.5 cm long, contains a hard endosperm and a
small embryo whose presence is indicated by a little circular germinal pore in the mid­
dle of the ventral part of the seed. While germinating, the tubular seedling grows out,
leaving the seed by way of the germinal pore. The rootlet completes its development
outside. The sole cotyledon remains inside, forming a haustorium, which communicat­
es with the seedling by means of a long tubular organ. There is no period of dormancy
in the germination of date palm seeds, therefore they can be sown at any time of the
year.

GROWTH

The date palm is a monocotyledonous plant, the terminal bud of which is the only
centre of vegetation. The development of the height of the date palm and the annual
production of leaves depend very much upon factors relating to variety and environ­
ment. In general, male date palms grow more rapidly than the female trees which bear
the fruits.
The productive capacity of a strong date palm ranges from 10-20 leaves a year. At
the adult stage the crown consists of 60-150 leaves, depending upon variety and en­
vironment (Nixon, 1950; Tothill, 1948). The extensive root system is situated deep in
the ground and includes many adventitious roots. An adult palm has roots which may
extend to over seven metres from the stem and reach a depth of more than six metres
(Fonteney, 1960).

VEGETATIVE MULTIPLICATION

The growth of suckers from the basal part of the stem makes it possible to propagate
the date palm vegetatively. A sucker is an exact replica of the tree from which it has devel­
oped. Once the suckers have been separated from the parent stem, they are capable of
forming their own root systems and of developing into productive palms. For cen-

245
J. H. M. OUDEJANS

turies on end date palms have almost exclusively been propagated vegetatively. Gene­
rative propagation as a rule results in a very heterogeneous progeny in many respects
inferior to the mother plants.
The technique of separating offshoots from their parent palms calls for great care
and skill. As suckers with a root system of their own are more likely to grow on after
separation from the parent plant, rooting is stimulated before separation (Lefèvre,
1961; Munier, 1955; Nixon, 1966 and Piquer, 1959). Modern cultivation methods
consist of letting the suckers develop on the parent tree for three to five years and
sometimes this period is much longer.
The proper cultivation technique is to keep the suckers in a nursery bed for 12-18
months and then to transplant them into the field (Nixon, 1966).

Biology of flowering

The date palm is dioecious, although in exceptional cases monoecious unisexual

flowers or even monoecious bisexual flowers seem to occur (Brown, 1924). Sex reversal
is also observed, viz. the transition from males that have long been producing pollen
to female fruitbearing trees (Ahmed, 1959).
Generally speaking, there are no secondary characteristics from which the sex of a
young tree before the first flowering can be inferred, although in some varieties ex­
perienced observers can differentiate male and female palms by their vegetative cha­
racters.

Inflorescence and flower

The inflorescence of either sex shows a branched cluster of spikes inserted on the
flower stalk completely surrounded by a fibrous leaflike sheath called the spathe. The
spathe of a female inflorescence is broader and longer than that of a male inflorescence.
The former is thick and pale brown while the latter has a rusty powdery coating. The
spathes slit lengthwise just before the first flowers open. After the spathes burst the
male inflorescences tend to hang slightly and their flower stalks do not grow much
longer, remaining erect or semi-erect in certain varieties.
The 40-50 cm long peduncle of the male inflorescence is flattened, smooth and
glabrous, ending in numerous spikeiets, each of them bearing 20-50 fragrant white
flowers. The oval <3 inflorescence, 25-60 cm long, may contain as many as 100 or 150
spikes which are seldom longer than 15 cm (fig. 2B). The far bigger female inflorescen­
ce may reach a length of 1.20-2.00 m (fig. 2A).Its smooth axis bears 25-100 spikes on
the top, the length of the axis ranging from 15-90 cm. Small globular flowers are si­
tuated rather close together along the spikes (fig. 3 and 4A).
Eight to ten thousand flowers are said to occur in a very big female inflorescence,
over 5000 fruits being counted on a particularly big unthinned bunch (Nixon, 1934).

246
 date palm

Fig. 2 Female (A) and male (B) inflorescence of the date palm, spathes just opening. (Photograph
U.S. Department of Agriculture).

247
J. H. M. OUDEJANS

Storage of pollen

Dry date pollen stored in a dry room at moderate temperatures will retain its via­
bility for two to three months. High temperatures adversely affect the longevity of
date pollen. Well dried pollen placed in an airtight container can be kept in cold stora­
ge (-18 °C) from one season to the next with very little loss of viability (Nixon, 1959).

Receptivity offemale flowers

The receptivity of female flowers reaches its optimum within three to four days after
the opening of the spathe but fair settings may be obtained in some cases up to 8 or 10
days (Péreau-Leroy, 1957).

TECHNIQUE OF ARTIFICIAL POLLINATION

Artificial pollination as a cultivational practice for improving fruit setting

The pollen is obtained by cutting strands of male flowers from a freshly opened male
inflorescence (Ahmed, 1959; Nixon, 1966; Péreau-Leroy, 1957). The flowering male
spikes are shaken over a mature female inflorescence whose split spathe has been re­
moved. The spikes are subsequently inserted upside down between the branches of the
female flower cluster. If the female palms should flower earlier than the male ones,
stored dried pollen is used. In this case the flowers are pollinated by putting pieces of
cotton dusted with pollen between the strands of the female flower cluster or by shaking
a muslin bag with dried and crushed male flowers over the newly opened female in­
florescences (Brown, 1924; Piquer, 1959).
Pollination is invariably carried out by specialists whose knowledge is transmitted
from father to son. This method is referred to as 'traditional' and must go back to
antiquity as even the cuneiform texts of Ur (about 2300 B.C.) made mention of it.
The above method of pollination implies that the labourers have to climb each
palm five to ten times in 40 days, which entails so much labour that during that period
one man can pollinate 250 female trees at the most (Mason, 1927 ; Wertheimer, 1957).
Pressure sprayers are now used in the modern plantations of California, South Algeria
and South Tunisia. These implements have long spray tubes, so that female inflores­
cences can be reached up to the height of 10 m when the trees are pollinated from the
ground. The latter procedure enables two men to treat 1000 palms a day.

Artificial pollination for breeding purposes

The technique of artificial pollination to obtain crosses for breeding purposes is

essentially the same as that mentioned in the foregoing section. Special care must be
taken to avoid contamination. Pollen is collected from male inflorescences that were

250
DATE PALM

251
J. H. M. OUDEJANS

bagged before the spathes opened. For pollination only unopened or bagged female
spathes are selected. Between handling different kinds of pollen the worker's hands,
arms and instruments must be cleaned thoroughly with water and soap or with 80 %
alcohol (Nixon and Furr, 1965).

FRUIT

After fertilization, the branches of the female inflorescence elongate to form a large
pendulous bunch of fruits.
One palm can produce 10-30 bunches. It is generally held that a date palm comes
into full bearing at the age of 10-12 years. It produces most when it is between 20 and
50 years old. The fruit is a one-seeded berry, 3-8 cm long and highly variable in shape,
colour and flavour.

THE OCCURRENCE OF METAXENIA

In date palms the tree whose pollen is responsible for the fertilization exercises a
direct influence on the somatic tissue of the fruit outside the embryo and the endo­
sperm. This influence, which Swingle (1928) called 'metaxenia', finds expression in
early or late ripening and in the difference in the size and colour of the fruits and seed
of one and the same tree pollinated with pollen from different varieties (Ahmed,
1959; Comelli, 1960; Munir and Niaz, 1961). In California Nixon (1928, 1934,
1935) made a detailed study of the phenomenon of metaxenia by pollinating a number
of varieties with pollen of diversified origins, including seedlings of well-known
varieties.
After the pollination of the well known Deglet Noor cultivar a marked difference
was observed between the pollen of seedling No. 4 of the variety Fard and that of the
variety Mosque. As compared with the former pollen the latter increases the size of
fruits by an average of 3-4 mm. What matters more than the increase in fruit size -
since the latter is influenced to a much greater extent by the bunch thinning commonly
practiced in date cultivation - is the effect on time of ripening. The fruits obtained
after the pollination of Deglet Noor with pollen of Fard No. 4 were found to mature
on average 15 days earlier than those resulting from pollination with Mosque pollen.
When applied in California, the procedure of artificial pollination using pollen of a
suitable male parent proved so successful that the variety Deglet Noor, which nor­
mally brings 25-30 % of its fruits to maturity late in the season, could be harvested
early.

SELECTION OF MALE PARTNERS

The observations mentioned above lend support to the belief widely held in many
Old World date-growing countries that some males are better than others for pollinat­

252
 DATE PALM

ing certain varieties. It would thus seem advisable to develop clones from good males
in order to be able to produce a desirable kind of pollen in quantity.
Many date growers are beginning to realize the value of selection among male trees.
In this respect, the following points should be observed (Nixon, 1966) :
1. Flowering season. In order to be eligible, a male tree should preferably flower si­
multaneously with its prospective female partner(s).
2. Size and structure of flower clusters and quantity of pollen. Large inflorescences
producing a great number of flower strands and flowers easily shedding an abundant
quantity of pollen are to be preferred. Male trees differ greatly in the last mentioned
respect. Pollen yields ranging from 267-754 g per year have been reported for male
palms in South Algeria (Monciero, 1950; Wertheimer, 1957).
3. Compatibility relations and metaxenia-inducing potential. These two features can
only be checked by means of systematic test crosses. The male candidate should be
well compatible with the prospective female partner(s) and, in addition, have a favoura­
ble effect on the time of maturity, quality and size of the resulting fruit. The latter
trait is not of such immediate importance, as it is affected to a much greater extent by
the bunch thinning generally practiced in commercial date cultivation (Nixon, 1966).

Improvement by breeding

VARIABILITY OF INITIAL MATERIAL

Date seedlings display a wide range of variability and the degree of variation en­
countered among the date cultivars (varieties) grown in different date-growing coun­
tries is also enormous. These facts are indicative of a wide genotypical diversity as
hardly anything else could have been expected from an obligate cross fertilizer like the
date palm. This large variation led Chevalier (1932) to postulate that, if one original
species of the date palm Phoenix dactylifera L. ever existed, it must have been modified
by hybridization with other species in different parts of the range in which it was origi­
nally to be found: P. sylvestris Roxb. in Pakistan, P. reclinata Jacq. and variant types
in North-East Africa and P. canadensis in North-West Africa.
Experimental evidence (Nixon, 1935) demonstrating that Phoenix species can be
intercrossed may be interpreted as supporting the above concept. Natural hybrids of
P. dactylifera and P. reclinata producing fruit of good quality were reported by Munier
(1951), who also found hybrids between P. dactylifera and P. canariensis in Spain
(1957). The diversity in the genetic composition may be expected to be particularly
great in date varieties from these different regions.

CLONAL SELECTION AND BREEDING

The date palm populations in the Old World originally consisted of trees grown
from seed, as is still the case in primitive agricultural communities today.

253
j. h. m. oudejans

In the course of years, or even of centuries, outstanding palm trees were detected
and vegetatively propagated into a clone. As the palm tree is one of the earliest cultivat­
ed plants to be submitted to a form of man-aided natural selection for some thousands
of years, it is highly likely that not only the numerous local varieties but also the well-
known widely distributed varieties originate from chance seedlings. As the date gro­
wers do their empirical work in isolation, identical local varieties may occur under
two or more names or entirely different varieties under the same name. Sometimes
attempts have been made to select within existing cultivars (Mason, 1927; Nixon,
1950).
The selection of clones mentioned above was merely empirical. Attempts to obtain
improved varieties by means of a deliberate well-planned programme of breeding have
only comparatively recently been initiated at a limited number of research centres. The
possibility of improving the date palm by breeding was suggested by Mason as early as
1908. A clear and concise review of the breeding work completed or in progress has
been given by Nixon and Furr (1965). These authors report that inbreeding experi­
ments within the progeny of the Deglet Noor cultivar carried out at the University of
Arizona yielded no satisfactory results and were discontinued after three generations.
The El Arfiane station in Southern Algeria conducted a date-breeding programme
for about two decades prior to the departure of the French personnel in 1962. The
objectives were to produce a line of Deglet Noor that would breed sufficiently true to
type for propagation of the variety by seed to be feasible and to produce new varieties,
particularly males, that would flower early and others that would yield large quanti­
ties of high-quality pollen.
The U.S.D.A. date-breeding project (Indio, California), which has been in opera­
tion for 20 years has the following aims (Nixon and Furr, 1965):
1. To produce, by backcrossing, males that approach the parent variety in genetic
composition.
2. To use males from advanced backcross progenies for intervarietal crosses in order
to produce new and better fruiting varieties.
3. To select superior seedlings, male or female, that appear in any generation and
that show potential for development of new varieties.
In the date palm the peculiar situation exists that all commercial varieties consist
exclusively of female individuals. Hence, unless special measures are taken, intervariet­
al crossing, a common breeding procedure for combining desirable characteristics,
cannot be effected directly. If it is desired to make certain crosses it is necessary first
to obtain, for each variety, the male counterpart that as closely as possible approaches
the varieties in question as to genetic composition. This is achieved by crossing the
female variety with some available male tree and by backcrossing the male individuals
within the resultant Fj to the female parent. This backcrossing procedure is repeated
for three or more generations, the commercial variety always being used as the re­
current parent, until the backcross progeny bears sufficient resemblance to the recur­
rent parent. Once this stage has been reached, male trees selected from the backcross

254
 DATE PALM

progeny of, for instance, the Medjool variety can be used for intervarietal crosses
with a reasonable expectation that they will transmit mainly Medjool characteristics
to the progeny. The possible transmission by the male of fruit characteristics - a
specific female trait - is still a matter of some uncertainty. More or less reliable in­
ferences in this respect can be made by observing the female palms in the family from
which the male is taken.
The backcrossing programme carried out by the U.S. Date and Citrus Station,
Indio, California, has up to 1965 yielded 37 backcross families of which eight have
advanced to the BC3 stage and two to the BC4 stage. Males from some advanced
backcross progenies have been used for making intervarietal crosses, which will at­
tain the fruitbearing stage in the next few years, when the first selections can then be
made.
The most serious drawback in date-palm-breeding is the time required. The average
cycle from seed to flowering has been found to be 6.5 years. Nixon and Furr(1965)
estimate that at least 30 years may be required to make three backcrosses and to raise
the intervarietal cross to the stage at which the first offshoots from outstanding trees
can be obtained.
An additional drawback is the slow increase rate in vegetative propagation owing
to the limited number of offshoots - ranging from 5-25 - produced by a date palm.
In order to obtain enough offshoots for a small-scale clonal progeny test some addi­
tional generations are required. If it is also taken into account that a date palm does
not reach full production until it is 10-15 years of age, it is easy to understand that
date breeding must inevitably be a long-term project.

Attainments and problems for further work

Considering the results obtained by breeding other cultivated plants, it is justified

to expect that in the date palm further progress may also be achieved although this
plant owing to its dioecious nature evidently presents specific difficulties which, as such,
constitute a challenge to the breeder.
In the continuation of the breeding work increasing emphasis may be placed on
specific characters such as the mode of branching of the fruit stalk, position of the
seed inside the fruit and resistance to diseases, pests and adverse environmental con­
ditions. Among the diseases and pests the Bayoud disease, caused by Fusarium albedin-
is (Malançon), which has destroyed two thirds of the Maroccan date groves deserves
special attention. Individuals resistant to this disease have been found and are being
increased vegetatively (Munier, 1962; Nixon, 1959). Resistance to Graphiola leaf spot,
in localities where the relative humidity is continuously high, has been observed
(Nixon, 1959). By selection and breeding it should be possible to obtain varieties more
suitable to marginal areas where because of high humidity, Graphiola leaf spot is a
serious problem.
Resistance to date mites has also been reported. This will be an important character

255
J. H. M. OUDEJANS

in breeding varieties for desert conditions (Nixon, 1959).

Considerable time and effort will be required to achieve all these objectives.
It has been emphasized from various quarters (Nixon, 1959) that some form of co­
operation between the date-growing countries would definitely promote the progress of
breeding programmes, especially if such co-operation were to be sponsored by some
international agency.

References

AHMED, M. S., 1959. Date palm pollination in the East. First FAO international technical meeting on
date production and processing. Tripoli, Libya, December 1959. Background paper nr. 12.
BEAL, J. M., 1937. Cytological studies in the genus Phoenix. Bot. Gaz., 99: 400-407.
BROWN, T. W., 1924. Date palm in Egypt. Min. of Agr. Egypt. Bull. nr. 43.
CHEVALIER, A., 1932. Les productions végétales du Sahara et de ses confins nord et sud: passé - pré­
sent - avenir. Revue de Botanique appliquée et d'Agriculture Tropicale, 12: 669-924.
CHEVALIER, A., 1952. Recherches sur les Phoenix africaines. Rev. Int. Bot. Appl. et d'Agriculture
Trop., 32 (nr. 3555): 205-233.
COMELLI, A., 1960. Le palmier dattier en Israel. Fruits, 15: 223-231.
FONTENEY, U. DE and FONTENEY, V. J. DE, 1960. Date growing in Australia. J. Australian Inst, of Agr.
Sei., 26: 246-257.
KAUL, K. N., 1951. Some interesting features of the distribution of palms in relation to their origin.
The Indian J. of Gen. and Plant Breeding, 11: 108-110.
LEFÈVRE, F., 1962. Multiplication du palmier-dattier à la station de Kankossa-Maurétanie. Fruits, 17:
129-131.
MASON, S. C., 1908. Date growing in Southern California. Official Report of the 34th Fruit Growers'
Convention of the State of California: 170-178.
MASON, S. C., 1927. Date culture in Egypt and the Sudan. U.S. Dept. of Agr. Bull., nr. 1457.
MONCIERO, A., 1950. La fécondation mécanique du palmier dattier. Bull, d'information, 4 (38-1):
81-88, Office Tunisien de Standardisation.
MUNIER, P., 1951. Contribution à la mise en valeur du Sahara Soudanais Français. Fruits d'Outre
Mer, 6: 104-108.
MUNIER, P., 1953. Sur l'origine du palmier-dattier. Fruits d'Outre Mer, 8: 47-52.
MUNIER, P., 1955. Le palmier-dattier en Maurétanie. Inst. des Fruits et Agrumes coloniaux, Annales
1955, nr. 12.
MUNIER, P., 1957. Le palmier-dattier en Espagne continentale. Fruits d'Outre Mer, 12: 269-279.
MUNIER, P., 1962. Sur la présence du faux dattier, Phoenix atiantica Chev., en Adrar maurétanien.
Fruits d'Outre Mer, 17: 208-210.
MUNIR, A. and NIAZ, A., 1961. Hastening date ripening by the use of selective pollen. Proc. 13th
Pakistan Sei. Conf. Dacca 1961, Sect. Ill: 11-19.
NIXON, R. W., 1928. The direct effect of pollen on the fruit of the date palm. J. Agr. Res., 36: 97-128.
NIXON, R. W., 1934. Metaxenia in dates. Am. Soc. Hort. Sei. Proc., 32: 221-226.
NIXON, R. W., 1935. Metaxenia and interspecific pollinations in Phoenix. Proc. Am. Soc. Hort. Sei.
33: 21-26.
NIXON, R. W., 1950. Imported varieties of dates in the United States. U.S. Dept. of Agr., circ. nr. 834.
NIXON, R. W., 1959. Pollination, breeding and selection of date palms. First FAO international tech­
nical meeting on date production and processing. Tripoli, Libya, December 1959. Background
paper nr. 11.
NIXON, R. W., 1966. Growing dates in the United States. Agr. Inform. Bull. U.S. Dept. Agr., 207:
1-50. Revised edition 1966.

256
 date palm

NIXON, R. W. and FURR, J. R., 1965. Problems and progress in date breeding. Date Growers' Institute
Report, 42: 2-5.
PÉREAU-LEROY, P., 1957. Fécondation du palmier-dattier. Fruits d'Outre Mer, 12: 101-105.
PIQUER, G. J., 1959. Culture du palmier-dattier en Egypte. First FAO international technical meeting
on date production and processing. Tripoli, Libya, December 1959. Background paper nr. 5.
SWINGLE, W. T., 1928. Metaxenia in the date palm. Jour. Heredity, 19: 257-268.
TOTHILL, 1948. Agriculture in the Sudan: 368-386. Oxford University Press, London.
WERTHEIMER, M., 1957. La pollination du palmier-dattier. Fruits d'Outre Mer, 12: 305-313.

257
FIG

Ficus carica L.

WILLIAM B. STOREY AND IRA J. CONDIT

University of California, Citrus Research Center, Riverside, California

Systematics

TAXONOMY

The systematic botany of the common fig, Ficus carica L., has been treated by
various authors including Eisen (1901), Condit (1957) and others. It is one of the
thousand or more species of Ficus widely distributed in tropical and subtropical
countries. Related genera which include species having edible fruits are Artocarpus,
Cudrania, and Morus, all in the family Moraceae. Ficus carica belongs to the subgenus
Ficus as recently classified by Corner (1965). Earlier botanists placed it in Eusyce. The
fruit or receptacle borne by figs is known botanically as a syconium, a name derived
from the Greek sykon (fig. 3). It may be defined as a form of inflorescence in which the
flowers are borne on the inner wall of a hollow receptacle.

ORIGIN AND DISTRIBUTION

The common fig is believed to have had its origin in southeastern Asia from which
it became gradually distributed to Caria in Asia Minor (hence the specific name
'carica'), to countries bordering the Mediterranean Sea, and even into England. Fol­
lowing discovery of the New World, fig trees became established in both North and
South America. In 1769 the Franciscan fathers planted trees first at San Diego and
later at other Mission stations now in the State of California. There are now about
20,000 acres of orchard trees producing an average of one ton of dried fruit per acre.
Thousands of trees are grown in dooryards in California, in southern and southeastern
United States, while in Texas some orchards produce fresh figs for preserves.

Biology of reproduction

The species, Ficus carica, is characterized by trees bearing fruits of two distinct
sex forms: (a) fruits with long-styled pistillate flowers only and (b) monoecious, with
both short-styled pistillate and staminate flowers in the same fruit (fig. 1). The pistillate
form is represented by several hundred named horticultural variétés as described by

259
WILLIAM B. STOREY AND IRA J. CONDIT

Fig. 1 Pistillate flowers of the fig; short-styled (left) and long-styled (right).

Condit (1955). The monoecious or caprifig type is represented by only a few dozen
named varieties, a small fraction of these being grown commercially as good pollen
producers.
The pistillate forms have been classified into three groups or types: the Common,
the San Pedro, the Smyrna. The characters of these three types may be designated as
follows:
Common Type. Trees of this type may or may not produce a first or breba crop
borne on wood of the previous season's growth. They do produce a more or less
profuse second crop or even a third crop late in the season in axils of leaves of cur­
rent growth. Both crops are produced by parthenocarpy.
San Pedro Type. Trees produce a first crop which is parthenocarpic. Figs of the
second crop drop prematurely unless the flowers are stimulated by pollination and
fertilization.
Smyrna Type. Trees generally produce a main or summer crop only although a
few brebas may reach maturity by parthenocarpy. Figs of the main crop are non-
parthenocarpic; they fail to set and mature unless flowers are pollinated and fertilized.
Caprifig Type
The trees of the second sex form, the monoecious or caprifig type produce at least
three crops annually, the figs of each crop being inhabited by the fig insect, Blastophaga
psenes. The first, the mamme crop, is initiated in the late fall, remains on the tree
during the winter with the insect in the larval stage, and matures in the spring. The
second, the profichi, is the most prolific crop, appearing on wood of the previous
season's growth. It matures in early summer with the staminate flowers producing
pollen profusely. In nature the pollen becomes attached to bodies of the fig insects
as they emerge and is thus carried to receptacles of either the caprifig where the
insects oviposit in short-styled flowers or to other figs having long-styled flowers
unsuitable to oviposition. This process of transfer of pollen is commonly known as
caprification which results in the production of fertile seeds or achenes. These so-cal­
led seeds have recently been classified as drupelets by Crane and Baker (1953).

260
 fig

Breeding

BREEDING HISTORY

The selection of natural seedlings long ago accounts for the origin of fig varieties the
best of which have been propagated asexually and maintained for centuries under
cultivation. Actual fig breeding has been carried on in various countries and by sev­
eral individuals over a long period.
There follows a summary of past attempts to develop better varieties by hybridiza­
tion and selection. (See Condit (1928) for details and list of references). In Algeria
Trabut (1922) obtained fertile seeds by crossing Ficus palmata Forsk. and F. pseudo-
carica Miq. with pollen of F. carica. Numerous seedlings were grown but no subse­
quent report on their behavior has been noted. In Italy Pellicano (1907) referred
to the Dottato as a decadent variety and both he and Guglielmi (1908) suggested the
development of new seedlings to replace it. In his book 'Fig culture', Van Velzer (1909)
stated that in Texas fig seedlings generally produce worthless wild figs. In Georgia
W. B. Hunt (1911, 1912) pollinated figs by hand and grew some seedlings vigorous in
growth but only one was selected for propagation. Dr. Gustav Eisen reported in 1901 :
"no variety of the common fig has been originated in California and any statements
of valuable varieties having been raised from seeds in England or elsewhere in Europe
should be accepted with doubt."
Several reports are available of attempts to produce good seedlings from seeds of
imported Smyrna-type figs. One of these near Loomis, California, involved 139 seed­
lings, 74 of the Caprifig and 65 of the Smyrna type. Several of the seedlings were given
variety names and distributed for trial by the United States Department of Agricul­
ture. Of the seedlings and of numerous others grown by G. P. Rixford (1918,1926) and
W. T. Swingle (1908,1912) for the Department, none became commercially important.
Another project in fig breeding was carried on at the Yuma Station of the Department,
involving 1600 seedlings from crosses of the Smyrna type. After eight years' work it
was reported by Noble (1922) that 384 trees or 24% of the seedlings produced fruit
but the quality was inferior to that of the parent varieties. Luther Burbank (1914)
stated in his autobiography: "I have grown seedlings in abundance, but ninety-nine
out of one hundred produce worthless fruit. You plant seeds of the white fig and
you are quite as likely to get black or brown figs as white ones."
In California the fig growers themselves urged the University to carry on work
in fig breeding as attested by a resolution passed at a Fig Institute meeting held
in 1924 urging development of varieties resistant to splitting and souring, and of some
kinds better adapted to the production of fruits for drying, for home use, or for the
fresh fig market. Such a project had already been initiated in 1922 when Mission and
Kadota fruits were hand pollinated. Several hundred of the seedlings were first grown
at Davis but in 1928 cuttings from them were transferred to the Citrus Research Cen­
ter, Riverside, where records of fruit production were obtained during ensuing seasons.

261
WILLIAM B. STOREY AND IRA J. CONDIT

This project was undoubtedly the first ever initiated on a large scale to develop im­
proved varieties by assembling a wide range of variety material and by applying the
valid practices of plant breeding.

BREEDING METHODS

Methods in fig breeding are simple indeed. Few fruits if any can be equally produc­
tive of fertile seeds by hand pollination. Pollen is easily secured from the male parent,
a caprifig with desirable characters, by splitting the fruits lengthwise and allowing
the two halves to dry overnight when the pollen can be sifted out. Pollination is
accomplished by means of a glass tube drawn to a point with a rubber nipple attached
to the opposite end. An opening is made in the side of the fig by a glass rod or nail
so that air can escape as the pollen is puffed through the eye. Figs pollinated in June
mature in August when the fertile seeds can be washed from the pulp. The testing of
seedlings may be expedited by the following procedure: plant seeds in October,
transplant small seedlings to greenhouse bed at 10 cm spacing, allow to grow during
winter, treat each seedling as budstick from which two buds are cut and each inserted
into sucker wood of mother trees prepared in advance by heavy pruning. Such a
practice was advocated by John Wright (1894) in 1891 when he wrote: "To ensure
early fruiting of seedlings they may be grafted or inarched on old trees when they may
fruit in the third year. On their own roots the seedling may be much slower in fruit
production."
During the past three decades over 25,000 seedlings representing 280 crosses have
been grown and fruited at Riverside. Seedling characters have been noted as to crop
(first and second), size, shape, color (exterior and interior), size of apical orifice,
parthenocarpy or non-parthenocarpy. Seedlings which showed promising characters
were tested in commercial fig districts of the San Joaquin Valley. One seedling, a cross
between Yerdone (White Adriatic) and Caprifig No. 72-80, and named Conadria
(Condit, 1956b), has proven to be more resistant to spoilage than Verdone. According
to the latest estimates there have been planted 660 acres of the Conadria mainly for
production of dried figs. As mentioned above, trees of this and some other seedlings
are remarkably vigorous as well as precocious. For dooryard planting in southern
California the Conadria and some as yet unnamed seedlings far excel certain varieties
hitherto grown.
One of the significant results of the above fig breeding project is the development
of several parthenocarpic caprifigs. They resemble common figs in size, shape, internal
and external color, productivity, and, to some extent, palatability. At the same time,
they produce abundant viable pollen. Such caprifigs are proving valuable for breeding
purposes as some of the seedlings already show characters worthy of commercial tests.

262
 FIG

Fig. 2 Interspecific hybrid. Left: leaves of F. pumila; center: leaves of pollen parent; right: leaves of
hybrid.

GENETICS

The genetics of parthenocarpy is shown by the following four crosses :

1. nonparthenocarpic x nonparthenocarpicall nonparthenocarpic
2. nonparthenocarpic X parthenocarpic -> 1 nonparthenocarpic : 1 parthenocarpic
3. parthenocarpic x nonparthenocarpic -» all nonparthenocarpic
4. parthenocarpic X parthenocarpic -> 1 nonparthenocarpic : 1 parthenocarpic
The allele for parthenocarpy is dominant. Nonparthenocarpic individuals, perforce,
are homozygous for the recessive allele. The parthenocarpic parents used in the cros­
ses above were heterozygous. One might expect, therefore, that cross (3) would result
in a progeny ratio of 1:1, and cross (4) in a progeny ratio of 3:1. These ratios are
not realized, however, because the allele for parthenocarpy is lethal to female gametes
containing it. Consequently, only those ovules containing the nonparthenocarpic
allele can function in syngamy in parthenocarpic figs.
The above genetic ratios were determined in 1965 by Wadie Fadel Saleeb, at the
University of California, and reported in his dissertation for a Ph. D. degree. In it he
uses the terms persistent and nonpersistent in preference to parthenocarpy and non-
parthenocarpy on the premise that the syconium is a vegetative structure; i.e. a
compound, complex peduncle, and that, in many cases, it does mature and ripen despite
the fact that the flowers within may be completely abortive.
The genetic segregations for sex occur in simple Mendelian ratios as follows :
fig X caprifig 1 fig : 1 caprifig
caprifig X caprifig -> 1 fig : 3 caprifigs

263
WILLIAM B. STOREY AND IRA J. CONDIT

Fig. 3 Left: syconium of F. pumila, the stamens undeveloped; right: syconium of hybrid with pro­
fuse pollen from stamens.

The fig is homogametic for sex, the caprifig heterogametic with the caprifig cha­
racters dominant. The only caprifigs used so far, except the parthenocarpic ones men­
tioned above, have been varieties grown commercially as pollen sources. All are
heterozygous for sex. Theoretically, homozygous caprifigs exist among seedlings of
caprifigs, but this has yet to be demonstrated.

INTERSPECIFIC HYBRIDS

The production of interspecific hybrids in Ficus or related genera offers a fertile

field for research. The first such hybrid was reported by Condit (1950) as a cross of
Ficus pumila L., an evergreen vine fig with pollen of F. carica, a deciduous tree fig.
Some seedlings were deciduous, some evergreen; most were vine like; many produced
fruits with which backcrosses were made (figures 2 and 3). This hybrid was reproduced
in 1965 by Storey and seedlings are again being grown.
Ficus carica hybridizes readily with F. palmata and F. pseudo-carica. However,
only one seedling, the Brawley caprifig, has been deemed worthy of a variety name. In
southern Russia Arendt (1959a, 1959b and 1964) has recently reported the results
of hybridization between F. afghanistanica Warb., F. carica, Morus alba, and other
members of Moraceae. He obtained seeds from which fruiting plants were obtained.
They possessed the normal chromosome complement (2n = 26) and were highly
fertile.

264
 FIG

Attainments and problems for future research

ATTAINMENTS

The following is an enumeration of the attainments in research relating to fig breed­

ing and genetics since the inception of projects in 1922 by the Experiment Station of
the University of California carried on mostly at Riverside.
1. Publications by Condit other than those cited above: Morphology of flowers in
Ficus carica (1932); Chromosome number determinations (1928); Characters useful
in identification of varieties (1941); Bibliography of the fig (1956a); Monograph of fig
varieties (1955); Description of a new variety, Conadria, probably the first and so far
the only commercially valuable seedling to result from planned fig breeding (1956b).
2. Development of parthenocarpic caprifig clones from early crosses. These have
characteristics of edible figs and produce viable pollen for use in further breeding.
3. Successful hybridization of three closely related species as noted above.
4. Genetics of sex determination by W. B. Storey, the results not yet published.
5. Determination of the genetics of parthenocarpy vs. nonparthenocarpy.
6. Induction of polyploidy both in germinating fig seeds and in vegetative buds of
several fig varieties with the use of colchicine. The seedlings have not yet fruited.
7. The establishment and maintenance of a collection representing over one hun­
dred species of Ficus with these objectives : hardiness and adaptability to California
environment; possible hybridization with F. carica ; rootstock affinity with F. carica
for resistance to such pests as nematodes, an objective already attained with three
species : F. gnaphalocarpa, A. Rich., F. racemosa L. and particularly with F. cocculifolia
Baker.

PROBLEMS FOR FUTURE RESEARCH

The field for future research in fig breeding is wide open. Some of the objectives in
such breeding may be outlined.
1. Selection of seedlings with light-green skin color and amber pulp, characters
desirable for both canning and drying.
2. Seedlings with small ostiole or eye (fig. 4) more or less closed by scales which
prevent entrance by species of Drosophila, Carpophilus, or thrips.
3. Seedless fruits or those without hard drupelets but still having good flavor and
quality.
4. Seedlings productive of fruits with skin which does not readily check or so tender
as to become bruised in handling.
5. A Smyrna-type fig, or one developing by parthenocarpy, equal in quality to the
Calimyrna (Sari Lop) but with a small eye, the fruit resistant to splitting and less sus­
ceptible to internal spoilage. Much of the quality of the Calimyrna is due to the nutlike
flavor of the embryo in the fertile seeds. Some Common Figs tend to develop endo-

265
WILLIAM B. STOREY AND IRA J. CONDIT

Fig. 4 Fruits of six varieties of figs showing typical eye or ostiole. Top left: small eye more or less
closed by scales ; bottom left : large open eye easily penetrated by insects.

sperm parthenogenetically as shown by Condit (1932). This tendency might be en­

hanced by the use of growth regulators as pointed out in a series of papers by Crane
(1952, 1953 and 1959) and associates.
6. Continuation of studies on induced polyploidy on seedlings and varieties with
the use of colchicine.
7. Continuation of attempts to produce interspecific hybrids such as already produc­
ed between Ficuspumila and F. carica.
8. Selection and testing of seedlings and hybrids which show resistance or immun­
ity to root troubles such as that caused by nematode infestation.

References

ARENDT, N. K., 1959a. Result of pollination of Ficus afghanistanica. Ialta Gosud. Nikitskii Botan.
Trudy, 29: 197-201.
ARENDT, N. K., 1959b. New fig varieties. Ialta Gosud. Nikitskii Botan. Trudy, 29: 235-250.
ARENDT, N. K., 1964. Fig breeding in the Crimea. Ialta Gosud. Nikitskii Botan. Trudy, 37: 190-214.
BURBANK, L., 1914. Luther Burbank. Vol. 4: 297. Burbank Press. New York.
CONDIT, IRA J., 1928. Chromosome number and morphology in seven species of Ficus. Univ. Califor­
nia Publ. Botany, 11: 233-244.
CONDIT, IRA J., 1928. Fig breeding. Jour. Heredity, 19:417-424.
CONDIT, IRA J., 1932. The structure and development of flowers in Ficus carica. Hilgardia, 6:443-481.
CONDIT, IRA J., 1941. Fig characters useful in the identification of varieties. Hilgardia, 14: 1-68.
CONDIT, IRA J., 1950. An interspecific hybrid in Ficus. Jour. Heredity, 41: 165-168.
CONDIT, IRA J., 1955. Fig varieties: a monograph. Hilgardia, 23: 323-538.

266
 fig

CoNDiT, IRA J., 1956a. A bibliography of the fig. HiJgardia, 25: 1-662.
CONDIT, IRA J., 1956b. Promising new seedling fig, Conadria. California Agr. 10 (6): 4.
CONDIT, IRA J., 1957. The fig. 222 pp. Chronica Botanica Co. Waltham, Mass.
CORNER, E. J. H., 1965. Check-list of Ficus in Asia and Australasia. The Garden's Bull. Singapore,
21: 1-186.
CRANE, J. C., 1952. Ovary-wall development as influenced by growth regulators inducing partheno-
carpy in the Calimyrna fig. Bot. Gazette 114: 102-107.
CRANE, J. C. and BAKER, R. E., 1953. Growth comparisons of the fruits and fruitlets of figs and straw­
berries. Amer. Soc. Hort. Sei. Proc. 62: 257-260.
CRANE, J. C., BRADLEY, MURIEL and LUCKWILL, L. C., 1959. Auxins in parthenocarpic and non-
parthenocarpic figs. Jour. Hort. Science 34: 142-153. (Includes list of other references.)
EISEN, G., 1901. The fig. United States Dept. Agric. Div. Pom. Bull. 9: 317 pp.
GUGLIELMI, G., 1908. Coltivazione industriale del fico nel Leccese. Boll. Arbor. Italiana Anno, 4:155.
HUNT, B. W., 1911, 1912. Fig breeding. Georgia Univ. Bull., 11: 146-148; 12: 106-110.
MASSEY, W. F., 1903. North Carolina Agr. Exp. Sta. Bull. 184: 123.
NESTERENKO, G. A., 1948. Sad i Ogorod 1948: 20-25. Abs. in Plant Breeding Abs., 19: 205.
NOBLE, E. G., 1922. Figs. United States Dept. Agr. Circ. 221: 28.
PELLICANO, A., 1907. II fico nel circondario di Gerace. Boll. Arbor. Italiana, J.- 148.
RIXFORD, G. P., 1918. Smyrna fig culture. United States Dept. Agr. Bull. 732: 36-39.
RIXFORD, G. P., 1926. Work with seedling figs. Pacific Rural Press 112: 696.
SALEEB, WADIE F., 1965. Genetics and syconium persistence in Ficus carica. 79 pp. Ph. D. dissertation.
Univ. of California, Riverside.
SWINGLE, W. T., 1908. Breeding new and superior figs. California State Fr. Grower's Conv. Report
34: 187.
SWINGLE, W. T., 1912. Cooperative distribution of new varieties of figs. United States Dept. Agr. Bur.
PI. Ind. No. 438: 1-6.
TRABUT, L., 1922. Sur les origines du figuier. Revue de Bot. Appl., 11: 395-396.
VAN VELZER, A. C., 1909. Fig culture. 218 pp. Houston, Texas.
WRIGHT, J., 1894. The fruit grower's guide. Vol. 2: 170-200. 6 Vols. Virtue & Co, London.

267
KAPOK TREE

Ceiba pentandra Gaertn.

A. C. ZEVEN

Institute of Plant Breeding, Wageningen, The Netherlands

Introduction

Before the war, the kapok tree was an important commercial crop, because its fibre
was used extensively in life-jackets and life-belts, in clothing for aviation, in linings
for mackintoshes, as a mattress and cushion-filling material and in sound insulation
for aircraft. In recent years, however, synthetic substitutes such as glass fibre, foam
rubber and man-made fibres have become available to take over many of these uses
(Industrial Fibres, 1965) and, in consequence, the international trade in this fibre
is on the decline. Where the kapok tree is grown, the fibre is much sought after for
mattresses, padding and stuffing.
It is difficult to estimate the total production of kapok, as only export figures are
available. In 1965, Thailand exported 50% (about 160 tons) of the total world trade,
followed by Cambodia (30 %), but this figure includes the kapok produce of Laos and
Vietnam. Indonesia, which before World War II accounted for 90% of the world
trade, had only 6% in 1965. Other countries in South-East Asia and tropical Africa
export only small quantities.

THE FIBRE

The cells of the inner epidermis of the epicarp form the fibre. It lies in small cush­
ions against each seed. Not all the epidermis cells grow into a fibre. Sometimes some
neighbouring cells grow together into a 'double hair' and a 'group of hairs'. The fibre
is about 1 to 2 cm long, with a diameter of 10 to 30 a (with a mean of about 20 jj.).
The air-filled lumen is broad and the wall rather thin. It breaks easily, which to­
gether with the smoothness of the outer-surface, is a disadvantage. Owing to its
smoothness, it cannot be spun.

Taxonomy, geographical distribution and origin

TAXONOMY

The kapok tree belongs to the Adansoniae of the Bombacaceae. In 1524, Oviedo

269
A. C. ZEVEN

mentioned its Caribbean name Ceyba or Seyba, and Miller used this vernacular name
in 1739 as a generic name. Linnaeus, when studying indica-material, gave it the name
Bombax pentandrum, the specific name being based on an illustration published in
Rheede tot Draakenstein's Horti Malabarici (Voorhoeve, 1965). In 1791 Gaertner
split up the genus Bombax into Bombax and Ceiba, which accounts for the present
botanical name Ceiba pentandra (L.) Gaertn.
Owing to its polyploid nature the kapok tree is very variable and this has made se­
veral taxonomists believe that C. pentandra was a group of species. Its savannah-type
habit has also contributed to this confusion. However, such subdivision is not valid,
because the supposed species hybridize freely and the characteristics used to distin­
guish the species described follow a simple or polygenic pattern of inheritance (cf.
p. 281). Thus, the species were, in fact, the ecotypes or local populations with the
most frequent characteristics described.
The main subdivisions of C. pentandra originally made by P. de Candolle on a
species level and reduced in 1924 by Bakhuizen van den Brink (1924) to the level of
botanical varieties are: var. caribaea (D.C.) Bakh. and var. indica (D.C.) Bakh.
For the African kapok tree, Ulbricht (1913) founded the botanical varieties var.
clausa, the type with indéhiscent fruits, and var. dehiscens.
Kapok is collected from other Ceiba and some Bombax species.

Chromosome number

The chromosome number ranges from 2n = 72 to 88 (Heyn, 1938; Tjio, 1948;

Baker, 1965, and our own investigations). Our investigations showed that the chromo­
some number of the progeny of a selfing varied, the mother plant being of the indica
type. It is very likely that this species is a polyploid and it seems that 2n = 72 is the
lowest possible number of chromosomes for a plant to grow.

GEOGRAPHICAL DISTRIBUTION

C. pentandra var. caribaea occurs wild and semi-wild in America (Southern Mexico,
the Caribbean Islands, Central America and tropical South America north of the
southern limit of the Amazon basin) and in Africa (the west coast from Senegal to
Central Africa). In Java the 'Reuzenrandoe' (giant kapok) bears some characteristics
typical of the caribaea variety. C. pentandra var. indica - the cultivated form - is
found in Ceylon, Thailand, West Pakistan, South India, Cambodia, South Vietnam,
Malaysia, Philippines, Indonesia and some islands of the Western Pacific (Chevalier,
1931, 1937; Toxopeus, 1950; Van der Pijl, 1956). In Africa some trees bear indica
characteristics.
Temperature and rainfall are limiting the spread of the kapok tree. Night tem­
peratures of below 17 °C retard the germination of the pollen grain and the growth of
the pollen tube and, thus, the flower drops before the pollen tube has reached the

270
 KAPOK TREE

Fig. 1 Kapok trees with a heavy fruit production. (Photograph T. G. Garrido)

ovary (Toxopeus, 1939). Rainfall should be about 1500 mm per year with a four-
month period of 150 to 350 mm distributed over 10 to 25 days. In some drier areas the
demand for water may be met by ground water supplies. Although rainfall is inade­
quate for the kapok tree in the Mekong delta, it is extensively grown there on river
banks. In Celebes it also grows on a lake bank which is annually flooded (Toxopeus,
1939,1943b, 1950).

ORIGIN

Toxopeus (1942a, 1948a, 1950) believed that the kapok tree originated in an area
which was later divided by the Atlantic Ocean, so that this species is a native of both
America and Africa. He based his conclusion mainly on the great variability of this
plant and on the high frequency of dominantly inherited characteristics in these two
continents.

271
A. C. ZEVEN

Bakhuizen van den Brink (1933) and Chevalier (1949) upheld the American origin
of the kapok. In pre-Columbian times seeds might have been transported by sea or air
currents to Africa. Chevalier believed that the present great variability of the kapok
tree in Africa would have arisen as a result of various introductions in historical times.
The distribution of the Ceiba species also points to an American origin, because C.
pentandra is the only Ceiba species with an extra-American distribution. If it is true that
the kapok tree arose as a result of the polyploidization of one or more Ceiba or related
species (cf. p. 270), then this tree can only have arisen in that area where these Ceiba
or related species were formerly or are at present found. This supposition would also
point to an American origin.
Callen (1965) refers to the pre-Colombian existence of the 'Ceiba, silk-cotton tree'.
However, this is not the kapok tree but the Ceiba parvifolia Rose mentioned by Smith
(1965).
Bakhuizen van den Brink (1933), Toxopeus (1941, 1943a, 1948a, 1950) and Merrill
(1954) all believe that the kapok tree was introduced into India from Africa by the
Sabean Lane and from India into South West Asia by Malaya, perhaps between
500 B.C. and 500 A.D. If so, the kapok with some seeds must first have been transport­
ed to North Eastern Africa. This might have been by the trade route through the Su­
dan zone. The slight variability of this crop in Asia indicates that only a few introduc­
ed seeds grew into productive trees in that region. During its cultivation, selection for
desired characteristics must have been applied, which has further reduced its variabi­
lity. The kapok tree was already being cultivated in Asia long before 976 A.D., be­
cause relief carvings found in Java show the indica-type (Steinmann, 1934). Toxopeus
(1941) referred to another carving of 850 A.D. which probably also shows the kapok
tree.
Toxopeus (1948b) suggested that, as the kapok population of former Indo-China
was less variable than those of Java or Sumatra, Indo-China must have received its
kapok from either of these two islands or from both, or possibly, even from Thailand/
Burma. The kapok tree of Celebes most likely came from Java. After 1900, some new
material was introduced.
For instance, material from Africa and America was brought to Java for experiment­
al purposes; Togo material was introduced into the Mekong delta and from there to
Madagascar (Toxopeus, 1939; Montagnac, 1952); and indica material was also
brought to East Africa (Cantzler, 1942).

Botanical notes

TREE HABIT

According to tree habit the kapok tree can be classified into two groups, viz. the
wild and semi-wild caribaea and the cultivated indica. The caribaea group can be sub­
divided into :

272
 KAPOK TREE

caribaea-forest type: a high (30-35 m) unbranched trunk, with a high crown and
often with big buttresses. Some spineless types with indéhiscent fruit and white kapok
have been selected for cultivation. They are named 'Togo'.
caribaea-savannah type: a tree with a short (up to 10 m) unbranched trunk and a
very broadly spreading crown. It often has big buttresses which are particularly con­
spicuous because of the short trunk.
Both types occur in Africa and America and may display the different characteris­
tics discussed on page 281, but the frequency of these characteristics may vary from
one local population to another. For instance, the savannah type in Togo and Liberia
is spineless (Ulbricht, 1913; Voorhoeve, 1965), but in Gabon it is spiny (Raponda-
Walker and Sillans, 1961). In general, the tree produces dehiscent fruit with a grey
kapok.
In the savannah area, where this type is grown in market places, it is propagated by
cuttings (Dalziel, 1937). Such asexual propagation undoubtedly leads to the incidence
of mixtures of clones or even a single clone in a certain area. Furthermore, it should
be investigated whether the savannah tree habit is a result of cuttings obtained from
plagiotropic branches, hence resulting in trees with a 'horizontal' growth habit.
The indica type is a tree with a total height up to 25 m and is not as large as the
caribaea type. It does not have big buttresses and it produces indéhiscent fruits con­
taining a white kapok. Spiny and spineless types are found. The trees are cultivated
around villages, on farmer's plots or sometimes on commercial plantations. The indica
type can be subdivided according to habit into subtypes, viz. 'pagoda' and ianang'.
The 'pagoda' habit is characterized by a tree shedding its lower branches, thus having
an unbranched stem with a crown, whereas the 'lanang' tree has a crown touching the
ground because it retains its lower branches.
The indica type progressively produces new storeys of three branches until the main
stem branches.

SEED, ROOT AND FLOWER

The brownish-black seed has a diameter of almost 0.5 cm, the germination is epi-
geic and the roots of the seedling are rather fragile (Francken and Schlieben, 1942). The
juvenile leaves of the indica types are green, while those of the caribaea types are red­
dish or green.
The root system of the 'Java' kapok is quite horizontal, reaching as far as 10 m
(Toxopeus, 1950). The caribaea types have a much finer branched root system and
they have a wider spread. Both these characteristics make these types more resistant
to drought.
The flowers are grouped in umbel-like inflorescences in the leaf axils of one-year-old
shoots on the ends of leafless branches. Toxopeus (1943b, 1954) estimated that there
were up to 60,000 flowers per average indica tree in Java. Most inflorescences have
about two to eight flowers (Grist, 1923). The corolla has five petals with hairs on the

273
A. C. ZEVEN

outside. The inside is bright yellow with a greasy shine. The five stamens and petals
are fused at the base. The filaments are about 2.5 cm long, while the style varies from
2.5 (homostyly) to 3.5 cm in length (heterostyly). The position of the stigma is eccen­
tric, especially in the case of the heterostylic flowers. The style has a thickened base
which is connected to the ovary by a thin strand (Toxopeus, 1948b; Jaeger, 1954).

TOPOPHYSIS

Only parts of the main stem can be used for cuttings and buds, because material
taken from branches is inclined to grow horizontally (plagiotropy). This is not al­
ways so, because buds taken from branches sometimes display a vertical growth
(Haigh, 1941). Cuttings and scions originating from buds taken from the main stem
always grow vertically (orthotropy).

ASEXUAL PROPAGATION

Asexual propagation, especially budgrafting, is necessary if material does not

breed true or if a reduction in the time of the vegetative phase is required. It may also
be applied when multicauly is desired. Patch-budding and patch-shield-budding are
the best methods of budgrafting.
The use of cuttings is only widespread in the savannah regions of West Africa, where
they are planted in places such as markets, etc. (Dalziel, 1937).
Grist (1923), Mercado (1934), Toxopeus (1936, 1950), Den Doop (1937), Huitema
(1937), Pacumbaba (1940), Haight (1941) and Montagnac (1952) have given detailed
information on the different methods of budgrafting, together with their advantages
and disadvantages.

ROOTSTOCK

Only a few data have been collected on the relation between scion and rootstock.
Scion dominance was found in the case of indica 'lanang' and indica 'Java', i.e. the
'Java' rootstock produced heavier roots than those of the 'lanang' (Bolt and Bolt,
1933). The rootstock dominance was observed of indica 'Java' over caribaea 'Su­
riname'. The late and irregular bearing characteristic of the 'Suriname' was also found
in the 'Java' scion (Huitema, 1937, 1938; Toxopeus, 1950). On the other hand, no
such dominance was found in the case of the rootstock indica/caribaea 'Reuzenran-
doe', which is also a late and irregular yielder (Toxopeus, 1950).
'Suriname' produced a heavier root system than did 'Java' and it was thought that
'Suriname' was a good rootstock type for poor soils or for dry areas. The yield of
indica 'Japara' on 'Japara' trees was found to be higher than that of 'Japara' on
'Suriname' (Montagnac, 1952). Related species were also tried, but without success
(Montagnac, 1952). The only practical application for the grafting of a scion on to

274
 KAPOK TREE

a rootstock has been to obtain a reduction in the vegetative period of the former, but
rootstocks may in future be used as the bearers of highly yielding material.

Floral biology and fruit development

FLORAL BIOLOGY

Flowering takes place on leafless branches at the beginning of the dry season. Flo­
wer buds open about 15-20 minutes after sunset. Within ten minutes the five petals
of the hanging flower have parted, turning their upper ends upwards (Toxopeus,
1948b, 1950; Jaeger, 1954; Baker and Harris, 1959). The straightening of the filaments
and style takes more time. The inside of the calyx secretes a nectar, which runs off the
corolla. The next morning and during the following day the petals show the first
signs of wilting, they slowly droop and shrivel, and lose their greasy shine. They final­
ly become yellowish or dirty brown. The same holds true for the filaments and style.
At the end of the day wilting is complete and the flower, with stamens and style, drops
down.
Reports differ on the ripening of the anthers and stigma. Toxopeus (1950) found the
'Java' kapok tree to be slightly protandrous, but Jaeger (1954) observed in Africa that,
before the opening of the flower, the anthers had already dehisced and the stigma had
been covered with pollen.
Garrido (1955) reported that the stigma is already receptive on the morning of the
day of opening. Chevalier's (1937) statement that the stigma had already emerged
before the flower had opened has not been confirmed by Toxopeus and Jaeger.
Indica trees start flowering in the fourth year after sowing. Prior to flowering, they
shed all their leaves. The flowering of caribaea trees, on the other hand, may be delay­
ed until their eleventh year. Leaf-shedding may in this case apply to the entire tree or
be restricted to some branches that shed their leaves either in sequence or all at once.
According to Dalziel (1937) the savannah type flowers more regularly than the forest
type.

POLLEN AND POLLINATION

The yellow pollen grains are rather big (diameter 70 to 80fi, Jaeger, 1954) and can
be stored at room temperature for several days, although the germination capacity
decreases.
Van der Pijl (1936, 1936-1937) pointed out that the kapok flower is a typical bat-
flower. Chiropterophily (bat-pollination) of single or small groups of kapok trees is
reported in certain parts of Java, Indonesia, in both 'Java' and 'Suriname' trees
(Van der Pijl, 1935), in Madagascar (Montagnac and Ramena, 1961), in several coun­
tries in West Africa (Jaeger, 1954; Baker and Harris, 1959) and in South America
(Baker and Harris, 1959; De Carvalho, 1960).

275
A. C. ZEVEN

Nectar-lapping bats do not occur, however, in the whole area covered by the kapok
tree, e.g. not in certain parts of Java (Toxopeus, 1936) and not in some Western Pacific
islands (Van der Pijl, 1956). One cause of their absence from these regions might
be that during the part of the year that the kapok tree does not flower no or insufficient
food is available (Van der Pijl, 1936-1937). Bat-pollination does not take place in large
plantations because the bats find it difficult to enter the crowns of the trees.
Bats visit the flowering trees in the first part of the night, promoting mainly self-
pollination (autogamy and geitonogamy), but where simultaneously flowering trees
are growing near to one another, although they may be even as much as 90 m apart
(Baker and Harris, 1959), cross-pollination (allogamy) may occur.
Where no nectar-lapping bats are present, pollination is accomplished during the
night by contact between anthers and stigma, which contact is caused by the wind
(Toxopeus, 1948b, 1950). However, the construction of the flower indicates that this
method of pollination is 'unnatural'. In this case, autogamy and geitonogamy
are common, whereas allogamy may only occur when simultaneously blooming
flowers of different trees touch each other. This might sometimes happen in plantations
or on compounds. In the event of strong winds or heavy rains at the time of the
dehiscence of the anthers, pollination is, however, unlikely to occur. During the
night some moths may visit the flowers, but such visits do not play a very important role.
Soon after sunrise, bees and other day insects come, lapping nectar and collecting
pollen (Toxopeus, 1948b, 1950; Jaeger, 1954). They will have soon pollinated all the
flowers in cases where there are sufficient bees in comparison with the number of flo­
wering trees. In plantations bee pollination is of little importance (Toxopeus, 1948b).
Here too self-pollination is common. Other animals are reported to collect nectar and,
by doing so, accomplish pollination, e.g. birds (Fruwirth, 1923), squirrels (Toxo­
peus, 1935) and monkeys (Jaeger, 1954).

ARTIFICIAL POLLINATION

Artificial pollination is done by hand (Opsomer, 1932; Garrido, 1955; Ferwerda,

1966). In Java the recommended time to emasculate flowers is between 15.00 and
16.00 h, but emasculation can also be carried out between noon and 15.00 h. Garrido
(1955) states that in the Philippines the best period for emasculation is between 16.00
and 18.00 h.
The stamens are easily removed with tweezers, without damaging the stigma. The
emasculated flower is then bagged and care must be taken not to enclose any flowers
which are going to bloom within the next four days.
Pollination should be done at night or early the next morning. The pollen grains
are easily collected from the father plant and should be dusted and properly rubbed
on the stigma. Two days after pollination the bag must be removed to allow the
free development of the fruitlet and later the fruit. A bamboo scaffold around the
tree facilitates work with flowers on high branches.

276
 kapok tree

FERTILIZATION

Fresh pollen grains germinate immediately after the arrival on the surface of the
stigma. As the flowers remain on the tree for about 20 hours, the pollen tubes have
to grow to the ovary prior this period. The flowers pollinated by bees in the early
morning have only ten hours for fertilization. This may well be sufficient, as under
favourable conditions it takes about eight hours for the pollen tube to reach the ovule
(Toxopeus, 1948b, 1950). The growth of the pollen tube is at its optimum at night
temperatures of about 20 °C, whereas at temperatures below 17°C the germination
and growth are very retarded and no fertilization takes place (Toxopeus, 1939).
About 72% of the 50,000 flowers of a single standing tree have been found to be suf­
ficiently pollinated (Toxopeus, 1943b, 1954) and in the case of plantation trees, this
percentage has been found to be 15 % (Toxopeus, 1948b).
In a plantation spontaneous cross-pollination rarely occurs. Only in those cases
where a few flowering trees are standing close together it it possible for insects to cause
cross-pollination, in which case up to 16 % are involved (Toxopeus, 1950).

FRUIT SET AND FRUIT DEVELOPMENT

The ovary contains many fertile and many sterile ovules, and it appeared to Toxo­
peus (1948b, 1950) that a certain number of ovules would have to be fertilized in order
to ensure a good initial development of the fruit, this threshold value possibly varying
with the number of fertile ovules and being connected with the fruit length and,
consequently, with the variety. Varieties can have the same number of fertile ovules
or the same fruit length and still have a different threshold value. The threshold value
of a variety with a mean fruit length of 14 cm is 60%, which means that the condition
of development is such that at least 60 % of the fertile ovules have to be fertilized.
Four days after flowering a number of fruitlets drop, a phenomenon called the
'early fruit fall'. These fruitlets have originated from ovaries of which an inadequate
number of fertile ovules have been fertilized.
This can be caused by the failure of sufficient pollen grains to reach the stigma, by
a retardation in the germination of the pollen grains and the growth of the pollen tubes
or by the existence of too low a number of fertile ovules in comparison with the thres­
hold value.
'Late fruit fall' occurs about 16 days after flowering, its cause being physiological
(Toxopeus, 1943b, 1954); i.e. the tree is not able to promote the development of all
fruitlets into fruits. Of a total of about 50,000 flowers on a solitary indica tree about
14,000 (28 %) drop without any growth of the ovary.
During the early fruit fall about 26,000 (52 %) and during the late fruit fall about
6000 fruitlets (12%) are shed, so 64% of the remaining fruitlets are dropped, leav­
ing 4000 fruits (8 %) to be developed (Toxopeus, 1943b, 1954).
In a plantation where the percentage of pollinated stigmas was found to be only

277