L. D.

SPARNAAIJ

grouped in pairs on the basis of the number of bunches (ordinary: n, high: N,) the
average bunch weight (w, W), the fruit-to-bunch (f, F) and the mesocarp-to-fruit ratio
(p, P); one dura (Dl) and one tenera (Tl) are used as standards.
The combinations chosen are limited to those expected to do best if additive in­
heritance is assumed to be operative. The early results from the NIFOR programme
have since demonstrated clearly that additive inheritance is indeed the rule for all
yield and fruit and bunch quality factors (excepting, for the time being, the oil-to-
mesocarp ratio). This justifies a further reduction in the number of crosses be­
tween parents and might thus lead to an improvement in the speed and efficiency
of breeding. Details of a simplified procedure are given in the section on inheritance
(page 377).

The combinations between a pair of dura parents and a pair of tenera parents which
produce the best tenera progeny can be reproduced on a commercial scale from dura
and pisifera palms selected from the sellings of the individual parents but also from
palms selected from the crosses between two parents forming a pair. Dependence on
the (unreliable) progenies from selfings is thus reduced.
This selection and seed production procedure is illustrated in fig. 16 by means of
the following example: Let us assume that two tenera X tenera progenies and four
dura X progenies have been selected as the most productive in the comparative
trials: T3 x T5, T4 x T5, Dl x T3, Dl x T4, D3 x T3 and D3 X T4. Seed and
pollen trees for commercial seed production can be selected from the selfed progenies
of the three tenera and two dura parents. When inbreeding depression prevents or
restricts selection, however, or when the demand for seed cannot be met from the
selfings, use can also be made of the crosses between the parents forming a pair, i.e.
T3 x T4 and Dl x D3. The choice is thus as follows:

Source of dura parent Source of pisifera parent

T5 selfed to be combined with T3 or T4 selfed (or T3 X T4)
T3 selfed with T5 selfed
T4 selfed with T5 selfed
(T3 x T4) with T5 selfed
Dl selfed with T3 or T4 selfed (or T3 X T4)
D3 selfed with T3 or T4 selfed (or T3 X T4)
(Dl X D3) with T3 or T4 selfed (or T3 x T4)
In the succeeding generations the procedure as described above can be repeated an
indefinite number of times using parent palms selected from the best tenera x tenera,
dura x tenera and dura x dura progenies and/or new introductions.
In addition to this and from the second generation onwards, a continued inbreeding
of tenera and dura palms selected from the selfings in the first generation is planned.
Simultaneous dura x pisifera comparative trials will indicate which combination of
'pure lines' is to be used for seed production.

366
 OIL PALM

INTRODUCTION

PROSPECT) ON

PARENT TREES

P3XT3
P3XT4

[ !
i U ,
PISIFERA POLLEN TREES DURA SEED TREES

SEED PRODUCTION ( D X P EXTENSION WORK SEEP|

PLANNED
INBREEDING
PROGRAMME
|TENERA| ]PISIFERA]

1 1
CONTINUED TEST OF CONTINUED
IN-BREEDING DXP IN- BREEDING
COMBINATIONS

Fig. 16 The NIFOR breeding programme schematically represented. The shaded line encloses the
different trials planted simultaneously for one generation. Within the trials, selected progenies (shad­
ed) and progenies used in seed production are indicated.

The IRHO1 breeding programme in the Ivory Coast is very similar to that of
NIFOR in Nigeria. An important difference between the IRHO and the NIFOR
programmes is that NIFOR groups its parent palms according to their distinctive
production characteristics, whereas IRHO groups them according to their origin.
IRHO also uses a far greater number of parent palms but combines them in fewer
crosses. The various origins and their crosses have been studied in detail as regards
their vegetative and yield characteristics and their reaction to climate (IRHO, 1959
and 1960; de Berchoux and Gascon, 1965; Bénard, 1965). The 'Expérience Interna­
tionale', an important exchange programme for seed and pollen organised by IRHO
and involving the Congo (INEAC), the Ivory Coast, Dahomey and the Cameroons

*) IRHO Institut de Recherches pour les Huiles et Oléagineux, Paris, with oil palm research
centres in the Ivory Coast, Dahomey, the Cameroons and Congo (Brazzaville).

367
L. D. SPARNAAIJ

(IRHO) and Malaya (SOCFIN) has provided a wealth of data on this problem. As
a result of these activities, IRHO now strongly recommends the use of Deli X Afri­
can pisifera seed for commercial plantations.

SELECTION

Selection in the nursery

The long pre-nursery and nursery period (11-16 months) provides ample oppor­
tunity for selection before field planting. At an even earlier stage, in the germinator,
selection on time of germination is possible, but experiments to date (Sparnaaij, 1955;
Menendez and Blaak, 1964) have not shown that this procedure has any effect on
performance in the field. Selection in the pre-nursery is not promising either, as at this
stage most observed differences reflect only differences in the date of germination.
Chlorotic and malformed seedlings should, of course, be removed before transplanting
to the nursery.
Nursery selection can be either negative, consisting of the removal of abnormal or
unhealthy plants, or positive, aimed at selecting plants or progenies which are likely
to be more precocious and, perhaps, more productive. The advantages of removing
abnormal plants are obvious. A full description of abnormalities in pre-nursery and
nursery seedlings has been published by IRHO (Bénard, 1964; Bénard and Sérier,
1965). The advantages of the positive selection of individual plants are more question­
able. The most vigorous plants can easily be detected by comparative leaf measure­
ments and, because of their advanced development, these palms are, in general, the
first to start fruiting. Their early yields are, for the same reasons, usually somewhat
higher than those of unselected palms, but nothing can be said about their mature
yields. The results of a statistical experiment at NIFOR in Nigeria show that the 50%
best palms selected in the nursery have a small (4%), but non-significant advantage
over the unselected ones the first five years of production. Even if this advantage is
maintained in later years, it is still too small to justify the doubling of seed production
and, consequently, the lowering of selection standards for seed trees.
Attempts to correlate the number of leaflets in leaves of very young nursery seed­
lings with yields in the mature palms appear more promising. Palms with a higher
leaflet number retain this characteristic in later years and show a significantly higher
average bunch weight, although the yields do not differ noticeably (WAIFOR, 1961 and
1964). Observations by de Berchoux and Gascon (1965) confirm that there is a signi­
ficant correlation in many types of palm between average bunch weight and number
of leaflets.
Of more interest to the plant breeder is the comparison of progeny performance in
the nursery and in the field. Observations at WAIFOR (1965) indicate that there is a
significant correlation between the average height of a progeny in the nursery and early
yields. A selection for (genetic) precocity thus appears practicable. It is, however, too

368
 OIL PALM

early to determine any relationship between the average height of progenies in the
nursery and the yield at maturity.

Selection on yield

The oil yield is a composite figure, the major elements being the bunch number, the
average bunch weight, the fruit-to-bunch ratio, the mesocarp-to-fruit ratio and the
oil-to-mesocarp ratio. In defining selection criteria a choice must be made as to the
priority to be given to any of these elements. Such a choice must be based on the
variation and on the heritability of the various factors.
The variation is greatest in the bunch yield factors (bunch number and average
bunch weight). The highest observed individual bunch yield in a selection field is
usually several times higher than the field average. The maximum oil-to-bunch value
that is theoretically possible is only 1J to 2 times the present day commercial figures.
The scope of yield selection is therefore greater than that of bunch quality selection.
In most breeding centres, however, priority has been given to the improvement of
fruit and bunch quality because this is simpler to effect. Reliable data become available
at a much earlier date (one year is usually sufficient) and heritabilities are higher. Since
the average fruit and bunch quality has risen to such a high level that further spectacu­
lar increases are unlikely, the bunch yield is at present the most important selection
criterion.
In a breeding programme such as that outlined in the preceding pages, yield selection
has to be carried out both at the progeny level (comparative progeny trials) and at the
individual level (selection of seed trees). In order to be effective, selection must be
based on reliable yield records and on some knowledge of the inheritance of yield
and its components.
Progeny trials are designed to determine the relative yields of a number of progenies
under a given set of external conditions. Replication and randomisation eliminates to
some extent the confusing effect on relative progeny yields of soil variations within the
trial area, but it cannot eliminate the effects of soil variations on the comparison of
individual palms. Progeny selection should therefore, whenever possible, precede
individual selection and any high yielding individuals found in poor progenies should
be regarded with suspicion.
Yield recording over a period corresponding to the economic life span of a planta­
tion undoubtedly provides the best basis for selection. In practice, however, selection
must usually be based on much more limited data if a rapid succession of generations
is to be achieved.
At what age and over what period must yield records be taken to provide a reliable
estimate of the performance of progenies or individuals over the whole of their econo­
mic life span?
How can relative performance in a progeny trial be translated into the relative value
of a progeny (or individual) for further breeding or seed production?

369
L. D. SPARNAAIJ

In trying to answer these questions we must consider the following factors affecting
selection efficiency:
- Age and mutual competition for light
- Soil conditions
- Climatic conditions
- Inheritance of yield and its components

Age and competition for light

From the scarce data available on yield progression in palms free from mutual com­
petition it appears that the number of bunches is highest during the first two years of
production and remains fairly constant afterwards at a slightly lower level. The average
bunch weight, however, continues to rise regularly until the palms are about 18 years
old.
In plantations it is impossible to separate age effects from the effects of increasing
competition for sunlight. The leaves of adjacent palms begin to overlap 7-8 years after
planting (9 m triangular spacing) and the amount of sunlight received, in particular by
the lower leaves, is progressively reduced. This only becomes apparent in the yields
9-10 years after planting (because of the 2\ years' delay between a change in external
conditions and the effect on bunch number). Fig. 17 illustrates the yield progression
of a progeny under wide spacing conditions (14 m triangular spacing) without light
competition and the same progeny in two other fields planted under normal spacing
conditions. It appears that the younger palms (Expts. 3-1 and 22-1) reach the same
yield level as the 4 years' older palms at the age of 8-9 years. After that, the yields of
the widely spaced palms continue to rise beyond the level of those planted at normal
density. In the 15th year after planting the yield per palm is twice that of the normally
spaced ones, whereas annual yield variation is much less pronounced under the wide
spacing conditions.
Not all progenies suffer from increasing light competition to the same extent. This
can be demonstrated clearly in the NIFOR spacing trial, where four different triangu­
lar spacings cause competition of increasing severity. The pre-competition yields (first
five years) of eight progenies are compared with the mature yields obtained in the four
spacings. The results are quite remarkable, as is shown in table 6.
At the widest spacing (13 mA) the progenies maintain strictly the same order of
production per palm, but with increasing competition the positions begin to change
until at the closest spacing (7 m A) the order is practically reversed. It appears that
progenies with high yields in the early years (generally those with the highest bunch
numbers) suffer more than the lower yielding progenies from light competition. What
is even more interesting from a plant breeder's point of view is that the first years of
production give an accurate indication of the yield potential of the progenies, i.e. the
yield levels they can reach under optimum light conditions.
The progenies in the spacing trial were selfed dura of only average productivity. The

370
 OIL PALM

Fig. 17 Yield progression in different plantings of the same progeny (Calabar 256, selfed), at NIFOR-
Nigeria.
a. planted 1941, normal density, fertile plots of expt. 5-1
b. planted 1941, normal density, infertile plots of expt. 5-1
c. planted 1946, normal density, fertile area, expt. 3-1
d. planted 1945, low density, 13 m triangular spacing, expt. 22-1.

high yielding progenies of modern breeding programmes are likely to show the compe­
titive effect even more seriously and, because of improved planting practices, at an
earlier age. The oldest statistical dura X pisifera progeny trial at NIFOR may illustrate
this. In table 7 the yields are given for ten different progenies (chosen on the basis of

Table 6 Progeny yields in early years averaged over all types of spacing and yields in later years for
four different triangular spacings, in kg per palm. Expt. 22-1 NIFOR- Nigeria. Planted 1945.

1955-1961
progeny 1950-54 difference
all spacings 13 m A 9mA 8mA 6.5 m A 13 m - 6.5 m

8 176 804 613 457 213 591

5 171 849 610 486 184 665
1 161 776 540 280 114 662
10 158 774 533 348 167 607
7 117 682 572 472 244 438
12 107 609 387 350 226 382
9 97 586 459 455 332 254
4 80 459 354 300 235 223

371
L. D. SPARNAAIJ

early fruit and bunch analysis data) covering a wide range of early yield averages. It
can be seen that in the 2nd + 3rd + 4th year of production the yield variation between
progenies is greater than and the average yield level the same as that in the 6th + 7th
+ 8th year of production. The four best yielding progenies in the early years - those
with the highest bunch number - actually produce considerably less in the 6th - 8th
year of production than in the 2nd - 4th year. As maximum competition is only reach­
ed at the age of 18-20 years, it is expected that progeny yield differences will level out
even more in the later years. Under these circumstances the leeway gained in the early
years may not be lost during the economic life span of the palm and selection based
on early yields would be effective.

Table 7 Yields in different periods for ten different dura x pisifera progenies per palm and per year
in number and weight of bunches (kg). Expt. 4-1 NIFOR - Nigeria. Planted 1952.

1957--1959 1961--1963 1956-1963 yield difference

progeny
1961/63-1957/59
number weight number weight number weight

1 10.3 49.6 6.3 61.1 8.2 50.2 + 11.5

9 13.1 78.4 6.2 67.6 9.8 67.9 - 10.8
10 7.5 55.3 4.2 60.8 5.9 52.6 + 5.5
12 9.5 46.7 5.6 56.1 7.5 46.8 + 9.4
14 9.9 70.0 5.8 66.8 8.1 62.8 - 3.2
18 9.5 61.3 5.1 69.7 7.5 59.4

00
+
24 14.7 78.9 6.7 67.5 10.7 68.3 - 11.4
25 11.8 70.8 6.1 60.9 8.9 61.3 - 9.9
31 11.1 59.0 8.8 73.9 9.9 62.8 + 14.9
49 11.1 61.5 6.0 68.9 8.3 59.3 + 7.4

In the above examples, progeny yields show no further increase with age from
about nine years after planting and differences between progenies, which are impor­
tant up to that age, are greatly reduced or even reversed. This situation can only be
explained as being due to the operation of a yield ceiling by the environment, viz. the
surrounding competing palms and the limited amount of sunlight. The existence of this
ceiling, which varies from year to year under the influence of the weather conditions,
has also been demonstrated by Blaak (1965) in a different way. He shows that within
progenies bunch number and bunch weight vary independently until at a certain ceiling
level a strict negative correlation between the two yield components becomes operative.
The described interaction between age and light competition makes it difficult to
suggest an effective selection procedure. It is clear that yields in the early years (when
taken together) provide a sound basis for selecting the progenies with the highest yield
potential. As an estimate of the actual yields to be expected in a mature plantation,
however, they have only a limited value. An examination of these limitations - the

372
 OIL PALM

action of the yield ceiling and the relative sensitivity of high sex ratio progenies - leads
us to expect that the best results will be obtained with progenies which, in the early
years, have a high yield produced by a moderate number of bunches of high average
bunch weight. This is also suggested by Blaak's figures (1965) for the correlations
between early (2nd-4th year) and mature yields in some progeny trials. For low sex
ratio Deli progenies the correlation is + 0.80, in African dura X tenera progenies it is
+0.59 and in modern high sex ratio dura X pisifera progenies it is only +0.4.

Soil conditions

In modern progeny trials, the statistical design assures that progeny differences are
not confounded with soil differences. Soil variations within and between the plots of
one progeny cannot be eliminated and may interfere with individual palm selection.
Differences of 50 % - 100 % between plots (replicates) of the same progeny in the same
trial have been observed and in such cases corrections must be made to allow for
differences in plot averages before selection is carried out. Soil studies have indicated
that progenies with very variable plot yields do not necessarily show the highest yields
in the most fertile plots ; the contrary may even be the case. This appears to be due to
the depressing effect of high nitrogen levels in the soil. Sensitivity to high soil nitrogen
levels is found in particular in high sex ratio progenies and under conditions of severe
light shortage. This complicated phenomenon has been discussed more extensively in
an earlier paper (Sparnaaij, 1960). It has the wider implication that high sex ratio prog­
enies may show a more severe competition effect when planted in more fertile soils.
In figure 17 the yield progression is shown for the same progeny in both fertile and
less fertile plots (as determined on the basis of leaf production in the first years after
planting).

Climatic conditions

Not all the progeny trials in a breeding programme can be planted at the same time.
It may, therefore, become necessary to compare the progeny yields of plantations of
different ages. From the section on age effects it would appear that age differences need
only to be taken into account in the first four or five years of production. Once the
trials to be compared are all more than eight years old, a direct comparison in the same
year is valid. In cases where the years when yield was recorded do not coincide it is
necessary to correct for climatic differences between years. The most important weather
factor involved is effective sunshine (sunshine during periods of adequate water sup­
ply). A method has been worked out (Sparnaaij, Rees and Chapas, 1963) for the
estimation of the effect of this factor on yields.
The effects of geographic and climatic differences on relative progeny performance
can be quite striking. Within Africa these effects are fairly limited, and breeding and
selection in central research stations appears quite justified. It is not surprising, how­

373
L. D. SPARNAAIJ

ever, that the Deli type, which in Sumatra and Malaya, where sunshine and rainfall
are more plentiful, yields a normal number of heavy bunches, fails to reach normal
yields in Africa owing to a too low bunch number. African palms planted in Malaya
tend to produce rather too many bunches, which remain too small to equal the
yields of the local Deli palms.

INHERITANCE OF YIELD AND ITS COMPONENTS

It is assumed by all oil palm breeders that bunch number and bunch weight are in­
dependently inherited as additive quantitative factors, and most breeding schemes
are based on this assumption. Pronk and Westenberg (1955) were the first to formulate
this hypothesis on the basis of the results of crosses between Deli dura palms and
imported dura and tenera palms from Africa. Gascon, Noiret and Bénard (1966) have
given strong support to this hypothesis by showing that the frequency distribution for
the number and average weight of bunches for Deli dura X African tenera palms is
exactly intermediate between the frequency distributions observed in Deli and African
palms, as can be seen from figure 18. The yield product (number of bunches N x
average bunch weight W) of these hybrid progenies is actually higher than that of the
parents. This has a simple arithmetical explanation, as is best illustrated by a numeri­
cal example :
Yield product Deli = Yx = Nx x W1 = 6 X 18 = 108 kg per palm
Yield product African = Y2 = N2 x W2 = 14 x 8 = 112 kg per palm
XL _l_ N W. 4- W
Yield product Deli x Afr. = YF = 1 2 x — - = 10 x 13 = 130 kg
2 2
t; (Ni - n 2) (w2 - w,)
It can be calculated that YF - Yp = —
' 4

It follows that the maximum increase of progeny yields over the mean parental
yields (Yp) is obtained when the differences in the two yield components are maximum.
The best progeny yields YF can thus be expected from the combination of two high-
yielding origins of very different yield composition. The success of the Deli x African
pisifera cross in commercial seed production is, in fact, based on the wide difference in
yield composition of the two origins rather than on a heterosis effect.
It is, of course, tempting to apply the same principle to individual palms. Individual
yield data are however strongly influenced by the environment and give only a poor
estimate of the genotypic values which are required for the above calculations. Gascon,
Noiret and Bénard (1966) have tried to overcome this difficulty by using as parental
values the average values for all the within-origin progenies of the parents. This elimi­
nates the unknown environmental error due to soil heterogeneity but at the same time
replaces it by an equally unknown genetic error due to the influence of the various male
parents. A more accurate estimate of genotypic parental values is obtained from test

374
 OIL PALM

Fig. 18 Variation in number of bunches, average bunch weight and bunch yield for dura palms of
different origins at La Mé - Ivory Coast (after Gascon, Noiret and Bénard); Deii x Deli,
La Mé x La Mé, ////// La Mé x Deli.

crosses with a common test palm (P0). The procedure is demonstrated below by
means of a numerical example.
1. Cross the parents with the test palm and determine the yield components:
N _i_ N W + W
for progeny (Px X P0) the yield product = 1 —5 X —1 ^ 0 = 9 X 12;

N +N W + W
for progeny (P2 X P0) the yield product = ——-—0 X —2 -—° = 6 x 21;

375
L. D. SPARNAAIJ

The difference between the yield components of these two progenies is half the differ­
ence between the yield components of the parents Pi and P2:
N, + N„ N„ + N N, - N„
-J—' ?_ ° = -J ? = 9-6 = 3
2 2 2
W2 - W 1
Similarly ^— = 21-12 = 9

2. Cross the two parent palms P! X P2 :

N,v + No W, + W,2
Yield product YF = ——? x —~ =7x21

As both the sum and the difference of the yield components are now known, they
can be calculated as follows: Nx = 10, W1 = 12; N2 = 4, W2 = 30
3. Any palm for which the genotypic values for the yield components have been
calculated as above can be used as a test palm for further parent palms. This reduces
the procedure to a single test cross, e.g. using palm Pj as the test palm for P3:
N3 + Ni1 W3 + W,1 N„3 + 10 W3 + 12
Yield product = —- X = x — —
2 2 2 2
If the actual yield product is 8 x 14, we calculate that N3 = 6 and W3 = 16.

The estimates of the yield components calculated in this fashion, though much more
reliable than actual individual yield data, remain subject to the influence of the age and
environment of the test progenies. It follows that they can only be used to compare the
yield composition of palms of the same age group (preferably 5th to 8th year after
planting) and from the same area.
The validity of this mode of calculation is dependent on the validity of the additive
inheritance hypothesis. The NIFOR programme has provided sufficient data to test
this hypothesis statistically. It was possible to demonstrate for 37 dura x tenera pro­
genies, derived from 8 tenera and 11 dura parent palms, that the variance due to the
dura parents and the variance due to the tenera parents were both highly significant
and that the interaction - i.e. the deviation from the exact additive influence of both
parents - was very small for all yield and fruit quality factors (excepting the oil-to-
mesocarp ratio, on which too few data were available). The calculated coëfficients
of variation, which are a measure of the variation due to chance plus the variation due
to an interaction between dura and tenera parents, were for :
number of bunches 3.4 % (2 years' yield data)
average bunch weight 6.2 % (first year's yield)
fruit-to-bunch ratio 1.2 %
mesocarp-to-fruit ratio 1.6%
kernel-to-fruit ratio 7.3 %
shell-to-fruit ratio 6.2 %

376
 OIL PALM

For the fruit-to-bunch ratio and for the number of bunches, the two factors most
strongly influenced by external conditions, the number of progenies used for statistical
analysis had to be reduced (to 29 and 20, respectively) to exclude progenies growing
under different conditions.
As an example, the data for the mesocarp-to-fruit ratio are reproduced in table 8.

Table 8 Mesocarp-to-fruit ratios for 37 dura x tenera progenies in the NIFOR breeding programme

dura tenera parents

parents 32.364 L3208 32.2612 3.1035 1.3352 32.3005 1.2229 3J504

201.32 81.2 79.8 78.8

3.2538 83.1 81.0 79.6 77.7
203.93 79.1 79.4
32.2824 83.7 83.3 79.2 77.1
5.642 83.0 81.2 80.7 80.8
5.368 80.8 78.4 76.3 75.5 72.3
G 98 78.2 78.9 76.3
3.361 75.2 76.3 74.6 73.1
32.658 77.4 74.1 73.8
1.53 76.3 73.8 71.4
2.1997 77.5 71.0

These results, thus, confirm that all yield and quality factors (excepting, for the time
being, the oil-to-mesocarp ratio) are additively inherited and much more strictly so
than was expected. Consequently it seems advisable to amend the present breeding
procedure and to endeavour to obtain a reliable determination of the genetic values for
as many parent trees as possible rather than to test a large number of promising com­
binations between a limited number of parents.
Theoretically, one estimate per parent would be adequate, but, to achieve reason­
able reliability, it is preferable to allow for at least two separate estimates of the genetic
values. A possible programme for 8 dura and 8 tenera parents (cf. the present NIFOR
system on page 365) is the following, comprising a total of only 35 crosses in addition
to the sixteen selfings :
DI X D2 T, X T2 DLF D2 X Tg, T6, TX
D3 X D4 TG X T4 D3, D4 x T7, T8, TJ
OF, X D6 T5 X TG D6, D6 X T1; T2
D7 X DG T7 X T8 D7, DG X TS, T4, T,
Di X T2, TB, T4, T7, TG

By maintaining the same system of pairing as for the NIFOR programme and limit­
ing the dura x dura and tenera X tenera crosses to crosses within pairs, the differences

377
L. D. SPARNAAIJ

between parents are accentuated and the potential progeny performance is increased
in accordance with the formula given on page 374.
The fact that the above mentioned results were obtained from data recorded in the
first year of production (except in the case of the number of bunches, for which two
year's data were used) means, furthermore, that the whole selection procedure can be
accelerated, as the periods during which yields are recorded and fruit and bunches
analysed - on a progeny basis - can be shortened. In the case of progenies selected
for seed production, recording on an individual basis should, of course, be carried out
over longer periods.
In a system in which the relative value of all combinations must be calculated on the
basis of the observations recorded for a few, the accuracy and reliability of the data is
essential. Progeny trials should not be too restricted, should be adequately replicated
and should all contain at least two standard progenies which can be used for the
correction of any differences between trials. All three progenies on which a calculation
of genetic values is based should preferably be planted together in one year and in one
trial. It is therefore advisable not to separate the dura x tenera from the tenera x
tenera progenies as has been the rule. For practical reasons it is also better to plant the
dura x dura crosses and the dura selfings, which both require the same type of indivi­
dual recording and analysis, together in the same trials. The tenera selfings could be
planted in the guard rows.
It remains to be seen whether this method of estimating yield components can be
used for mature palms beyond the age of eight years. At that stage the disturbing in­
fluence of light competition begins to obscure the genetic relationships between parents
and progenies. Raising the potential yield by a more effective choice of parents does
not necessarily raise the yield ceiling set by the environment.
This places the oil palm breeder in a difficult dilemma. Should he aim at producing
material that is adapted to present plantation practice, i.e. material that can make the
most efficient use of the limited light supply? Or should he try to raise the production
potential of his material as expressed in the first four or five years of production?
Either approach can be defended but, in the long run, the latter is the more promising,
provided that the agronomist co-operates with the plant breeder in creating an en­
vironment in which the high yield potential can be realised. In practice this means the
use of a wider spacing.
A temporary solution could be to breed very flexible material that combines both
high bunch weight and reasonably high bunch number factors and to test all progenies
under two different spacing conditions. A procedure of this kind has already been
adopted by NIFOR in Nigeria. The dura x tenera comparative trials are so planted
that half the palms in each plot receive considerably more light than the other half.
In these trials the fields are marked out normally but every fifth row (in a north-south
direction) is eliminated. The progeny plots cover four rows, two of which are 'inside'
rows and two 'outside' rows. The greater the difference between the inside and the
outside palms in a progeny, the wider its optimum spacing. The information so ob-

378
 OIL PALM

Fig. 19 Progeny trial planted according to the four-row system of planting. (Photo NIFOR).

tained could lead to the adoption of a specific spacing or a progressive thinningsystem.

The four-row system of planting (fig. 19) has the added advantage that it greatly faci­
litates the selection for 'dumpy' characteristics. In closed plantings it is almost im­
possible to find palms or progenies with an inherently slower height increment (see
page 384).

Selection on the basis of fruit and bunch quality

"Because the ratio of oil and kernel to bunch is transmitted to the progeny whatever
the environmental conditions, the bunch qualities in plantations are always similar
to those of the parents used for seed production" (IRHO, 1962 Annual Report).
The limited influence of environment and, consequently, the high heritability (see
page 359) facilitates selection on the basis of fruit composition. There are important
seasonal fluctuations but, as long as observations cover one or more whole years, these
will not affect progeny comparisons. When comparing individual palms which may
produce their bunches in different seasons, more than one year's data may be required.
In an extensive selection programme it is not possible to analyse all bunches for one
or more years. The procedure adopted by NIFOR is to take a sample from each pro­

379
L. D. SPARNAAIJ

geny, once each month, during the first year and to do a complete analysis of selected
progenies in a later year. The analysis techniques should be as streamlined as possible,
as the capacity of the analysis section can easily become the bottleneck in a selection
programme. A full description of modern analysis techniques is given by Blaak, Spar-
naaij and Menendez (1963).
A problem which often arises in practice is to select dura seed trees for the produc­
tion of tenera seed. It is, therefore, desirable to express the observed fruit composition
values of the dura in terms of tenera values. As dura and tenera palms often occur in the
same progenies (D X T and T X T) the obvious course has been to study the correla­
tions between dura and tenera siblings. If a sufficiently wide range of progeny values is
included, these correlations are fairly high and significant (Gascon and de Berchoux,
1963), but within the limited range of progeny values encountered in breeding material
such a correlation is virtually non-existent. In fact, it is frequently found in progeny
trials that the average values for tenera palms in a progeny are largely independent of
the average values for the dura palms.
In the following example, the average mesocarp-to-fruit ratios for dura and tenera
palms in 15 dura X tenera crosses from the NIFOR breeding programme demonstrate
that the lack of agreement between the dura and tenera values is not accidental but
is genetically determined (table 9).

Table 9 Average mesocarp-to-fruit values for dura and tenera palms in 15 dura x tenera crosses of
the NIFOR breeding programme.

tenera parents
dura
32.364 32.2612 1.3352
parents
dura tenera dura tenera dura tenera

5.642 51.8 83.0 52.3 80.7 55.2 80.8

32.2824 49.7 83.7 51.8 83.3 - -

5.368 51.7 80.8 51.1 78.4 54.3 76.3

3.2538 47.1 83.1 - - 51.6 79.6
G 98 43.9 78.2 50.9 78.9 51.8 76.3
1.53 - - 45.0 76.3 50.4 73.8

This unexpected phenomenon should be taken together with another remarkable

fact, viz. that in the NIFOR breeding programme as a whole the dura palms in the
dura X tenera progenies are considerably better than the dura palms in the tenera X
tenera progenies from the same tenera parents, whereas their tenera siblings are clearly
inferior to the tenera offspring in the tenera X tenera progenies. The average dura
fruit quality was much better in the dura X dura progenies than in the tenera X tenera
progenies.

380
 OIL PALM

The explanation for this is that the fruit composition in the tenera is not determined
by a single factor, shell thickness, as is the case for the dura, but by two factors, viz.
the potential shell thickness (the extent of the fibre ring corresponding to the shell in
the dura siblings) and the degree of lignification of the potential shell area. Dura parents
have been selected mainly on the basis of the shell-thickness factor and are best in this
respect, whereas the tenera parents are selected primarily because of their low ligni­
fication. This latter factor, which is operative only in the tenera, causes the indepen­
dent variation.
A third factor which plays a role in this concept is the kernel ratio. The kernel size
varies considerably and this has a much more important effect on the tenera (in which
the percentage kernel is of the same order of magnitude as the percentage shell) than
on the dura.An increase in kernel size in tenera palms - the shell-thickness remaining the
same - will lead to an increase in shell percentage and vice versa. The dura parent in a
dura X tenera cross does, however, contribute to the kernel size of the tenera and
thus, indirectly influences the shell percentage. It could be seen in the NIFOR dura X
tenera trials that the dura parents with relatively big kernels (which decreased in size
when being transmitted to the tenera progeny) produced better tenera offspring than
could be expected simply on the basis of their shell content, whereas dura palms with
relatively small kernels gave disappointing results. In extreme cases (Deli dura palms
with thick shells and small kernels crossed with tenera palms with relatively big ker­
nels) this may even lead to a complete lignification of the fibre ring and the appearance
of a surplus of 'dura' palms in the progeny.
In view of the complexity of these relationships it is advisable to calculate the gene­
tic value of tenera and dura parents exclusively on the basis of the performance of their
tenera offspring. The analysis of dura palms in dura X tenera and tenera X tenera pro­
geny trials would appear to have a very limited value.
The oil-to-mesocarp ratio is more influenced by age and environment than the
other fruit quality factors. Both in Sumatra (Janssen, 1959) and in Nigeria, it has been
observed that the oil extraction figures are very low in bunches from very young
palms, particularly in Deli palms (WAIFOR, 1964). Valid comparisons of progenies or
individuals can only be made in plantations of the same age. The seasonal variation is
considerable and affects not only the ratio of oil and moisture to mesocarp but also
the third component, the fibre-to-mesocarp ratio (WAIFOR, 1962); the assumption that
the oil + water content of the mesocarp is a constant quantity cannot, therefore, be
maintained. It was on this assumption that the earlier, indirect analysis methods
(consisting of simple moisture determinations and the subtraction of the results from
a constant) were based. The present-day direct extraction methods are described by
Servant and Henry (1963) and by Blaak, Sparnaaij and Menendez (1963).
The oil-to-mesocarp ratio has long been a selection factor of secondary importance
because its determination is very time-consuming and, as a result, its variability and
heritability are virtually unknown. Now that most breeding centres have raised the
mesocarp-to-fruit ratios of their material to very high levels, the oil-to-mesocarp

381
L. D. SPARNAAIJ

should become the primary selection factor, as it offers the greatest scope for the im­
provement of bunch quality.

Selection of pisifera parents

A selection problem arising from the present seed-production procedure is the

choice of pisifera parents. Pisifera pollen intervenes in the production of virtually all
commercial seed and the limited number of pisifera pollen trees have as much influence
on the quality of the seed as all the dura seed trees put together. Pisifera selection
should, therefore, be even more strict than dura selection. But the pisifera is usually
sterile, in the sense that is it unable to produce ripe bunches (fig. 20), and the normal
selection criteria cannot, therefore, be applied. To overcome this difficulty, the use of
fertile pisifera parents has often been advocated. It has been suggested, however, that
the degree of fertility of a pisifera is inversely related to its inherent fruit quality and
its sex ratio (Sparnaaij, 1960). The fact that fertile pisifera palms are to be found al­
most exclusively in the progeny of relatively thick-shelled tenera has been one of the
reasons for this suggestion. Recent analyses in Nigeria (WAIFOR, 1964) made in a num­
ber of inheritance trials have confirmed this hypothesis and given support to the
WAIFOR policy of the preferential use of sterile pisifera palms for seed production
(fig. 21 p. 384).
While this is, in itself, an important conclusion, it does not solve the problem of the
selection from a group of sterile pisifera. Fortunately, in the present breeding system
this problem arises only in the last stages of selection, i.e. after the tenera X tenera
progenies to be used for reproduction have been chosen. As the number of pisifera
pollen trees required is relatively small, it is practicable to base the selection of pisifera
on more than the usual amount of observations. Progeny testing can be done on a
limited scale, particularly when the progenies can be used in agronomic experiments
on the same station, but it is very time-consuming. A more practical approach is to
base selection on the careful observation of the yield components, particularly on the
sex ratio and the size and structure of the inflorescences. This is quite possible, because
pisifera bunches normally abort after anthesis. In this way a fair estimate is obtained
of the major yield factors, viz. the bunch number and the average bunch weight. An
idea of the fruit quality can be obtained by treating the inflorescences at the anthesis
stage with growth-active substances such as a 2, 4, 5, tri-chlorophenoxy propionic
acid, this causing parthenocarpic fruit setting. The extent of the fibre zone can give
an impression of the inherent shell-thickness factors, even though the shell itself is
absent (de Jong, 1967, private communication). This procedure has the added advan­
tage that it may unmask any sterile tenera or dura palms which might otherwise be
taken as pisifera and cause incorrect segregation.
A totally different approach is to mix the pollen from all pisifera palms in a selected
progeny on the assumption that the pollen mixture would represent the average value
of the progeny. This procedure has been in use in the Congo and in Nigeria but it was

382
 OIL PALM

Fig. 20 Pisifera palm carrying a large number of exclusively female inflorescences of which the older
ones have aborted. (Photo NIFOR).

abandoned when it was discovered that differences in the speed of germination of the
pollen defeated the object of mixing.
In view of the very large progeny of any pisifera pollen tree, it is of great value to

383
L. D. SPARNAAIJ

DURA x PISIFERA

8 7 6 5 4 3 2mm fertile sterile

\b I3 2 0.1mm
Fig. 21 Schematic representation of the paren­
tage of very thick-shelled, normal and very thin-
TENERA shelled tenera.

test them for disease resistance before using them for seed production. Susceptibility
to 'blast' is usually quite evident in normal nurseries, although annual variations in
'blast' incidence may necessitate repeated tests. Simple inoculation methods for testing
nursery seedlings have been developed for dry basal rot (WAIFOR, 1963) and wilt
(Prendergast, 1963). Moreover, field observations in the Ivory Coast have confirmed
the value of the nursery test for wilt-resistance (IRHO, 1964).

Selection on the basis of growth rate

The expression of growth rate is very much influenced by environment. Under

normal spacing conditions, palms with an inherently slow growth either etiolate to
reach the same level as the others or they become overshadowed and their yield capa­
city cannot be assessed. These difficulties can be overcome by selecting on a progeny
basis and by using the four-row system of planting (see page 378).

THE RESULTS OF OIL PALM BREEDING AND FUTURE PROSPECTS

Breeding and selection in the oil palm has been going on for more than 40 years in
Sumatra and Malaya and for only a slightly shorter period in Africa. The major
objective has been, throughout, to increase the yield of oil per hectare. How successful
the oil palm breeders have been is difficult to estimate, as the extent of the contri­
butory effects of improved agricultural practices and of improved planting material
cannot easily be gauged separately.
The highest yield levels have been reached in the Asian oil palm areas, where clima­
tic conditions are at an optimum. The increase in the course of subsequent generations
has, however, been more spectacular in Africa, where early plantings consisted of
unselected bush palms.
In Asia, the first plantations were established on the basic of seed derived from the
palms first introduced at Bogor, which happened to be of reasonably good yield and
fruit composition. Figures quoted by Janssen (1959) suggest that the first selection of

384
 OIL PALM

this material, based on number of bunches and thickness of mesocarp, resulted in a

yield increase of 20 % in the case of young palms to 50 %-70 % in the case of mature
palms. At that time the yield of unselected material at the AVROS research station in
Medan was no more than 1.2 ton of oil per ha as compared with 2 ton of oil for select­
ed material. Introductions from Sumatra planted in Malaya in the same period now
yield about 3 tons per ha as against 4 tons for post-war Deli dura material. This would
suggest that since the first selection of the basic material around 1920, which resulted
in an increase of, very roughly, 65 %, a further increase of about 35 % has been achieved
in later generations. This corresponds to a total increase due to breeding and selection
of roughly 120-125%. In addition to this, the improvement in plantation practices
would have added about 50 % to the general yield level.
The advent of the tenera in Asia has further raised commercial yield levels. With
extraction around the 22% level and bunch yields up to 20-23 tons/ha, the yield of oil
reaches 5 tons per ha.
In Nigeria, present yield levels are 6-8 tons of bunches for unselected bush palms
and 10-12 ton bunches for improved dura X pisifera palms. Owing to the poor fruit
quality of the average bush palm, the actual increase in oil yield has been from 0.6-1
ton to 2.5 ton per ha. In the Ivory Coast, the IRHO expects oil production in the new
Deli dura X pisifera plantations to exceed 3 tons per ha.
In view of the fact that there have been so few generations of breeding and selec­
tion, particularly in Africa (where there have been no more than three), more improve­
ments are likely to be introduced in the future resulting from more systematic pros­
pecting, a better understanding of yield inheritance and an adaptation of cultural
practices to the increased genetic yield potential.

References

BEIRNAERT, A., 1935. Introduction à la biologie florale du palmier à huile. Publ. de l'INEAC. Série
scientifique 5.
BEIRNAERT, A. and VANDERWEYEN, R., 1941. Contribution à l'étude génétique et biométrique des
variétés à'Elaeis guineensis Jacquin. Publ. de l'INEAC, Série scientifique 27.
BÉNARD, G., 1964. Palmier à huile - Sélection en pépinière (Conseil de l'IRHO, no 28) Oléagineux,
19 (4): 243-246.
BÉNARD, G. and MALINGRAUX, C., 1965. La production de semences sélectionnées de palmier à huile à
l'IRHO. Principe et réalisation. Oléagineux, 20 (5) : 297-302.
BÉNARD, G. and SÉRIER, J. B., 1965. Palmier à huile. Sélection en prépépinière. (Conseil de l'IRHO no
39). Oléagineux, 20 (4): 237-239.
DE BERCHOUX, CHR. and GASCON, J. P., 1965. Caractéristiques végétatives de cinq descendances
A'Elaeis guineensis Jacq. Oléagineux, 20 (1): 1-7.
BLAAK G., 1965. Breeding and inheritance in the oil palm; Part III: Yield selection and inheritance.
J. Nigerian Inst, for Oil Palm Res., 4 (15): 262-283.
BLAAK, G., SPARNAAIJ, L. D. and MENENDEZ, T., 1963. Breeding and inheritance in the oil palm;
Part II: Methods of bunch quality analysis. J. W. Afric. Inst, for Oil Palm Res., 4 (14): 146-155.
BROEKMANS, A. F. M., 1957. Growth, flowering and yield of the oil palm in Nigeria. J. W. Afric. Inst,
for Oil Palm Res., 2 (7) : 187-220.

385
L. D. SPARNAAU

CARRIÈRE DE BELGARRIC, R., 1951. Note sur la sélection du palmier à huile à Sumatra. Oléagineux, 6
(2): 65-71.
CHAPAS, L. C, 1961. Plot size and reduction of variability in oil palm experiments. Empire Journ. of
Exper. Agric., 29: 212-224.
FERRAND, M., 1946. Découvertes récentes dans la génétique du palmier à huile et leurs conséquences
quant à la sélection de ce végétal. C. R. des Séances de l'Ac. d'Agr. de France, 32: 75-79.
GASCON, J. P., BÉNARD, G. and BREDAS, J., 1965. Sélection des semences, des plantules et des plants
chez le palmier à huile. The oil palm. Papers presented at Tropical Products Institute Conference,
London: 12-16.
GASCON, P. and DE BERCHOUX, C., 1963. Quelques relations entre les dura et tenera d'une même
descendance et leur application à l'amélioration des semences. Oléagineux, 18 (6): 411-415.
GASCON, J. P. and DE BERCHOUX, C., 1964. Caractéristiques de la production d'Elaeis guineensis
(Jacq.) de diverses origines et de leurs croisements. Application à la sélection du palmier à huile.
Oléagineux, 19 (2) 75-84.
GASCON, J. P., NOIRET, J. M. and BÉNARD, G., 1966. Contribution à l'étude de l'hérédité de la produc­
tion de régimes d'Elaeis guineensis Jacq. Oléagineux, 21 (11): 657-661.
GASCON, J. P. and PREVOT, P., 1957. Orientation de la sélection du palmier à huile à l'IRHO. Comptes
Rendus de la Conférence Franco-Britanique sur le Palmier à Huile. Bull. Agron. Fr. d'O.M., 14: 42.
GRAY, B. S., 1965. The necessity for assisted pollination in areas of low male inflorescence production
and its effect on the components of yield in the oil palm (Elaeis guineensis). The Oil Palm. Papers
presented at Tropical Products Institute Conference, London: 17-31.
GUNN, J. S., SLY, J. M. A. and CHAPAS, L. C., 1961. The development of improved nursery practices
for the oil palm in West Africa. J. W. Afric. Inst, for Oil Palm Res., 3 (11): 198-228.
HARTLEY, C. W. S., 1957. Oil palm breeding and selection in Nigeria. J. W. African Inst, for Oil Palm
Res., 2(6): 108-115.
HARTLEY, C. W. S., 1967. The oil palm. Longmans, London.
I.R.H.O., 1959, 1960, 1964. Institut de Recherches pour les Huiles et Oléagineux. Rapports Annuels.
JAGOE, R. B., 1952. 'Deli' oil palms and early introductions of Elaeis in Malaya. Mai. Agric. Journal,
35(1): 3-11.
JANSSEN, A. W. B., 1959. 'The productivity of the oil palm' and 'Oil palm productivity and genetics'.
Chemara Res. Station. Miscellaneous notes on estate agriculture in Sumatra. No. 6 and No. 7.
LABRO, M. F., GUENIN, G. and RABÉCHAULT, H., 1964. Essais de levée de dormance des graines de
palmier à huile (Elaeis guineensis Jacq.) par des températures élevées. Oléagineux, 19 (12) : 757-765.
MENENDEZ, T. and BLAAK, G., 1963-64. Plant Breeding Division Annual Report 1963-64. Twelfth
Annual Report W. Afric. Inst, for Oil Palm Res.
NOIRET, J. M. and GASCON, J. P., 1967. Contribution à l'étude de la hauteur et de la croissance du stipe
d'Elaeis guineensis Jacq. Oléagineux, 22 (11): 661-664.
NOIRET, J. M., GASCON, J. P. and BÉNARD, G., 1966. Contribution à l'étude de l'hérédité des caracté­
ristiques de la qualité du régime et du fruit d'Elaeis guineensis Jacq. Oléagineux, 21 (6): 343-349.
PICHEL, R., 1957. L'amélioration du palmier à huile au Congo Belge. Comptes Rendus de la Confé­
rence Franco-Britanique sur le Palmier à Huile, 1956. Bull. Agron. Fr. d'OM., 14: 59-66.
PRENDERGAST, A. G., 1963. A method of testing oil palm progenies at the nursery stage for resistance
to vascular wilt disease, caused by Fusarium oxysporum Sehl. J. W. Afric. Inst, for Oil Palm
Res. 4 (14): 156-175.
PRONK, F., 1955. Productie verwachtingen van het Deli-tenera type (Yield expectations of the Deli
tenera type). Bergcultures, 24: 379-381.
PRONK, F. and WESTENBERG, M. G., 1955. AVROS circulars and unpublished report.
REES, A. R., 1959. The germination of oil palm seed. 'The cooling effect' and 'Large scale germination'.
J.W.Afric. Inst, for Oil Palm Res., 3 (9): 76-94.
REES, A. R., 1965. Some factors affecting the viability of oil palm seed in storage. J. Nigerian Inst, for

386
 OIL PALM

Oil Palm Res., 4 (15): 317-324.

SERVANT, M. and HENRY, J., 1963. Détermination de la richesse en huile de la pulpe des fruits de palme
Oléagineux, 18 (5): 331-341.
SPARNAAIJ, L. D., 1955. Seedling selection in oil palm nurseries. J. W. Afric. Inst, for Oil Palm Res.,
1 (3): 51-63.
SPARNAAIJ, L. D., 1960. The analysis of bunch production in the oil palm. J. W. Afric. Inst, for Oil
Palm Res., 3 (10): 109-180.
SPARNAAIJ, L. D. and GUNN, J. S., 1959. The development of transplanting techniques for the oil palm
in West Africa. J.W.Afric. Inst, for Oil Palm Res., 2 (8): 281-309.
SPARNAAIJ, L. D. and MENENDEZ, T., 1962. Recent advances in the theory and practice of oil palm
breeding in Nigeria. Proc. XVIth Int. Hort. Congress: 458-463.
SPARNAAIJ, L. D., MENENDEZ, T. and BLAAK, G., 1963. Breeding and inheritance in the oil palm. Part.
I: The design of a breeding programme. J.W.Afric. Inst, for Oil Palm Res., 4 (14): 126-145.
SPARNAAIJ, L. D., REES, A. R. and CHAPAS, L. C., 1963. Annual yield variation in the oil palm. J. W.
Afric. Inst, for Oil Palm Res., 4 (14): 111-125.
VANDERWEYEN, R. and MICLOTTE, H., 1949. Valeur des graines d'Elaeis guineensis Jacq. livrées par la
station de Yangambi. Publ. de l'INEAC, série technique no 37.
VANDERWEYEN, R. and ROELS, O., 1949. Les variétés d'Elaeis guineensis Jacq. du type albescens et
l'Elaeis melanococca Gaertner. Publ. de l'INEAC, série technique no 28.
VANDERWEYEN, R., ROSSIGNOL, J. and MICLOTTE, H., 1945. Contributions à l'étude d'Elaeis guineensis.
Unpublished report INEAC, Yangambi.
WAIFOR., 1961-1964. West African Institute for Oil palm Research Annual Reports.
WAIFOR., Notes on the botany of the oil palm. 1. The seed and its germination; 2. The seedling; 3.
The stem and the stem apex; 4. The leaf. J. W. Afric. Inst, for Oil Palm Res., 1 (1955), 3; 2 (1956),
5; 3 (1961), 11 and 3 (1962), 12.
WORMER, TH., 1958. Croissance et développement du palmier à huile. Oléagineux, 13 (14): 385-393.
YAMPOLSKY, C., 1922. A contribution to the study of the oil palm. Bull. Jard. Bot. Buitenzorg, Série
III, vol. 5(2): 107-174.
ZEVEN, A. C., 1965a. The origin of the oil palm (Elaeis guineensis Jacq.). A summary. J. W. Afric.
Inst, for Oil Palm Res., 4 (15): 218-225.
ZEVEN, A. C., 1965b. Oil palm groves in Southern Nigeria. Part. I. Types of groves in existence. J.
Nigerian Inst, for Oil Palm Res., 4 (15): 226-250.
ZEVEN, A. C., 1967. The semi wild oilpalm and its industry in Africa. Agric. Res. Rep. Nr. 689. Pudoc,
Wageningen. 178 pp.

387
PAPAYA

Carica papaya L.

WILLIAM B. STOREY

University of California, Citrus Research Center, Riverside, California

Systematics

TAXONOMY

The papaya (Carica papaya L.) is a dicotyledonous species of small, semiwoody,

tropical fruit trees that are prized for their palatable melonlike fruits. It belongs to the
family Caricaceae which consists of the following genera: Carica, Cylicomorpha,
Jacaratia, and Jarilla. Carica contains about 21 species, all indigenous to tropical
America. Cylicomorpha contains two species native to tropical Africa. Jacaratia con­
tains six species, all native to tropical America. Jarilla consists of a single species
native to central Mexico.
Carica is the only genus having species that are cultivated for their fruits. The other
three genera have a number of species, however, that are cultivated as ornamentals.
The fruits of C. candamarcensis Hook, f., C. monoica Desf., C. pentagona Heilborn,
C. erythrocarpa Heilborn, C. goudotiana Solms-Laubach, and C. quercifolia Benth.
and Hook, are edible. They are seldom eaten in the fresh state, however, because they
lack the ready palatability of the papaya. Some have been assembled into collections
by plant breeders, plant science institutes, and experiment stations for use in attempts
to produce hybrids between one or more of them and C. papaya.

CENTER OF ORIGIN

The concensus of opinion among botanists is that the papaya originated in the low­
lands of Central America somewhere in the region between southern Mexico and
Nicaragua.
The origin of the papaya as a cultivated fruit tree is lost in antiquity. Certainly, it
must have been domesticated by an early civilization in tropical America. By the time
it became known to European botanists and gardeners, it had already varied under
cultivation into a large number of diverse types.

389
WILLIAM B. STOREY

DISTRIBUTION

Following discovery of the New World, the papaya was distributed along tropical
trade routes by travelers on the ships of explorers and of traders of various maritime
nations. It reached Panama as early as 1535, Puerto Rico by 1540, and Cuba soon
thereafter. By 1611, it was being grown in India, and by 1800 was widely distributed
among the numerous islands of the South Pacific Ocean. Today it is grown extensively
throughout the tropical and extra-tropical regions of the world both as a plantation
tree and as a favorite fruit tree for the home garden. In the past 60 years the fruit has
continually increased in popularity, and the tree has gained importance as a plantation
crop in Hawaii, South Africa, Australia, India, Ceylon, the Philippines, and a number
of countries in tropical America and southeastern Asia.
A compelling reason for the high esteem in which it is held by peoples inhabiting the
tropics, in addition to palatability, is its characteristic of producing fruit without inter­
ruption the whole year around. In this respect it outdoes the great majority of tropical
fruit species, for these, like fruit plants of the temperate zones, have rather distinct,
comparatively short crop seasons. A papaya tree in a favorable environment and
provided with good culture begins producing ripe fruits in about a year from the time
it appeared as a newly germinated seedling, and can live and continue to bear for 25
years or more.

MATING SYSTEM

The papaya is a polygamous species of plant. For practical convenience, the trees
are generally classified into three primary sex types : 1. staminate, or male ; 2. herma­
phrodite, or bisexual; 3. pistillate, or female.
Male trees are characterized by long, pendulous, ramified, cymose, many-flowered
inflorescences consisting either exclusively or preponderantly of staminate flowers.
Hermaphrodite trees have relatively short, few-flowered inflorescences consisting
mainly of bisexual flowers. Female trees have short inflorescences usually consisting
of only five or six flowers that are pistillate exclusively.
In nature, probably, and to some extent as a result of Man's interference, isolated
populations or strains may consist of males and females only (dioecious), of herma­
phrodites and females only (gynodioecious), and of all three sex forms, (trioecious, or
polygamous).
Pollination apparently is effected largely by wind, but insects are believed to play a
part also. Wind especially favors dissemination of the pollen of male trees. Insect ac­
tivity probably results in the transfer of pollen from male and hermaphrodite trees
about equally. Some forms of hermaphrodites are self-pollinating in the mature un­
opened flower bud, precluding pollination by other trees later when the flowers are in
anthesis.
Weather conditions and other factors can determine whether wind pollination or

390
 PAPAYA

insect pollination is the more effective on a given day. Under the circumstances, popu­
lations quite different in composition can diverge from an original trioecious source.
In fact, if seed were taken from a single fruit and planted elsewhere it could immediate­
ly establish any one of the three classes of populations.
The papaya, although regarded as a cross-pollinating species, can be self-pollinated
without loss of vigor. The breeding system generally used, therefore, is hybridization
of selected phenotypes among various varieties and strains followed both by inbreed­
ing and backcross breeding procedures. The latter offers great promise because of the
longevity of the individual selected as the recurrent parent.

Physiology of development

GERMINATION OF SEEDS

Almost invariably papaya trees are grown from seeds. They can be propagated
vegetatively (Allan, 1964) but this is seldom done for commercial planting because the
expense is not justified by the relatively short economic life of a plantation.
Well-pollinated papaya fruits contain 300-700 viable seeds. These are easily re­
moved from the ripe fruit by scraping out the placenta to which they are attached. The
seeds germinate readily when sprinkled over the surface of a sandy potting soil in a
flat box and covered with a thin layer of sand, peat moss, or composted coconut fiber,
or in vermiculite, and kept damp in a warm place.
The papaya seed consists of a small laterally flattened embryo surrounded by endo­
sperm, and a seed coat consisting of a dark, hard, muricate endotesta and a translu­
cent sarcotesta containing a thin mucilaginous fluid. Experimental results have shown
that seeds with the sarcotesta removed germinate more rapidly and uniformly than
those with it left intact. Removal is easily accomplished by fermenting the seeds and
any adherent placenta in a dish of water for two or three days. The sarcotestas break
easily when the seeds are rubbed gently in a fine-meshed sieve or squeezed in a piece
of cloth. The seeds are washed to remove extraneous material by putting them in a
vessel containing water. The viable seeds sink, while the nonviable seeds, remnants of
the placenta and sarcotestas, and other debris float and can be skimmed off.
The seeds can be planted immediately, or can be stored if dried. The principal pre­
cautions are to do the drying in shade when they are air dried, and to hold the tem­
perature below 32°C when they are dried with artificial heat, to avoid the danger of
killing the embryos.
Seeds will remain viable in storage for about a year if kept at about 12°C in a
tightly capped jar.

GROWTH

The seedlings emerge in two to three weeks. About three to four weeks after ger-

391
WILLIAM B. STOREY

minating, they are planted in individual containers. In another three to four weeks
they are 15- 20 cm tall. At this time they are removed from the containers and trans­
planted into their permanent places in the field.

SHORTENING THE JUVENILE STAGE

Neither a method for shortening the juvenile stage nor a genetic factor which might
do so has been found. Some strains are more precocious than others, but all fall within
the normal range of four to eight months of growth required from the time of ger­
mination to the initiation of flower buds.
The present approach is to combine the precocity and low stature that are inherent
in some strains with the desirable fruit and tree characters of less precocious, taller
growing strains.

Biology of flowers

INFLORESCENCE

Depending on variety or strain, papaya seedlings begin to flower four to eight

months after germination. Once they have begun to flower they continue to produce
an inflorescence in the axil of each leaf without interruption if conditions for growth
are favorable. New leaves emerge at the rate of about two a week, or approximately
100 a year. If a fruit sets in each leaf axil, one can expect a yield of 100 or more fruits a
year. Yields in terms of weight depend upon the variety. Fruit weights of various
varieties and strains in cultivation range from 110 g to 9.5 kg and more.
The papaya inflorescence is a cyme. On male trees it is long, pendulous, and freely
branching. On hermaphrodite and female trees it is reduced to a few inches in length
and a few flowers in number.

FLOWERS

The flowers themselves are of three sex types: 1. staminate; 2. bisexual; 3. pistil­
late.
The staminate flower has a sympetalous corolla which forms a slender tube about
2.5 cm long, surmounted by a five-parted limb of about equal length. There are ten
stamens which are inserted at the throat of the corolla tube. The flower is unisexual in
function, for only a pistillode is present in the position which, in the bisexual flower,
is occupied by the gynoecium.
The bisexual flower is similar in structure to the staminate flower, but is larger in
overall size and broader in girth through the corolla tube. Typically, it has a 5-car-
pellate ovary with parietal placentation. Fruits produced from bisexual flowers tend
to be long-cylindrical, obovoid, or pyriform, depending on variety.

392
 PAPAYA

The pistillate flower has a large functional pistil but is entirely devoid of stamens.
Like that of the bisexual flower, the ovary is 5-carpellate with parietal placentation. It
is ovoid in shape, however, Superficially, the petals appear to be free from one an­
other; actually, they are inconspicuously fused at the base and adnate to the base of the
ovary. The fruits of pistillate flowers of different varieties range from oblate spheroidal
through spherical to oval in shape.
Between the extremes of the staminate flower at one end and the pistillate flower at
the other, there is a large number of intermediate and teratological forms. Lange (1961)
described seventeen. The variations appear in response to changes in environmental
conditions. They are responsible for the wide range of phenotypic variations which
occur among staminate and hermaphrodite trees. The most recent classification
(Storey, 1958) reported 15 forms of hermaphrodite trees, 15 forms of male trees, and
one form (possibly two) of female trees.
The variations in flower type result from two sets of genetic factors, one affecting
female fertility, the other causing stamens to become carpelloid. Both sets of factors
are influenced by a third set which determines the time of expressivity. Such instability
indicates that, although the primary sex type of the tree is determined genotypically,
phenotypic expressions of the alleles for the presence or absence of an androecium
and those for the presence or absence of a gynoecium are influenced by prevailing
environmental factors at the time of flower bud initiation. Critical studies of some
of the factors and their effects have been reported by Awada and Ikeda (1957) and
Awada (1958).
The variations in flower type and, therefore, in tree type in a hybrid progeny are a
source of annoyance, as well as a challenge, to a person trying to develop a stable,
uniform variety.

POLLEN ANDSTORAGE

Fresh pollen is preferred for hybridization. If, for some reason, one wishes to store
pollen, he can do so by putting it into cool storage at 10°C with the relative humidity
of the air held at about 10%. Under these conditions, pollen remains viable for six
months, or longer.

Improvement

VARIABILITY

The papaya exists in a great diversity of types, some of which offer great promise for
use in breeding because they possess attributes that are considered to be desirable
horticulturally.
There are few true-breeding varieties or strains of papaya. This is so because, almost
invariably wherever it is grown, seeds for succeeding generations are taken from open-

393
WILLIAM B. STOREY

pollinated fruits without regard to the pollen source. In order to maintain the identity
of a named variety or a desirable strain, either the trees from which seeds are to be
taken must be grown well isolated from trees of other varieties and strains, or the
flowers which are to produce the fruits for seed purposes must be hand pollinated with
pollen from the appropriate source and protected against contamination by unwanted
sources. If precautions are not taken against open pollination where two or more
varieties grow together, mongrelization sets in and varieties lose their identities in a
matter of only two or three generations.
The Solo variety of Hawaii, and possibly Hortus Gold of South Africa, Improved
Petersen of Australia, and Betty of Florida, owe their continued existence as entities
to the fact that the seeds for each succeeding generation are produced under controlled
conditions. The Solo is in its 55th year and approximately 25th generation of inbreed­
ing and selection since its introduction into Hawaii from Barbados in 1910. Meanwhile,
a number of lines of Solo which were selected for especially promising characters are
now in their fifth to eighth inbred generation. Hofmeyr (1938) observed that inbreeding
did not seem to cause loss of vigor in papaya. This was confirmed experimentally by
Hamilton (1954).

EXTENDING THE RANGE OF VARIABILITY

The literature contains several reports of successful hybridizations between various

species of Carica, including C. papaya x gracilis, C. papaya x cauliflora, C. papaya X
quercifolia, and C. papaya X peltata. There seem to be no follow-up reports, however,
and no report of a successful cross more recent than 1952. It is presumed, therefore,
that all hybrid material of this sort has been lost. Interspecific hybridization apparently
can be accomplished, and awaits the breeder who wishes to take advantage of it to
extend the range of variability.
Polyploids have been produced by treating seeds with colchicine, but none has been
used for further breeding. There are no reports, either, of discovery of desirable mu­
tants or of induction of mutations by radiation or other means.

CLONE AND ROOTSTOCK PROBLEMS

The papaya can be propagated clonally both as rooted cuttings, and as scions graft­
ed on seedlings and cuttings. This is seldom done, however, except for special pur­
poses, and rarely done for commercial planting. When it has been done, no serious
problems have developed with either rooted cuttings or the rootstocks of grafted
plants growing in the field.

BREEDING METHODS

The papaya exists in a few more or less purebreeding varieties that have been pro­

394
 PAPAYA

duced by plant breeders, and a large number of more or less homogeneous, fairly uni­
form strains that have developed in isolated regions or plantings through generations
of interbreeding among the individuals making up the population.
The breeding procedure most widely used is to assemble as many of the strains as
possible in the same location, and select those among them having phenotypic charac­
ters most closely approximating the desired combinations for hybridizing.
After the initial cross, two approaches are followed: 1. carrying the F, to F2 and
successive generations in order to enhance the probability of finding recombinations
of the desired characters by means of the pedigree method of breeding; and 2. by
means of backcross breeding procedures, especially if one parent already possesses a
high order of desirable characters which make it suitable as the recurrent parent.
Application of the knowledge of the genetics of sex determination in Hawaii in
developing superior lines of the Solo variety, and of gynodioecious hybrid lines from
an initial cross of Betty X Solo, emphasize the point that there is no justification for
developing dioecious varieties in which 50 % of the trees are males and, therefore, use­
less as producers of fruit.
Studies of fruit size and shape, precocity, and tree stature have provided information
that allows one to predict with a fair degree of certainty what the hybrid between in­
dividuals of two different strains will be like.

HYBRID VARIETIES

Probably the only bona fide varieties in existence are Solo and Bush of Hawaii, and
Hortus Gold of South Africa. These originated as intraspecific hybrids between dis­
tinct strains.
The source of Solo is known, although its parentage is not. The fruit from which the
seed was taken was purchased in a market in Barbados in 1910 by Gerrit P. Wilder,
who described it as small and bananalike. The resulting progeny and successive gene­
rations have had hermaphrodite trees with pyriform fruits weighing about 450 g.
Since neither the bananalike type nor the pyriform type occurs on Barbados, Wilder's
collection may have been an F2 segregant of the cross of a small wild type of papaya
known as 'lechosita', which is common in the West Indies, and the large commonly
cultivated type. Two inbred lines of Solo, Line 5 and Line 8 selected in 1939 and 1953
respectively, account for virtually all of Hawaii's annual production of 18,000,000 kg.
In 1936, papaya production in Hawaii averaged 11,120 kg per ha of marketable
fruit. In 1965, it averaged 44,480 kg per ha.
Bush is a red-fleshed Solo type which the writer developed in an experiment to
combine red flesh color with the pyriform shape and small size of Solo. The female
parent was a red-fleshed segregant of the cultivar Fairchild from Florida. The pol­
len parent was a selected Solo hermaphrodite tree. This hybrid has been improved
at the Hawaii Agricultural Experiment Station by five generations of inbreeding,
and enjoys some local sale.

395
WILLIAM B. STOREY

Hortus Gold is a dioecious variety which was introduced into the South African
trade by Hofmeyr in 1936.
The last published list of papaya varieties contained the names of 61 varieties and
strains (Alonso Olivé, 1952). Not listed are at least ten others which have been given
variety or strain designations in the past 15 years. It is quite likely that none of those
listed exists today in the form in which it was introduced excepting Solo, because no
one concerned with their perpetuation has taken the pains to protect them from mon-
grelization by replanting each successive generation only with seeds from controlled
pollinations or by growing them at a safe distance from sources of contamination.

SCREENING

Screening out of undesirable individuals and selection of desirable phenotypes is

important in papaya breeding. A major breeding objective is development of strains
in which every tree produces marketable fruit. Sterile and semisterile hermaphrodites
must be eliminated. So, also, must hermaphrodites that produce high percentages of
malformed fruits resulting from carpellody of stamens.
In order to weed out sex-reversing forms, one should keep the trees under observa­
tion for a full year, at least, and if feasible for two years. The latter is desirable because,
depending upon the time of planting and environmental factors, an undesirable in­
dividual may escape a single inspection by commencing flowering out of phase with
its normal pattern of behavior.
An important aspect of papaya breeding is that the work should be done in the
locality where the crop is to be grown. The papaya is sensitive to the effects of micro­
climates, and strains and varieties which have been bred for and do well in one locality
sometimes 'fall apart' in another locality which may differ quite imperceptibly in the
environment it provides.

GENETICS OF IMPORTANT CHARACTERS

Genetics of sex

In discussion of the mating system, the papaya was described as a polygamous

species of plant which is characterized by three primary sex types : male, hermaphro­
dite, and female. The genetics of sex determination, the hypothetical genes involved,
and the hypothetical structure of the sex chromosomes have been discussed fully else­
where (Storey, 1953; Horowitz, 1954).
Reduced to simplest terms, the genetics of sex determination may be likened to a
case of monohybrid inheritance involving three alleles with pleiotropic effects. The sex
homologues can be symbolized as M, MH, and m for male, hermaphrodite, and female,
respectively. Combinations of dominants, i.e. MM, MMH, MHMH are lethal to the
zygotes receiving them. Consequently, all existing males and hermaphrodites are

396
 PAPAYA

Table 1 Pollination combinations and resulting segregation ratios in progenies of Caricapapaya.

Segregation ratios

Pollination 9 2 <? Nonviable zygotes

{mm) {M^m) {Mm) {MM; MHM; M"MH)

1. ? X((; {mm x Mm) 1 1

2. Ç x £ ; {mm x M^m) 1 1
3. c? selfed {Mm ® ) 1 2 1
4. <? x (J; {Mm x Mm) 1 2 1
5. § selfed {MHm ®) 1 2 1
6. ? x {MHm x MHm) 1 2 1
7. y x (J; {M^m x Mm) 1 1 1 1
8. cJ x {Mm X M^m) 1 1 1 1

enforced sex heterozygotes. The sex genotypes can be represented as Mm, MHm, and
mm for male, hermaphrodite, and female, respectively.
Eight pollination combinations can be made. These are given in table 1, with their
corresponding segregation ratios. The seemingly paradoxical self-pollination of males,
pollinations between males, and pollinations of males by hermaphrodites are possible
because the males involved are sex-reversing forms which produce perfect flowers at
some time during the year.
Although the basic sex type is determined genotypically, certain male and herma­
phrodite trees undergo sex reversal in various degrees, as well as other sorts of mor­
phological variation, under the influence of seasonal changes in climate. The variations
which are known to occur have been summarized by Storey (1958).
The female is stable phenotypically, and excepting the occurrence reported by Hof-
meyr (1939) of several female trees in a single progeny reverting to stamen production,
is unknown to undergo sex reversal.
If one includes the exception reported by Hofmeyr, the number of heritable forms
in papaya, some phenotypically stable, others unstable, is 32, of which 2 are female,
15 are male, and 15 are hermaphrodite.
The sex of a papaya seedling cannot be determined until it produces flower buds
six to nine months after germination. Many methods have been proposed for separat­
ing young seedlings into their supposed sexes, but none has withstood the test of
biometrical analysis in controlled experiments.

Fruit size and shape

Papaya strains differ widely in size of fruit. Some strains have fruits less than 5 cm
in diameter and 50 g in weight. Other strains have fruits 50 cm or more in length and
10 kg or more in weight.

397
WILLIAM B. STOREY

Size preferences vary greatly among countries where the papaya is grown. Hawaiian
production is based upon the Solo variety which is typified by the pyriform fruit weigh­
ing 400-500 g borne by hermaphrodite trees. Its shape has become the trademark
of high quality, and its size allows for it to be packed and shipped with minimal bruis­
ing. South African preference is for the globular fruit weighing 1.25-2.50 kg which
is borne by the female tree of the variety Hortus Gold. In most other countries, large
size seems to be considered an attribute of desirability, and throughout most of Latin
America and the South Pacific islands, the weights of fruits grown for home use or
offered for sale are of the order of 2.5-6.0 kg.
Breeding for shape can be accomplished with relative ease, since this is correlated
with sex. If one wishes a strain with elongate or pyriform fruit, he can start with a
hermaphrodite tree of the desired type and cross it with a female tree having the other
traits desired. By the usual methods of sibmating and backcrossing the desired re­
combination of shape and other qualities can be achieved in a few generations in
hermaphroditic offspring. Multiple plantings of seedlings from self-pollinated her­
maphrodites, with thinning later as sex becomes known, is indicated if one wants his
plantation to produce the greatest possible number of fruits of this kind (table 2).

Table 2 Percentages of places with $ and § trees remaining in multiple plantings of ? <8> progenies of
C. papaya after reduction to a single tree favoring

percentages of $$ & per place after reduction

number of trees planted
per place ÇÇ 5?

1 33.33 66.67
2 11.11 88.89
3 3.70 96.30
4 1.23 98.77

Breeding for oval or round fruit can be done in the same manner, and the planting
of the type ultimately selected set out also with several plants to the space to be thinned
later, favoring the females (table 3).
Weight of fruits is determined genetically by multiple factors, and size of fruit in
terms of volume is highly correlated with it. Studies at the Hawaii Agricultural Ex­
periment Station have shown that the mean weight of the fruit of a hybrid lies at or
near the geometric mean of the weights of the parents rather than at the arithmetric
mean. With this knowledge, the breeder can choose the direction in which he wishes
to go toward attaining the desired size of fruit.
Many strains of papayas having otherwise desirable characteristics have ovarian
cavities which are deeply furrowed, making removal of the placenta and the seeds
difficult. Certain Solo strains are of this type. Some strains do exist which have un-

398
 PAPAYA

Table 3 Percentages of places with $ and $ or o trees remaining in multiple plantings of Ç x ^ or

? X <? progenies of C. papaya respectively after reduction to a single tree favoring $?.

percentages of $$ & ^ per place after reduction

number of trees planted
per place c?c? or

1 50.00 50.00
2 75.00 25.00
3 87.50 12.50
4 93.75 6.25

furrowed ovarian cavities resembling those of cantaloupes which allow for easy re­
moval of the placenta, leaving the surface with an attractive frosty appearance. Breed­
ing is being done to develop improved Solo strains with smooth, unfurrowed cavities.
Much progress has been made, but the hoped-for results have yet to be achieved.
Multiple genes appear to be involved in the genetic expression for this trait, also.

Flavour and odour

The flavour of most papaya strains is rather bland. Odour may vary from almost im­
perceptible to highly aromatic. Flavour is generally best when the fruit is fully ripe and
has a sugar content of 12-15% on the fresh-weight basis. The flavour and associated
odour of some strains of papaya are strong and musky, to the point where some per­
sons find them decidedly unpleasant.
Studies in Hawaii have shown muskiness to be due to the homozygous recessive
allele of a single gene, and so may be easily and quickly bred out of a line. In fact,
some of the most promising new types have come from a cross of Betty X Solo. Betty
is quite musky but high in sugar. Solo has the preferred size and shape. The new hy­
brids closely approach both attributes, and the unpleasant flavour and odour are gone.

Colour of flesh

Most papaya strains have chrome-yellow flesh, but red-fleshed strains are quite
common in some parts of the world, especially Latin America.
Hofmeyr (1938) showed that flesh colour is determined by a single pair of alleles,
R, yellow, and r, red, which is recessive to yellow.
About fifteen years ago, the writer set about to create a red-fleshed strain having
the size and shape of Solo, which has yellow flesh. The female of a red-fleshed strain
which was being grown in Florida was pollinated with pollen from a Solo hermaphro­
dite. The fruits of all F, plants had yellow flesh. By backcrossing the Fx to the red
parent and by selling some of the hermaphrodites, numerous red-fleshed hermaphrodit­
es were obtained. By inbreeding those that most closely approached Solo in shape, a

399
WILLIAM B. STOREY

red Solo strain was developed quite easily. Its has never become popular as a market
fruit, but it is still grown to some extent as a horticultural curiosity.
A point of interest in yellow-fleshed x red-fleshed crosses is that the yellow-fleshed
FI actually contains the biochemical chromogen for red colour but its expression is
blocked by lack of the activating enzyme.
It should be noted, also, that although yellow and red coloration are basic, each
ranges through a fairly wide band of shades due to the effects of modifying genes upon
their expression.

Precocity and stature

Two characters of great importance in papaya, in view of the short economic life
of a plantation, are earliness of bearing and height of initial fruiting.
Earliness of bearing, as reported by Nakasone and Storey (1955), is a factor of
number of nodes produced to the first flowering node. By way of illustration, the
variety Betty begins flowering at 25.4 i 2.25 weeks after it has produced 23.7 ± 0.51
nodes; Solo begins flowering at 32.8 ± 2.84 weeks at 49.2 ± 0.68 nodes; the Fx hybrid
between them flowers at 27.9 i 1.46 weeks at 32.1 ± 0.34 nodes.
Height of initial bearing depends upon the factors above to which is added the effect
of internode length. The internode length for Betty is 1.6 ± 0.11 cm; for Solo it is
2.95 ±0.13 cm; and for the hybrid it is 2.18 ± 0-06 cm.
When the characters are taken together, it is seen that Betty fruits initially in about
25 weeks at the 23rd node, and at 38 cm above the ground. Solo fruits initially in 33
weeks, at the 49 th node, which is 145 cm above the ground. The figures for the hy­
brid are 28 weeks, 32 nodes, and 71 cm above the ground.
The characters in question are determined quantitatively with the hybrid tending to
be intermediate between the parents.
Hofmeyr (1949) once mentioned a genetic dwarf, but there seems to be no published
record of its use in a breeding program.

Yield ofpapain-containing latex

In addition to being cultivated for the fruit as a source of food, the papaya tree is
grown also for its latex which contains a protein-digesting enzyme, papain. All parts
of the tree contain latex in an anastomosing canal system of cells under turgor pressure,
but the greatest amount and the easiest to collect is in the green fruit where it occurs
in an extensive canal system in the mesocarp of the ovary wall. After the fruit has reach­
ed maturity and begun to ripen, the latex and papain hydrolyze into reducing sugars,
and possibly other substances, and virtually none is to be found in the fully-ripened
fruit.
The latex is usually collected by catching it in a ceramic bowl after scoring the
fruit with a bone knife or similar sharp instrument. At first the latex is fluid and runs

400
 PAPAYA

down the scorings into the receptacle. In a few seconds, however, because of the drop
in turgor pressure it oozes from the wound and congeals on the surface of the fruit
from which it must be scraped.
Although papain is crystallized in pure form, the papain sold in the world market
actually is the dried latex with its complement of papain. Principal uses of papain are
for tenderizing meat, for clearing beer, for digesting putrefying tissue in gangrenous
wounds, and for exfoliative cytology for the detection of stomach cancer. Principal
producing countries are Ceylon, Tanzania, and Uganda. World production amounts
to about 275 metric tons a year. The United States of America is the chief importer
(Becker, 1958).
Attempts to breed varieties having higher contents of latex have come to naught.
The reason for this undoubtedly lies in the fact reported by Jones (1940) that the
amount of latex in a fruit is directly proportional to its size. The weight in all varieties
and strains investigated by Jones ranged from 0.7-1.0% of the weight of the fruit.
Dried latex weighs only 1.0 % of the fresh latex.
With no difference among varieties in terms of yields, breeding for higher-yielding
varieties within the species appears to be futile at this time. Some time a high-yield­
ing strain may be discovered or induced by one of the numerous methods available
to plant breeders. Since all species of Carica produce papain, the answer may lie in
interspecific hybridization. Until this is a fait accompli, however, the chief hope for
increasing papain yields lies in the development of strains of trees producing as large-
sized fruits as they can support, and spacing them in the planting for the highest pos­
sible production of fruit per unit of area. For this purpose, development of gyno-
dioecious strains is indicated so that every tree in the planting will be a fruit-produc­
ing tree.

What has been attained

1. A study initiated at the Hawaii Agricultural Station in 1936 to attempt to learn

the mode of sex determination soon showed that the male could be bred out of a
strain in the first generation simply by close-pollinating hermaphrodites.
This study was carried on primarily with Solo, a variety of unknown parentage
which originated in Barbados and was introduced into Hawaii in 1910. It soon became
the leading variety there, but, by the time the study was begun, it had all but lost its
identity for want of controlled pollination. Originally gynodioecious it had become
trioecious.
By 1938, the Hawaii Station was in the position both of advising seed producers on
the procedure for obtaining completely male-free seed and of furnishing seed to
growers desiring to make a start. The first planting produced about 50 % more fruit
to the acre than comparable plantings with fairly high percentages of males.
Selection of about 20 hermaphroditic trees on the basis of phenotype for develop­
ing inbred lines resulted ultimately in Inbred Line 5 which is the backbone of the

401
WILLIAM B. STOREY

Fig. 1 Plantation of Inbred Line 5 Solo papaya in Hawaii showing heavy bearing and absence of male
trees.

Hawaiian industry today. An acre of trees of this line produces about 120% more
fruit than the original stock did in 1936.
2. Since the sex of an individual seedling cannot be determined until it begins to
flower, knowledge of the genetics of sex determination is useful in that it does provide
the means of knowing what sex forms will appear in the progeny resulting from con­
trolled pollination and what their proportions will be. By planting several seedlings
to a place and applying the formula (pa + qb)n in which a is one sex type, b is the
other sex type, p is the segregation percentage of a, q is the segregation percentage of
b, and n is the number of plants to the place, one can thin to a single plant per place
later when the appearance of flowers makes the sexes known, favoring the preferred
sex type. Tables 2 and 3 show expectancies for representative pollinations. Three
seedlings to each place is considered to be the most efficient planting, since this leaves
12.5% males in the field to serve as pollinators in progenies resulting from female x
male pollinations, and 96.3% hermaphrodites in progenies of self-pollinated herma­
phrodites.

Dioecious varieties are grown in South Africa and most other countries, consequent­
ly the results expected from multiple plantings in which progenies have approximately
equal numbers of females and males are those given in table 2 (p. 398).
Consumer preference has developed in Hawaii for the pyriform fruit of the herma­
phroditic tree of Solo, consequently progenies are grown from hermaphroditic trees
only. Since the ratio of females to hermaphrodites is 1: 2, table 3 applies. A plantation
planted according to this plan is shown in fig. 1.
There has been a 150 per cent increase in fruit production per acre in Hawaii

402
 PAPAYA

since the adoption of controlled pollination for seed production in 1938 and of mul­
tiple planting in 1939, to which have been added good cultural practices and improved
methods of disease and pest control.
3. Gene behavior has been determined for ten contrasting pairs of phenotypic
characters. Most are unrelated to important horticultural characters, but two are of
interest.
The first is red-flesh color which Hofmeyr (1938) found to be recessive to yellow-
flesh color in monohybrid inheritance. Advantage has been taken of this knowledge in
Hawaii to develop a red-fleshed strain resembling Solo by crossing the female of a
red-fleshed Florida variety with the hermaphroditic form of Solo. This type has gained
some favor in the Hawaiian market.
The second is muskiness, an odor which many persons find disagreeable. This was
found to be due to a single dominant gene. This finding set the pattern for breeding
Solo to Betty, a variety with disagreeably musky fruit but which has desirable attribu­
tes in extremely early and low bearing and in high sugar content of the fruit.
4. Studies reported by Nakasone and Storey (1955) have led to the development of
low-bearing, semidwarf lines. Some of these now approach commercial grower ac­
ceptance in all respects excepting fruit size. The fruits tend to run to 900-1350 g in
weight instead of the preferred weight of 400-500 g. Since the more important aspects
of the inheritance of fruit size are known, it would seem to be just a matter of time
to achieve the objective.

Problems for the future

Some of the major problems which need to be solved by the papaya breeder are
enumerated below.
Mosaic resistance. A virus disease of papaya called papaya mosaic is the scourge of
the industry in a number of regions of the world. It has literally destroyed papaya
production on the plantation basis throughout the American tropics, and frequently
cuts short the life of home-grown trees to a single crop season or less.
Adsuar (1947) began investigating the disease in Puerto Rico in 1937. He suggested
that the diseases called bunchy top and die-back are different manifestations of the
same virus. He determined the green citrus aphid (Aphis spiraecola) to be the principal
if not the only vector in its transmission. One approach to control of the disease would
be by control or, preferably, eradication of the vector. A second approach would
be development of highly resistant or immune strains of trees. The attempts to pro­
duce hybrids between C. papaya and other species in Jamaica and Puerto Rico were
motivated by the need for resistance rather than control in the belief that many of
the wild species are highly resistant if not immune to the disease. Although nothing
of value to the industry seems to have materialized, this line of endeavor should be
pursued diligently.
Elimination of sterile hermophrodites. Despite the progress that has been made in

403
WILLIAM B. STOREY

t *

ièm?
• w?
S >

Fig. 2 Hermaphroditic papaya tree with a high degree of female sterility.

developing continuously fruitful lines, some lines continue to produce seasonally

female-sterile hermaphrodites (fig. 2). Such hermaphrodites need to be bred out
of lines entirely so that only trees which are capable of producing to full capacity are
planted in order to achieve maximum production from the acreage devoted to a planta­
tion.
Elimination of carpellody of stamens. The tendency for the stamens in hermaphro­
ditic flowers in certain strains to assume the structure and function of carpels is called
carpellody of stamens. In papaya it is a seasonal occurrence. It can result in high
percentages of'cat-faced', malformed, unmarketable fruits (fig. 3). It must be elimina­
ted if one wishes to obtain the highest possible income from a plantation.
Need for a homozygous hermaphrodite. A homozygous hermaphrodite, which hy-

404
 PAPAYA

Fig. 3 Hermaphroditic papaya tree with a high percentage of misshapen fruits caused by carpellody
of stamens.

pothetically would produce progenies consisting entirely of hermaphrodites when self-

pollinated or crossed with other hermaphrodites and with females, would obviate the
need for starting three to four times the number of seedlings required for planting a
given area with several trees to the place for thinning later to a single tree. Progenies of
this sort would effect economies in number of seedlings needed, planting and thin­
ning operations, fertilizer used, and water applied.
All known hermaphrodites are enforced sex heterozygotes because of the lethal
factor that kills the homozygous dominant embryo. Not clear at this time is whether
the lethal factor affects the zygote directly or whether death of the zygote is a second­
ary effect brought on by endosperms which fail to develop properly. If the latter,
tissue culture of dissected embryos suggests itself as an approach to the problem.

405
WILLIAM B. STOREY

Induction of polyploidy and various kinds of radiation might be approaches to the

desired end also.
Sex-linkage. Although several sex-linked vegetative characters are known, none is
close enough to the sex-determining locus to be of practical use. What is needed is a ju­
venile vegetative marker so closely linked with sex that it can be used to separate seed­
lings at a very early stage. This would obviate the need for growing three to four times
as many plants for six months before the unwanted sex can be eliminated as will
finally remain in the field.
Radiation treatments suggest themselves as an approach to the problem of inducing
sex-linked mutations.
Ovarian cavity. Great improvement has been made in fruit size and shape in the
past two decades, especially in standardizing inbred lines of the Solo papaya. A
remaining problem is development of fruits from hermaphroditic trees having ovarian
cavities that are circular in transverse section. Many strains have deeply-grooved,
star-shaped cavities which make seed removal difficult and spoil the appearance of the
interior ovary wall. Circular cavities permit easy removal of the seeds by taking the
placenta at one end and stripping it off gently. The surface of the flesh is left with an
undamaged, attractive, frosty appearance.
Early and low bearing. Excellent progress has been made along these lines in
Hawaii in recent years, and the work is continuing. These characters are important in
prolonging the economic life of a plantation and in reducing the time and labor, and
therefore, the expense involved in harvesting the fruit.
Inflorescence and fruit setting. Development of pistillate and hermaphroditic trees
having moderately long peduncles, i.e. 7.5-10.0 cm or more, which set only a single
fruit, is desirable to obviate overcrowding. Fruits that are too closely packed on the
trunk exert pressure on one another in the course of development. This results in a
high percentage of the fruit being flattened or otherwise misshapen, and, therefore,
unmarketable in places where grading standards are enforced.
Increase in yield of latex. The amount of latex and, consequently, of the enzyme
papain, is directly proportional to fruit size. The amount of latex that can be produced
to the acre depends therefore upon the weight of fruit reaching the firm-ripe stage, at
which the latex system begins to disintegrate. Fruit size, per se, can be self-defeating
for large fruits often fall prematurely from sheer weight and the inability of the pe­
duncle to withstand the strain.
Possibilities for breeding seem to exist in the induction of mutations by one means
or another or by successful hybridization with other species.
Genetic studies. Genetic studies of important economic characters should be con­
tinued and extended. Results to date, both from previous genetic studies and from
breeding for economic characters, indicate that failure to make more rapid progress
in approaching certain objectives may be the result of strong linkages in some in­
stances and of epistasis in others.

406
 PAPAYA

Addendum.
Subsequent to preparation and submission of this article, additional information relating to papaya
breeding has been published in ten papers making up the entire issue of Agronomia Tropical, Vol.
17, No. 4, October - December 1967. Included are: a complete taxonomie revision of the family
Caricaceae; a theory of the derivations of the unisexual flowers of Caricaceae; a report on inter­
specific and intergeneric hybridization in Caricaceae; some genetic and breeding aspects of papaya;
papaya breeding in Colombia; a study of virus inoculations; a report on insects effects on papaya; the
status of papaya breeding in Peru ; papaya breeding in Hawaii ; and a new concept of sex in papaya.

References

ADSUAR, J., 1947. Studies on virus diseases of papaya (Caricapapaya) in Puerto Rico. I Transmission
of papaya mosaic. II Transmission of papaya mosaic by green citrus aphid (Aphis spiraecola Patch).
Ill Progeny studies of papaya mosaic virus. Univ. of Puerto Rico Jour. Agr., 31: 248-264.
ALLAN, P., 1964. Pawpaws grown from cuttings - are more true to type - bear earlier - bear lower and
longer. Farming in So. Afr., 101: 1-6, (reprint).
ALONSO OLIVÉ, R. A., 1951. Observaciones sobre el cultivo y mejoramiento de la fruta bomba. Estac.
Exp. Agron. Cuba Bol. No. 67. 160 pp.
Aw ADA, M., 1958. Relationships of minimum temperature and growth rate with sex expression of
papaya plants (Carica papaya L.). Hawaii Agr. Exp. Sta. Tech. Bui. 38. 16 pp.
AwADA, M., and IKEDA, W. S., 1957. Effects of water and nitrogen application on composition,
growth, sugars in fruits, yield, and sex expression of the papaya plants (Carica papaya L.). Hawaii
Agr. Exp. Sta. Tech. Bui. 33. 16 pp.
BECKER, S., 1958. The production of papain - an agricultural industry for Tropical America. Econ.
Bot., 12: 62-79.
HAMILTON, R. A., 1954. A quantitative study of growth and fruiting in inbred and crossbred progenies
from two Solo papaya strains. Hawaii Agr. Exp. Sta. Tech. Bui. 20, 38 pp.
HOFMEYR, J. D. J., 1938. Genetical studies of Carica papaya L. I The inheritance and relation of sex
and certain plant characteristics. II Sex reversal and sex forms. So. Afr. Dept. Agr. and Sei. Sei. Bul.
No. 187. 64 pp.
HOFMEYR, J. D. J., 1939. Sex reversal in Carica Papaya L. So. Afr. Jour. Sei., 26: 286-287.
HOFMEYR, J. D. J., 1949. Inheritance of dwarfness - Carica papaya L. (abstract). So. Afr. Jour. Sei.,
45: 96-97.
HOROWITZ, S., 1954. Determinacion del sexo en Carica papaya L. Estructura hipotetica de los cromo-
somas sexuales. Agronomica Tropical, 3: 229-249.
JONES, W. W., STOREY, W. B., PARRIS, G. K. and HOLDAWAY, F. G., 1940. Papaya production in the
Hawaiian Islands. Hawaii Agr. Exp. Sta. Bui. 87.
LANGE, A. H., 1961. Factors affecting sex change in the flowers of Carica papaya L. Amer. Soc. Hort.
Sei. Proc., 77: 252-264.
NAKASONE, H. Y. and STOREY, W. B., 1955. Studies on the inheritance of fruiting height of Carica
papaya L. Amer. Soc. Hort. Sei. Proc., 66: 168-182.
STOREY, W. B., 1953. Genetics of the papaya. Jour. Hered., 44: 70-78.
STOREY, W. B., 1958. Modifications of sex expression in papaya. Hort. Adv., 2: 49-60.
WARMKE, H. E., CABANILLAS, E. and CRUZADO, H. J., 1953. A new interspecific hybrid in the genus
Carica. Amer. Soc. Hort. Sei. Proc., 64: 284-288.

407
PEPPER

Piper nigrum L.

P. W. F. DE WAARD AND A. C. ZEVEN

Royal Tropical Institute, Amsterdam, The Netherlands and Institute of Plant Breeding,
Wageningen, The Netherlands

Introduction

Pepper is one of the oldest and most important spice crops; it was already known to
early civilizations. The dried fruit of the vine (P. nigrum L.) constitutes the pepper of
commerce. In the 15th and 16th centuries it caused seafarers from West European
countries to search for other ways to Asia than the traditional route via the Mediter­
ranean, Red Sea and the Indian Ocean. At present, world production amounts to some
80,000 metric tons per year (Anon., 1965).
The crop is grown by small holders in several tropical countries. India, Sarawak
(East Malaysia) and Indonesia are the main producers, in descending order of impor­
tance, followed by Brazil as a country with a rising production. Madagascar, former
Indo-China and some other countries are of minor significance. On and around the
Malaysian Peninsula clones are grown with a production capacity of 5-8 tons/ha
of dry white pepper. In contrast, common Indian cultivars show a yield potential of
about 70 % lower.
In each of the major countries only one or a few selected clonal cultivars are plant­
ed. The resulting concentration of usually monoclonal areas entails the danger of
disastrous destruction by e.g. virulent pathogens as reported by Muller (1936) in
Indonesia and by Holliday (1961) in Sarawak.
Efforts to breed improved material date from comparatively recent times. This
work was mainly concentrated in India, where investigations were initiated in 1952
(Nambiar and Sayeed, 1962). In Puerto Rico investigations were started in 1953 by
Martin and Gregory (1962). In Indonesia Hasan Iljas (1960) published work of a
similar nature in which local highly productive clones had been used. In Sarawak
breeding work was initiated in 1962 with the object of creating cultivars which com­
bine existing production potential with resistance to foot root disease (De Waard, 1963).

Systematics

TAXONOMY AND SOME OTHER PIPER SPECIES

P. nigrum L. belongs to the Piperaceae of the Piperales which is one of the most

409
P. W. F. DE WAARD AND A. C. ZEVEN

Fig. 1 The pepper plant. (Photograph P. W. F. de Waard).

primitive branches originating from the Ranales. The family is dicotyledonous, but
the stem has characteristics lying intermediate between those of the dicotyledons and
those of the monocotyledons, i.e. the vascular bundles lie in two or more circles. This
was confirmed by anatomical work on P. nigrum in Sarawak (De Waard, 1967b).
The result is that any established graft develops slowly and in due course dies.
There are many Piper species. Counts range from 600 to 2900 (Koorders, 1908;
Trelease, 1930). They occur over widespread areas in the tropics. The most important
species for man are the following: (Atal and Ojha, 1965; Koorders, 1908; Martin and
Gregory, 1962; Melchior, 1964; Rutgers, 1949):
P. aduncum L., a soil conservant;
P. augustifolium Vahl., producing Folia Matica\
P. betle L., the betle vine; the leaves of this species are chewed, providing a stimulant
for the people;
P. cubeba L. f., the cubebe or tailed pepper, of which the augmented base ('tail') of
the ovary is used as a condiment;
P. guineense Schum., the guinea pepper;
P. longum L., (P. peepuloides Rox and P. retrofractum Vahl. (syn. P. officinarum C.
DC) parts of the unripe thick spike are used, as a condiment;
P. methysticum Forst, its root is used for a toxic soporific beverage;
P. nigrum L., the black pepper, which produces the pepper of commerce ;
P. ornatum N.E.Br., an ornamental plant.

CHROMOSOME NUMBER

The somatic number of chromosomes varies. For instance 2n = 48 (Sharma and

Bhattacharyya, 1959), 2n = 52 (Martin and Gregory, 1962) and 2n = 128 (Darling­

410
 PEPPER

ton and Wylie, 1955). This variation might be caused by polyploidization, but as the
number of chromosomes of only a few cultivars of P. nigrum and of only a few Piper
species have been determined it is not possible to give detailed data on the origin of
P. nigrum. Martin and Gregory (1962) found mainly bivalents and only a few quadri-
valents. This may point to an almost complete diploidization of an autopolyploid or
to allopolyploidization of the cultivars studied.
Sharma and Bhattacharyya (1959) suggested that the original basic number of
chromosome is 12; at present this varies from 12 to 16. The cultivars studied by Martin
and Gregory (1962) have x = 13. This variation in the number of chromosomes a-
mong the pepper cultivars is one cause of the weak F, plants often observed.

CENTRE OF DIVERSITY AND GEOGRAPHICAL DISTRIBUTION

Wild pepper plants grow in the shade of forest-trees on the slopes of mountains in
the Western Ghats, Malabar, S.W. India, at an altitude between 150 and 2400 m
(Gentry, 1955a). In early times it spread from this region to an area between 20 °N
and 20°S, within S.E. Asia, including other parts of South India, Burma, Thailand,
former Indo-China, Ceylon, the Philippines, Malaysia and Indonesia. Only relatively
recently was it introduced to parts of Africa (notably Madagascar, the Congo Re­
public, the Central African Republic, the Cameroons, Nigeria and Dahomey), the
Pacific Islands (the Carolines, Tahiti) and Central and South America (Puerto Rico,
Guadelupe, Jamaica and Brazil.)

DISTRIBUTION OF CLONES

Pepper cultivars were most probably spread by means of cuttings rather than by
seed, because fresh seed has a longevity of approximately seven days under normal
conditions. Rooted or unrooted cuttings, properly packed in moist sand, saw dust,
polyethylene etc., can be transported over large distances without much trouble.
Only a few of the best Indian clones travelled to localities outside the place of origin.
Marinet (1953) suggested that for this reason the cv. Bangka in Indonesia resembled
the cv. Kamchay in former Indo-China. It appears likely that the original introductions
were a mixture of clones, which would partly explain certain anomalies existing with
respect to differences in characteristics within the apparently similar cultivars. The
cv. Kuching in Sarawak was probably derived from only four parent plants of the cv.
Bangka. Indian cultivars were recently introduced into the Western Hemispere, whereas
the cv. Kuching has migrated to Fiji and the Carolines (Migvar, 1965).

Botanical notes

THE PLANT

The pepper vine is a perennial plant, which develops the following aerial parts :

411
P. W. F. DE WAARD AND A. C. ZEVEN

1. Terminal stems These are non-productive, orthotropic vinelike steins, divided

into 2-10 cm long internodes bearing adventitious roots on conspicuous nodes, which
cling to supports. These stems possess a monopodial growth habit.
2. Stolons or runners These are non-productive, thin, orthotropic vines with
10-25 cm internodes and inconspicuous nodes which have only a few roots; these
runners may develop on any stem node. After some time the runners bend over and
trail over the soil, where they may strike root.
3. Lateral branches These are plagiotropic fruit-bearing branches with a sym-
podial growth habit; these laterals show a periodical growth which is not found in
orthotropic stems.
The stems and branches bear alternate, shiny, dark green, ovate, thickish leaves.
Branches are reproduced from dorsal buds on the parent branch or by apparent apical
extension. Each bud is accompanied by a single leaf lateral to the bud. The raceme
invariably develops simultaneously on the current year's flush opposite a leaf. Two
successive primordials of a raceme are present within a single bud, giving rise to
abundant flowering.

FLORAL BIOLOGY

Most wild Piper species and some wild forms of Piper nigrum in the W. Ghats appear
to be dioecious (Koorders, 1908), whereas the majority of cultivars appear to be
monoecious (Hasan Iljas, 1960; Nanbiar and Sayeed, 1962; Martin and Gregory,
1962); for example, cvs. Kalluvalli and Bangka have perfect flowers. On the other
hand, the cv. Kuderavalli has hermaphrodite, female and male flowers, whereas the
cv. Uthirancotta appears to possess female organs only. However, Hasan Iljas (1960)
reports that stamens may be present in a rudimentary form, embedded in the tissue
below the surface. This would provide an explanation for the restricted hermaphrodi­
tism in some of the cultivars. Although female clones do apparently exist (Martin and
Gregory, 1962), floral characteristics may vary within homonymous cultivars. Male
plants appear to be rare, and they are easily recognized by their vigorous, vegetative
appearance (Hasan Iljas, 1960; Cobley, 1963; Koorders, 1908; Marinet, 1953). In
commercial cultivation a high ratio of hermaphrodite flowers is essential for high
potential production (Blacklock, 1954; Cramer, 1907).

MORPHOLOGY

The mature catkin varies from 5-2Ö cm in length, supporting 50-150 small sessile
flowers without perianth. Protruding stigmata each with one or two lateral stamen at the
most are arranged longitudinally in several weak spirals. Stigmata are usually situated
below one of the preceding stamen when a preponderance of perfect flowers is present.
Each stigma has three to five branches, 1 mm in length, and is located on a single-
celled ovary containing one ovule. The succulent papillae of 10 pt. in diameter are

412
 PEPPER

extremely sensitive to mechanical damage (Adanan, 1924; Hasan Iljas, 1960; Mar­
tin and Gregory, 1962).
A viscous condition indicates receptivity. The period of peak receptivity occurs
three to five days after emergence and it may be extended to some ten days, depending
on cultivar and environment.
The stamen pushes its way through the catkin tissue and appears as a white spheric­
al body on top of a short, thick filament. It is usually not more than 1 mm in length,
and is situated some 1 mm away from the stigma.
The fruit is a small indéhiscent berry with a diameter of about 4.5 mm (Kamerlingh,
1930).

DEVELOPMENT OF THE RACEME

The raceme begins to show positive geotropism several days after emergence. After
some 15 days when the immature raceme has increased in length flowers appear from
the basal portion onwards. A protogynic stage develops and may exist for five days.
Subsequently, stamens appear, usually first at the base of the spike; four or five days
later each stigma may be accompanied by one or two stamens. Development is fun­
damentally centripetal, but an irregular appearance or dominancy is frequently obser­
ved. Anther dehiscence within pairs is not simultaneous as a rule. Varietal characteri­
stics may partly control spike length, protogyny and staminal development; sometimes
the protogynic stage does not exist (Hasan Iljas, 1960). In India protogyny may be
extended over a period of seven to eight days (Anandan, 1924; Cobley, 1963). In
Puerto Rico three to eight days passed before anther dehiscence was observed (Martin
and Gregory, 1962). Protandry is rarely found.

POLLEN AND POLLEN PRODUCTION

Temperature and relative humidity appear partially to control the longitudinal

dehiscence of the pollen sac (Hasan Iljas, 1960; Martin and Gregory, 1962). Work in
Sarawak has indicated that opening usually takes place between 12.00 and 14.00 h on
days when a relative humidity of approx. 60% is attained at a temperature of 32 °C
and in conditions of bright sunshine (De Waard, 1967a). The mass of pollen may spill
freely over adjacent stigmas and other parts of the catkin.
Several investigators (Hasan Iljas, 1960; Martin and Gregory, 1962) reported that
pollen grains are small, the mean diameter being approximately 10 fx, irrespective of
cultivars.

The total amount of pollen per anther varies with the cultivar. In Indian cultivars
each spike yielded 500,000-700,000 pollen grains, 10 in diameter (Marinet, 1955);
other data suggest 100,000-300,000 pollen grains per spike (Martin and Gregory, 1962).
Erosive action, presumably caused by rainfall, spontaneous losses to the air, and re-

413
P. W. F. DE WAARD AND A. C. ZEVEN

Fig. 2 Abundant development of inflorescences

of a cv. Kuching vine. (Photograph P. W. F. de
Waard).

moval by mechanical agitation affect the rate of pollen drain from the anther (Martin
and Gregory, 1962) and the efficiency of pollination.
There is usually sufficient pollen available to fertilize all stigmas, if properly distri­
buted.
There is some uncertainty concerning the condition of the fresh pollen. In Sarawak
it was observed in the cv. Kuching that fresh pollen appeared in glutinous clusters
dispersable in water. In this medium, viability may be retained for a period of up to
three days (Menon, 1949). Accumulations of dew may cause the disintegration of pol­
len lumps (Martin and Gregory, 1962); drops collected from the racemes were found
to contain considerable quantities of pollen. Apparently water acted as the medium
for pollen distribution over the length of the spike. In contrast, Hasan Iljas (1960)
established the presence of dry, powderlike pollen in the cv. Bangka and suggested the
possibility of direct gravitational distribution.

NATURAL MODE OF POLLINATION

The protogynic development suggests cross-pollination. Available evidence is

inconclusive, but it points towards self-compatibility and self-pollination. The follow­
ing modes of pollination have been suggested.

414
 PEPPER

1. Insect pollination The flowers are not such as to presuppose pollination by in­
sects (Hasan Iljas, 1960). Wingless insects have occasionally been found on the race­
mes. These insects may therefore be potential pollinators (Martin and Gregory, 1962).
Similar observations were made in Sarawak. Insecticidal control measures during the
period of flowering may sometimes result in a poor yield (Marinet, 1953). However,
direct damage to the sensitive papillae may also be a causal factor.
2. Wind pollination The flower construction does not seem conducive to efficient
wind pollination. Tests showed that pollen transportation by wind is negligible
(Hasan Iljas, 1960). Moreover, natural cross-pollination of the female cv. Uthirancot-
ta gave a very poor fruit set. As insect visitation was not observed pollen from vines
1-15 m away may have affected this low fruit set (Martin and Gregory, 1962). Never­
theless, studies in Puerto Rico indicated that 32-64 % of the pollen on the raceme may
be dispersed to the air within 24 hours after exposure (Martin and Gregory, 1962).
Upon release the small grains may be subject to wind transportation.
Insect and wind pollination can evidently be considered to be accidental in the
majority of cases.
3. Geitonogamy A composite mode of self-pollination involves the combined effect
of rainwater (or dew drops), intermittent showers alternating with prolonged periods
of sun and wind, each promoting fertilisation. Dispersed pollen grains move along the
spike by gravity, thus causing geitonogamy.
Heavy and driving rains have an adverse effect on fertilization. Similarly, prolonged
droughts during flowering frequently result in low yields (Anandan, 1924; Govinda
and Venkateswaran, 1929; Marinet, 1953). In Indonesia, Hasan Iljas (1960) reported
the occurrence of geitonogamy, but in this case it was due to the action of gravity on
dry pollen. Free-hanging racemes inside polyethylene isolation bags displayed an
unrestricted fruit set, irrespective of insects or rain water. Positive geotropism, a
spiral arrangement of the flowers, a sequential ripening of the stigmas and a non-
chronological dehiscence of anthers stimulate geitonogamic fertilization in general.
An increase in relative humidity extended the period of stigma receptivity.
4. Autogamy At the time of dehiscence stigma branches may have curved very
close to the fresh clusters of pollen. Thus 'true' self-pollination can be effected in
hermaphrodite cultivars.
5. Pollination between neighbouring racemes Poor pollination may be expected
at the base of the flower spike. This may be attributed to a time difference of 9-11
days between stigma development and anther dehiscence. Towards the tip the ef­
fect of this difference decreases. No evidence for such variability has been found,
setting appearing at random over the entire length. Transportation of dispersed
pollen by gravity from neighbouring spikes may play an important role. The con­
sensus of opinion is that geitonogamy under the influence of gravity appears to
be the general rule in cultivars with a preponderance of hermaphrodite flowers.
Pollination may be assisted by autogamy, particularly when environmental conditions
promote the extended receptivity of the stigma. These conditions also entail pollen

415
P. W. F. DE WAARD AND A. C. ZEVEN

Fig. 3 Isolation cage. (Photograph by the courtesy of the Sarawak Department of Agriculture).

dispersion in water, which acts as the distributing agent. Pollen in drops moving
from overhead racemes on to lower spikes may assist pollination. Wind and insects
are of minor importance.

HAND POLLINATION

Little concrete information is available as to the techniques used for the artificial
pollination of pepper. Although breeding programmes have been in progress in
India since 1953 no details have been reported on the modes of hand pollination em­
ployed. Martin and Gregory (1962) briefly mention two different techniques. In one

416
 PEPPER

Fig. 4 Artificial pollination using a portion of the flower spike showing freshly opened anthers.
(Photograph by the courtesy of the Sarawak Department of Agriculture).

technique, ripe anthers were opened by means of a scalpel and the pollen was scooped
up and applied to the appropriate stigmas, but the efficiency of this method appeared
to be low. In the other technique donor and recipient spikes were brushed with a ca-
mel's-hair brush. Fertilization appeared successful. Emasculation was tried by means
of alcohol, hot water and excision ; but this may not be necessary in the case of pro­
longed periods of protogyny. Hasan Iljas (1960) suggested emasculation by employing
a suction pump. He stressed the need to use dwarfed plants for this method.
In Sarawak a method of hand pollination was developed which made use of the
extended period of protogyny in the cv. Kuching, the apparent preponderance of
self-pollination of hermaphrodite cultivars and the absence of efficient natural cross-
pollination (De Waard, 1967a). Prior to hand pollination all flower spikes present on
the receiving vine are removed to prevent geitonogamy between neighbouring spikes.
At three or more locations branches are selected, which exhibit actively growing
'apical' buds. After the removal of insects from the bud it is isolated in a bag of cheese
cloth stretched around a strong wire frame. Protracted isolation within waxed
paper bags tended to promote premature abscission of the spike. No other spikes
are allowed to develop apart from those selected for breeding work. As soon as stig­
mata appear on the proximal portion of the spike a number of these, usually two or

417
P. W. F. DE WAARD AND A. C. ZEVEN

three per spike, are appropriately marked. By this time the stigma is ready for pollina­
tion; stamens are absent at this stage.
From a father plant a spike is selected which possesses freshly opened anthers filled
with pollen. A portion of the spike bearing two or more of these ripe anthers is select­
ed, cut off and placed on the end of a long pin. Subsequently, the entire pollen cluster
is gently brought into contact with the young stigma. This method was found to be
quite successfull in 50 to 75 % of routine pollinations. After fruit set, the isolation bag
is removed.
Although in theory the method of isolation is not absolutely rain and air proof,
the chance of illegitimate cross-pollination is considered to be extremely small in the
cv. Kuching. Moreover, pollination by both visiting insects and wingless insects on
the spike is excluded by early isolation. Seed developing on hand pollinated stigmata
could therefore be accepted as true legitimate hybrids.

FRUIT SET AND RIPENING

The ovules on a raceme develop into three types (Martin and Gregory, 1962):
(1) completely developed fruit ; (2) underdeveloped fruit and (3) undeveloped ovules.
The ovules of underdeveloped fruits start growing but stop at a certain moment.
Martin and Gregory (1962) supposed that insect damage might be a cause.

418
 PEPPER

The undeveloped ovules are probably a result of the absence of fertilization. This
may be due to the fact that pollination is unsufficient, the pollen is of poor quality, the
stigmas have lost their receptivity before the stamens shed the pollen and the stigmas
have been damaged (Martin and Gregory, 1962). These authors found 13.5% of the
stigmas to be undamaged, 72.6 % moderately damaged and 13.9% severely damaged.
The ripening of the fruits is uneven. The time elapsing between flowering and ripen­
ing ranges from five to nine months the average being seven months in India (Me-
non, 1949).

GENERATIVE PROPAGATION

Progenies of self- or cross-fertilized plants are often composed of weak seedlings

(Creech, 1955). This may point to the existence of parents with different ploidy levels
or to the presence of lethal genes. For this reason, propagation by seed is of limited
importance only.
The seed itself, when dried in the shade without pericarp remains viable for seven
days and when it is stored at 5 °C its life may be extended to two weeks. Removal of the
seed coat slightly accelerates germination (Anon., 1954), although it makes no differen­
ce whether the seed is sown immediately after collection or after three days drying.
In India, on the other hand, storage for 20 days did not appear to affect the viability.
In this latter case, the presence or absence of pericarp was not clearly established.
Chavancy et al. (1953) and Marinet (1955) gave details on the sowing of seed and
the time of transplanting. The biggest seeds appear to be the best for germination
(Gillot and Van Dingenen, 1960).
Gentry (1955b) reported a possible case of apomixis. He found fruits on the Indian
male sterile cv. Uthirancotta, without the presence of pollen sources in the neigh­
bourhood. This finding is not supported by Martin and Gregory (1962) and by the
results of investigation in India (Anon., 1956).

VEGETATIVE PROPAGATION

Grafts

Grafts would be very useful because pepper suffers from several root diseases and
high yielding cultivars could then be grafted on resistant rootstocks, e.g. in Puerto
Rico P. aduncum was found to be almost resistant to Phytophthora root rot (Ruppel
and Almeyda, 1965).
However, the stem anatomy of pepper (see p. 410) does not seem to allow grafting.
Any successful graft will die at a later stage ('delayed incompatibility') (Chavancy
et al., 1963; Nambiar and Sayeed, 1962).
It is possible to obtain coalescence between P. nigrum scions and rootstocks of
other Piper species. For instance Gregory et al. (1960) obtained grafts between

419
P. W. F. DE WAARD AND A. C. ZEVEN

P. nigrum scions and rootstocks of five American species. Hasan Iljas (1960) obtained
(dwarfed) grafts between P. nigrum scions and P. hirsutum and P. ariifolium rootstocks,
but the scions will die in due course. In Sarawak only initial coalescence was found be­
tween rootstock and scions, when grafting two portions of the main stem of the same
plant of the cv. Kuching. No permanent fusion was obtained between wood of
different varieties (De Waard, 1967c).

Cuttings

The use of cuttings (from a single node to seven nodes long) is valuable for the
quick multiplication of material for breeding and extension work, which can then be
tested in various places.
Numerous reports have been published on the farmer's and the reseacher's methods
of preparing a rooted cutting from stem parts (Gillot and Van Dingenen, 1960;
Greene, 1951 ; Hughes, 1966; Winter and Muzik, 1963 to mention a few authors).
Although all shoot types (see p. 412) can be used as cutting material the terminal
shoots are preferred. The best method is to use a part of the orthotropic terminal vine
with a plagiotropic lateral. Some plants obtained from fruit-bearing laterals develop
into bushy types (Anon., 1961; Gentry, 1955a; Larcher, 1966; Martin and Gregory,
1962) because they cannot climb.

Diseases and pests

There are several diseases and pests which kill the plants or damage the fruits. The
main one is Phytophthora foot rot, other diseases and pests being of lesser importan­
ce.
Phytophthora foot rot is caused by species and strains belonging to the genus
Phytophthora. Muller (1936) investigated this disease in Bangka and the Lampongs
(Indonesia) and isolated the pathogen which he claimed to be Ph. palmivora var.
piperis.
Holliday and Mowat (1963) working in Sarawak also found a Phytophthora species,
but they disagree with Muller as to the precise type.
For further information on diseases and pests the reader is advised to consult:
Barat (1952), Bregman (1940), Leacher (1967), Menon (1949), Ruppel and Almeyda
(1965), Rutgers (1949), and De Waard (1964).

Breeding work

INTRODUCTION

Planned hybridization has only recently started. Previously, only clone evaluation
with special reference to disease resistance and yield had been carried out, and some

420
 PEPPER

voluntary seedling plants originating from a selfing or an illegitimate cross were ob­
served (Gentry, 1955a; Gregory et al., 1960; Lim, 1961 ; Marinet, 1953; Menon, 1949;
Muller, 1936).
The desirable properties which a cultivar should possess are listed below (Cobley,
1963; Menon, 1949; Muller, 1936):
1. no spikes or only a few produced before the plant is mature,
2. precocity,
3. regular yielding,
4. even ripening,
5. vigorous growth,
6. resistance to diseases and pests,
7. possession of abundant spikes,
8. possession of long spikes,
9. possession of many rows of flowers per spike,
10. relatively close setting of flowers within the row,
11. high ratio of hermaphrodite flowers to total number of flowers,
12. high fruit set,
13. production of berries of large size,
14. production of seed of large size,
15. high content of alkaloids and non-volatile ether extract.

The late start of 'organized' hybridization may be largely attributed to the existence
of plant-to-plant variability within cultivars. In India, a formal breeding programme
was undertaken in 1953 with the objectives of obtaining a cultivar of superior yielding
capacity with good resistance to diseases and pests. Some of the work done in this
respect in India, Indonesia and Sarawak is discussed in the following sections.

LIVING COLLECTION

In India and Indonesia particularly a considerable number of varieties has been

collected. In India, there are some 64 different cultivars of local origin, three exotics
and one P. attaneatum. Four cultivars appeared to be outstanding with a production
of 4 kg of fresh berries per plant per year; the exotics did not perform well. It is inte­
resting to note that some of the local cultivars yielded some 2 kg of fresh fruits per
vine, whereas in Sarawak the same cultivars yielded as much as 5-9 kg. Their inherent
yield potential is obviously much higher than that suggested by the Indian results.
In Indonesia some 13 local cultivars and wild pepper have been included in the
collection, but little information is available as to their performance (Hasan Ujas,
1960). Reports from Sarawak indicate that the cv. Bangka is able to yield 18-27 kg
of green berries. The cv. Belantung and Djambi produce only slightly less, but tend
to mature later (De Waard, 1963-1965).
The cv. Kuching shows a superior productivity, similar to that of the cv. Bangka.

421
P. W. F. DE WAARD AND A. C. ZEVEN

These two cultivars appear to be a valuable genetical source for improving the yield
of otherwise satisfactory cultivars.
Rapid growth of fruit, particularly in countries with a brief period of dry weather,
the production of large berries and hermaphrodite flowers are desirable characteristics.
Reports from India (Anon., 1964), Puerto Rico (Martin and Gregory, 1962) and
Sarawak (De Waard, 1967a) suggest that these latter two characteristics are genet­
ically controlled. Berry weight and size and the rate of fruit development appear tobe
superior for cvs. Balamcotta and Uthirancotta, but the size of the corns is small for
the latter and large for the former cultivar. Thus, large berries are not necessarily
associated with large corns. Hermaphroditism varies from cultivar to cultivar and
determines productivity to a large extent.

VINE PRODUCTIVITY

As the traditional cultivars in India frequently do not meet the production standards
demanded by a competitive world market, breeding for yield improvement seemed
essential. The first results appear to be somewhat disappointing but are not unex­
pected. Only 40% of the germinated FT plants survived; this ratio varied from
cross to cross and from year to year. Moreover, only 7 % of the surviving hybrids
yielded more than 0.5 kg of fresh berries seven years after planting which is of the
same order as the parental productivity.
Only two specimen in all the Fj plants of one cross cv. Uthirancotta X cv. Talipa-
ramba made in 1955, yielded 4.5-5.5 kg of berries, which is a larger yield than that
of the best parent. One plant of the F1 of cv. Uthirancotta X Kottanadan gave 3.5 kg
of berries. None of the other Fj-plants yielded more than 1.2 kg of berries in the fol­
lowing years. It was noted that the higher production of the few individuals within
the progenies coincided with highest sexual ratio.
When these yields are compared with a production of 18-21 kg of berries per vine
for cvs. Kuching and Bangka and with the higher production of Indian cultivars under
different ecological conditions it can be concluded that considerable progress may be
expected in India from breeding.

From this discussion it may be concluded that:

1. Fj plants as a rule possess a high plant-to-plant variability with respect to yield.
2. Comparatively speaking, the cv. Uthirancotta appears to be the most promising
mother cultivar within the Indian group of cultivars, if relatively highly productive
father plants are used in this respect; cvs. Kuching or Bangka, Arikottanadan, Kum-
bakhodi and Kutharavally A.R.S. may be introduced as breeding material.
3. The absence of knowledge on the genetics of this crop makes its breeding a ha­
zardous operation.

422
 PEPPER

RESISTANCE TO DISEASES AND PESTS

The cv. Kuching in Sarawak yields 18-22 kg of berries per vine per year and thus,
compares favourably with Indian cultivars. On the other hand it is highly susceptible
to the endemic and virulent 'foot rot' disease caused by a species of Phytophthora
(Holliday and Mowat, 1963). Field control is rather elaborate and on the whole, has
failed to be accepted by farmers. Similarly, grafting on resistant or tolerant rootstocks
was unsuccessful and disease 'escapes' did not exhibit any degree of inherent resistance.
In spite of the difficulties expected to be encountered, a hybridization programme was
initiated in Sarawak with the aim of introducing a resistance gene(s) into cv. Kuch­
ing. For this purpose the method of repeated backcrossing was adopted. The need for
long term research was envisaged owing to: (1) the poor survival of the F, progenies;
(2) the necessity of fieldtesting of productivity of late-maturing plants over a period
of several years and (3) lack of information available on the inheritance of the disease
(De Waard, 1963-1965).
Unfortunately, no highly resistant cultivars appeared to be present among the
available cultivars and wild species in Sarawak (Holliday, 1961 ; Turner, 1965).
The programme, which started in 1963 has not progressed very far as yet. Legitima­
te p! seed obtained in 1964 and 1965 was germinated, the rate of germination being
above 60%. Observations made after transplanting indicated a large variability in
habitus, vigour and leaf colour from seedling to seedling. Slow terminal growth
appeared to be associated with a stunted elongation of the internodes.
When the rooted cuttings are large enough, they will be inoculated to test their
resistance and the clonal progenies of suitable resistant individuals will be tested in
the field. Early estimates of the performance and yield capacity of F, plants can be
made by assessing the habitus, flowering abundance and characteristics of the raceme,
especially its sexual ratio and pollen production.
This procedure allows time to be saved by back-crossing promising Fj individuals
early in the 1st year of production. First, second and higher order backcross progenies
are treated in a similar manner to that adopted for the F, plants.
In India, investigations on plants which are resistant to 'wilt' are in progress (Nam-
biar and Sayeed, 1962), although the causal vector of this disease has not been men­
tioned. In this field a threefold approach is being made :
(1) selection of 'wilt' resistant cultivars; (2) breeding of 'wilt' resistant strains;
(3) grafting of sensitive cultivars on to resistant rootstocks.

Field studies with respect to (1) and (2) are in the initial stages, whilst work on (3)
has so far indicated that grafting appears to be of no significance, even if at all pos­
sible.
Hybridization for resistance to insect damage may be a possibility. Variability has
been observed in resistance to the flea beetle, Longitarsus nigripennis which causes
hollow berry (Menon, 1949). Similarly it has been shown that cvs. Kuching and Bangka

423
P. W. F. DE WAARD AND A. C. ZEVEN

were highly resistant to stem borers, in contrast to the poor resistance of neighbouring
Indian cultivars.
Both insect pests can be controlled satisfactorily by means of carefully planned
insecticide spraying schedules.
The information available so far gives rise to the following remarks :
1. The work on aspects of flowering and modes of pollination seems to have been
well-covered, although not exhaustively.
2. Only scanty data are available on methods of hand pollination.
3. Hybridization on this crop encounters many obstacles associated with hybrid
weakness. Parental selection based on obviously compatible numbers of chromosomes
may reduce adverse influence to some extent.
4. Information is lacking on the inheritance of even the most important production
components.

CLONAL SELECTION

The apparently much higher potential production of Indian cultivars, and the
existence of alternative cultivars which give a superior production, indicate scope
for selection and/or adoption of cultivation practices in such way as fully to utilize
the yield potential of these cultivars. In view of the lack of adequate information on
the genetics of pepper, much work need to be done on this subject.

Future breeding work

With regard to future breeding work the following procedure seems to be the most
appropriate.
Inventarization, description and determination of the number of chromosomes in
the clones facilitates the rejection of the synonyms and the correct identification of the
homonyms. High yielding clones and obviously compatible parent plants may be
selected from this material. Subsequently, good Fx material can be selected and the
resulting clones evaluated at various sites.

Note: Apparently succesful budgrafting is carried out in Sarawak so far without delayed incomp­
atibility, following research reported bij Garner (Exptl. Agric. 4 (1968): 187-192).

References

ANANDAN, M., 1924. Observations on the habit of the pepper vine with special reference of the repro­
ductive phase. Madras Agric. Dept. Yearbook: 49-69.
ANONYMOUS, 1954. Annual Report of the Indian Council of Agric. Res. for 1952-1953. 189 pp.
ANONYMOUS, 1956. Annual Report of the Indian Council of Agric. Res. for 1955-1956. Cyclostyled.
ANONYMOUS, 1961. Los Banos agronomists produce bushy black pepper. Coffee and Cacao J. 4: 45.

424
 PEPPER

ANONYMOUS, 1962. Ann. Progress Report for the year 1961-1962. Pepper Res. Sta. Panniyur, Tali-
paramba, India. Typescript.
ANONYMOUS, 1965. La multiplication du poivrier. Service Plantes à Épices et Huiles essentielles,
I.R.A.T. Cah. Agr. Prat. Pays Chauds 1: 41-42.
ATAL, C. K. and OJHA, J. N., 1965. Studies on the genus Piper IV. Long peppers of Indian commerce.
Econ. Bot. 19: 157-164.
BARAT, H., 1952. Étude sur le dépérissement des poivriers en Indochine. Arch. Rech. Agron., Cam­
bodge, Laos et Vietnam 13: 1-92.
BLACKLOCK, J. S., 1954. A short history of pepper culture with special reference to Sarawak. Trop.
Agr. Trin. 31: 40-56.
BREGMAN, A., 1940. De pepercultuur en -handel op Bangka (Pepper cultivation and pepper trade in
Bangka). Landbouw 16: 139-256.
CHAVANCY, A., LANFRANCHI, J. et GUINARD, A., 1953. Compte-rendu destravaux du Centre d'Expéri­
mentation Agronomique de Blao en 1950-1951-1952. Poivriers. Arch. Rech-Agron. Pastor. Viêt-
Nam. No. 19: 171-186.
COBLEY, L. S., 1963. An introduction to the botany of tropical crops. Longmans, London 357 pp.
CRAMER, P. J. S., 1907. De produktiviteit van peperranken (The yielding capacity of pepper vines).
Teysmannia 18: 343.
CREECH, J. L., 1955. Propagation of black pepper. Econ. Bot. 9: 233-242.
DARLINGTON, C. D. and WYLIE, A. P., 1955. Chromosome atlas of flowering plants. London 519 pp.
GENTRY, H. S., 1955a. Introducing biack-pepper into America. Econ. Bot. 9: 256-268.
GENTRY, H. S., 1955b. Apomixis in black pepper and jojoba? J. Hered. 46: 8.
GILLOT, J. et VAN DINGENEN, A., 1960. Observations sur la multiplication vegetative du poivrier (Piper
nigrum). Bull. Inform. INÉAC 9: 115-118.
GOVINDA, K. M. and VENKATESWARAN, P. A., 1929. Pepper cultivation on the West Coast. Agric.
Dept. Madras. Bull. no. 98.
GREENE, L., 1951. Abstracts of some articles pertaining to the cultivation of black pepper. USDA Off.
For. Agric. Relation Techn. Coll. Branch.
GREGORY, L. E., ALMEYDA, N. and THEIS, T., 1960. The black pepper research program in Puerto
Rico: Procs. of the Caribbean Region of the Amer. Soc. of Hortic. Sei. 8th Meeting: Puerto Rico,
May 29 - June 4, 1960: 64-65.
HASAN ILJAS, B., 1960. Beberapa tjatatan tentang biologi bunga lada (Some notes on the floral biology
of black pepper) (P. nigrum L).. Pemb. Balai Besar Penj. Pert. Bogor no. 157: 1-22. Cited from PI.
Br. Ab. 32: no. 1030, 1962.
HOLLIDAY, P., 1961. A root disease of black pepper in Sarawak. Rept. 6th Commonw. Mycol. Conf.
Kew 1960: 156-161.
HOLLIDAY, P. and MOWAT, W. P., 1963. Footrot of Piper nigrum L. (Phytophthora palmivora). Com­
monw. Mycol. Inst., Kew; Phytopath. Paper no. 5. 62 pp.
HUGHES, J., 1966. Vegetative propagation of pepper under mist spray. S. Pac. Bull. 16: 41-42, 56.
KAMERLING, Z., 1930. Bekende en merkwaardige Indische planten in woord en beeld I. De peper en
haar verwanten (Well-known and notable Indonesian plants, described and illustrated I. Pepper
and its relatives). Ind. Culturen 15: 382-391.
KOORDERS, S. H., 1908. Die Piperaceae von Java. Verhandelingen Kon. Akad. Wetenschappen,
Amsterdam 2e sectie 14 (4): 1-75.
LARCHER, J., 1966. Premier bilan de quatre années d'expérimentation pipéricole au Centre de Re­
cherches Agronomiques de Boukoko (RCA). Agron. Trop. 21: 615-631.
LEATHER, R. I., 1967. The occurrence of a phytophthora root and leaf disease of black pepper in
Jamaica. F.A.O. Plant Prot. Bull 15: 15-16.
LIM, J. J., 1961. Black pepper in Indonesia. Coffee and Cacao J. 4: 66.
MARINET, J., 1953. La culture du poivre dans l'Inde. Rapport de mission (14 janvier au 7 février 1953).

425
P. W. F. DE WAARD AND A. C. ZEVEN

Agron. Trop., Nogent. 8: 482-505.

MARINET, J., 1955. Étude économique et culturale du poivre du Cambodge. Agron. Trop. 10: 279-
360.
MARTIN, F. W. and GREGORY, L. E., 1962. Mode of pollination and factors affecting fruit set in Piper
nigrum L. Crop Sei. 2: 295-299.
MELCHIOR, H., 1964. 17 Reihe Piperales. In A. Engler's Syllabus der Pflanzenfamilien. II. Angiosper­
men. : 147-151.
MENON, K. KRISHNA, 1949. The survey of poilu and root disease of pepper. Indian J. Agric. Sei. 19:
89-136.
MIGVAR, L., 1965. Black pepper in the Caroline Islands. Agric. Est. Bull., 4, Div. of Agriculture
Saipan.
MULLER, H. R. A., 1936. Het Phytophthora-voetrot van peper (Piper nigrum L.) in Nederlandsch-
Indië (The Phytophthora foot rot of pepper in Netherlands East Indies). Med. Inst. Plantenz. 88:
1-73.
NAMBIAR, P. K. VENUGOPALAN and SAYEED, P. M., 1962. Progress report for the year 1961-1662.
Pepper Res. Scheme, Pepper Res. Sta., Pannigur, Taliparamba, Kerala, India. Typescript.
RUPPEL, E. G. and ALMEYDA, N., 1965. Susceptibility of native Piper species to the collar rot pathogen
of black pepper in Puerto Rico. Plant Dis. Rpt. 49: 550-551.
RUTGERS, A. A. L., 1949. Peper (Pepper). In (Eds.) C. J. J. van Hall and C. van de Koppel. De land­
bouw in de Indische Archipel (Agriculture in the Indonesian Archipelego) IIB. Genotmiddelen en
specerijen (IIB. Stimulants and spices) : 620-654. Van Hoeve, 's Gravenhage.
SHARMA, A. K. and BHATTACHARYYA, N. K., 1959. Chromosome studies on two genera of the family
Piperaceae. Genetica 29: 256-289.
TRELEASE, W., 1930. The geography of American peppers. Proc. Amer. Philos. Soc. 69: 309-327.
TURNER, G. J., 1965. Personal communications with P. F. W. de Waard
DE WAARD, P. W. F., 1963-1965. Pepper Annual reports 1963, 1964, 1965. Department of Agric.
Sarawak.
DE WAARD, P. W. F., 1964. Pepper cultivation in Sarawak. World Crops 16: 24-34.
DE WAARD, P. W. F., 1967a. A method of artificial pollination in pepper cv. Kuching. Unpublished.
DE WAARD, P. W. F., 1967b. Some notes of the anatomy of the stem of P. nigrum. Unpublished.
DE WAARD, P. W. F., 1967c. Efforts of grafting pepper. Unpublished.
WINTERS, H. F. and MUZIK, T. G., 1963. Rooting and growth of fruiting branches of black pepper.
Trop. Agric., Trin. 40: 247-252.

426
RUBBER

Hevea brasiliensis (Wild.) Müll. Arg.

F. P. FERWERDA

Institute of Plant Breeding, Wageningen, The Netherlands

Systematics

TAXONOMY

The genus Hevea is a member of the family Euphorbiaceae which also includes other
important tropical crop plants such as Ricinus (castor oil), Manihot (cassava) and
Aleurites (tung oil tree).
The genus Hevea according to older taxonomists (Pax and Hoffmann, 1931) com­
prises some 20 species all native to the Amazon basin, morphologically not very clearly
defined and readily intercrossing. A considerably smaller number of species is mention­
ed by other botanists (La Rue, 1926; Ducke, 1939). Baldwin (1947), on the strength of
an intensive survey of the wild material in the Amazon basin combined with cytologic-
al observations, arrived at the conclusion that the genus Hevea should be reduced to
nine clearly distinguishable species, of which H. brasiliensis, H. spruceana, H. bent-
hamiana, H. pauciflora and H. rigidifolia are worth mentioning. H. brasiliensis (Wild.)
Müll. Arg. is the only species of importance as a source of rubber. The value of the
other ones either lies in their resistance to Dothidella leaf blight or in their suitability
as a root stock for H. brasiliensis.

CHROMOSOME NUMBERS IN RELATION TO TAXONOMY

The somatic chromosome number of the main Hevea species was determined by
various research workers (Ramaer, 1935; Baldwin, 1947; Bouharmont, 1960) to be
2n = 36. In one clone of H. guianensis 2n = 54 was found and in a race of H. pauci­
flora 2n = 18. It was postulated (Bouharmont, 1960) that Hevea has an amphidiploid
origin, being derived from the cross between two primitive species possessing similar
genomes with 2n =18.

427
F. P. FERWERDA

Crop physiological data

HEVEA BRASILIENSIS - GENERAL

Hevea brasiliensis is a tree which when allowed to grow undisturbed may attain a
height of 20-30 m. It develops a straight smooth trunk with rather slender branches
forming a fairly open crown. Rubber trees are evergreens which shed their foliage
every year at a distinct time varying according to the climatic and edaphic conditions
of the region where they grow. Almost immediately afterwards the shed leaves are
replaced by new ones. Along with the new leaves the cymose inflorescences are produc­
ed mainly at the end of branches. Under the climatic conditions prevailing in the low­
lands of the South-East Asian tropics rubber plants start flowering in their fourth or
fifth year of life. Some more details about the structure of inflorescence and flower are
given in the section on floral biology (p. 431).
The arbitrary girth used as a criterion for tappability - 45 cm at a height of 100 cm
above the root collar - is attained as a rule in the fifth or sixth year of life in the main
rubber-growing districts of South-East Asia. At an age of approximately 15 years rub­
ber trees are in their prime as far as yielding capacity is concerned. The age at which
they start declining is difficult to define exactly as it is strongly influenced by climate,
soil, planting density, rate of thinning, tapping system and cultivational conditions.
The economic life span may be put at approximately 25 years. Once this age has been
reached removal of the old stand and replanting the field with new improved material
can be considered.
Some solitary rubber trees in botanical gardens are reported to have reached an
age of 75 years or more.

VEGETATIVE PROPAGATION

Hevea can easily be multiplied vegetatively by means of bud-grafting. The method

commonly used is that developed in Indonesia by van Helten, Bodde and Tas, modified
by Forkert and described in detail in various planters' manuals (Edgar, 1960). In
recent years the method of green budding has come to the fore which makes use of
very young (three months old and upwards) rootstocks and budwood of the same age.
Of the other methods of vegetative propagation marcotting and rooting of soft­
wood cuttings have met with some success in the case of young seedlings but until
recently attempts in rooting cuttings of established clones failed. (Wiersum, 1955;
Mclndoe, 1958).
An improved method of rooting leafy softwood cuttings in a mist propagation fra­
me has been developed by Tinley (1960). This method proved to be particularly suc­
cessful in the case of cuttings of clonal material. Large differences in rooting ability
were observed among the various clones tried, the score ranging from zero to 94%,
averaging around 60 per cent.

428
 RUBBER

The propagants so obtained mostly lack a deep going tap root making them liable
to be overblown by wind. If methods could be developed by which this disadventageous
trait is avoided or eliminated then vegetatively propagated root stocks may provide
a suitable material for studying rootstock-scion interactions. In literature no mention
is made of experiments approaching the rootstock problem from this angle.

ROOTSTOCK-SCION RELATIONSHIPS

The interaction between scion and rootstock is of great importance in a physiologi­

cally dual organism like a Hevea budgraft. The problems concerning the stock-scion
relationships and the relevant investigations, mostly carried out in Indonesia and
Malaya before World War II, have been extensively reviewed by Dijkman (1951). In
this chapter only those features will be mentioned having a direct bearing upon bree­
ding.
The influence of the rootstock most clearly manifests itself in the variability in
yield and growth vigour observed within clones and in the descrepancies in yield be­
tween mother trees and the clones derived from them. This lack of conformity also
manifests itself when seedlings and their derivative JT-buddings (see p. 445) of about
the same age and growing in the same field are compared.
The relation between a seedling and its derivative clone is mainly determined by
two factors:
a. its response to the budgraft operation per se, and
b. its response to the forced symbiosis with an alien rootstock.
The first factor could be ruled out by an experiment carried out at the West Java
Experiment Station (Dijkman, 1951). In this experiment a number of individuals of
two seedling families were budded with their own stem-eyes and an equal number were
left unbudded to serve as checks. No significant differences were found between the
average yields of the budded plants in comparison to their unbudded partners. This
experiment made evident that in the case of auto-grafts the budding operation in it­
self has no influence on the yield of the budgraft.
Thus, the stock-scion relationship remains as the chief source of divergence between
mother seedlings and their derivative clones. In an experiment five seedling families
comprising a few hundred individuals were compared with their J. T. clones budded
on unselected and consequently, variable seedling rootstocks (Ferwerda, 1953;
Paardekooper, 1956). It was found that the buddings lagged behind the seedlings as
regards girth increase and yielding capacity. Averaged over the first three tapping
years the seedlings were found to outyield the J. T. buddings by 25-30%. An analysis
of individual tree yields revealed that in 76% of the cases the seedling mother trees
surpassed their J. T. -buddings. In the remaining 24% of the cases the J. T. buddings
yielded as much as, or more than, their mother trees. A slightly different arrangement
of the numerical data (tab. 1) reveals that J.T. clones equalling or surpassing their
mother trees are practically restricted to the two lowest producing seedling classes.

429
F. P. FERWERDA

Table 1 Percentage of J. T. clones giving the same or higher yields than their respective seedling
mother trees.

yield classes of the seedlings number of instances (in %) in which the buddings equalled
or surpassed their seedling mother trees

fam. 1 fam. 4

1 (lowest) 39.5 66
2 9.5 24
3 3 0
4 (highest) 0 0

Consequently there is, in the highest yielding groups, a much greater contrast between
seedlings and derivative J. T.-buddings than in the lowest production classes. From
these facts it has been concluded (Tollenaar, 1941) that clones originating from seed­
lings belonging to the highest yield groups are most depressed by the stock.
In this connexion experiments conducted at the West Java Experiment Station
(P. W. J., 1939) and the Rubber Research Institute of Malaya should be mentioned
(R.R.I.M., 1939), in which sets of twin seedlings were compared of which one mem­
ber was budded with a clone while its counterpart was allowed to develop normally.
These experiments showed that, within a clone, there exists a certain relation between
the growth vigour and the yielding capacity of a budgraft, and that of its supporting
rootstock. The highest yielding buddings proved in general to stand on stocks with
a high yielding potential. Such a relation is quite understandable because a vigou-
rously growing stock that is capable of producing a rich flow of latex can best satisfy
the heavy demands made by a clone of high yielding capacity. Taking this into ac­
count it is not difficult to understand that the chances for clones derived from superior
seedlings to find a congenial partner as rootstock are much smaller than for those ori­
ginating from medium or poor material. This seems to be one of the main reasons
why clones derived from high yielding mother trees so often gave disappointing re­
sults.
In the course of years several seedling strains, mostly illegitimate clonal seedlings,
have been tested as to their suitability as rootstocks for commercial clones in Indone­
sia and Malaya (Schmöle, 1940; R.R.I.M., 1939). A limited number of vigorously
growing and good yielding seedling strains were found to be best suited as rootstock
material for commercial clones. These findings led to an increased use of reputedly
good rootstock seedling strains during the last years preceding World War II and also
thereafter.
In one rootstock experiment conducted at the A.V.R.O.S. experiment station in
Sumatra seedlings arisen by spontaneous hybridization between Hevea spruceana and
H. brasiliensis were found to be outstanding as rootstock for H. brasiliensis. Three

430
 RUBBER

Fig. 1 Branchlet with expanding new leaves and developing inflorescences. (Photo Firestone Com­
pany, Liberia).

wellknown older brasiliensis clones A.V. 49, A.V. 50 and A.V. 256 were compared on
two rootstocks : illegitimate H. brasiliensis and hybrid spruceana seedlings. Growth,
girth increase, thickness of virgin and regenerated bark and number of latex vessel
cylinders in the virgin bark were considerably better in the budgrafts on hybrid
spruceana stocks than in those on brasiliensis stocks. Yields were recorded for 5
tapping years. The clones budded on hybrid H. spruceana stocks were found to pro­
duce up to 30 % more than those on H. brasiliensis stocks (Schmöle, 1941). These
experiments were disrupted by the war and have not been resumed afterwards.

Despite their fragmentary nature the results obtained show what can be achieved if
a clone is grafted on a congenial rootstock. Research along these lines fully deserves to
be continued. Particularly if it should prove possible to remove the drawbacks now
still inherent to vegetatively propagated rootstocks (see p. 429) there may be interest­
ing and attractive possibilities for further specialization. Ultimately ways may be
found to designate, for each clone, the most suitable type of rootstock.

Floral biology

INFLORESCENCE AND FLOWER

After the annual leaf-shedding (wintering) the monoecious cymose inflorescences

appear along with new flushes, mainly at the end of branches. They consist of a

431
F. P. FERWERDA

Fig. 2 Branch tip with inflorescences in full bloom. (Photo Firestone Company, Liberia).

main axis which carries about 12 pubescent branches on which the flowers are distribut­
ed in a cymose arrangement.
The small greenish white flowers are of two types, male and female, the latter being
larger than the male flowers and confined to the tips of the branches (fig. 1-3). Clones
may vary considerably in terms of the numerical proportion of male and female flo­
wers (George a.o., 1967).
The rather strongly scented rubber flowers have no petals but only a five-lobed
perianth tube splitting into five tapering segments curving back slightly at the tip.
The female flowers are easily distinguishable from the male buds by their larger
size and the rounder shape of the basal part (see fig. 4). The ovary is topped by a two
lobed sessile stigma. The male flowers are more slender than the female flowers;
they contain ten stamens arranged in two series of five lying above each other around
a central column.
An inflorescence matures over a period of one or two weeks. Male flower buds ex­
pand earlier than female ones; dehiscence generally takes place in the second half
of the morning and is practically completed by noon. For this reason, the flower buds

432
 RUBBER

st

, <>»-
X'

VV

' 'j
•y* **

-<v

e ^ •

A v
^2^

>v»

«H
Fig. 3 Close up of an inflorescence with male (B)
and female flowers (A), expanded and as buds.
(Photo Firestone Company, Liberia).

of the male parents in crosses should be collected early in the morning.

POLLEN TRANSFER, MATING SYSTEM, LONGEVITY OF POLLEN

The pollen is non-powdery and tends to stick together. It had already been observed
by early research workers (Maas, 1919) that small insects must be the chief agents in
pollen transfer. Later, positive evidence was obtained that midges (fam. Heleidae)
play an important part in transferring Hevea pollen (Warmke, 1951, 1952).
Hevea shows no preference for cross-pollination over self-pollination, although the
percentages of fruit setting obtained in artificial pollinations tend to be slightly lower
after selfing than after cross-pollination. Some clones are known to be distinctly
self-incompatible (Dijkman, 1951). Bouharmont (1962), on the other hand, found no
appreciable differences in the developmental process after fertilization and in the
definite percentage of setting in both forms of pollination. From these facts this
author concludes that the majority of seeds on a particular tree will generally be
derived from self-fertilization, because intraplant transfer occurs more readily than

433
F. P. FERWERDA

Fig. 4 (upper half). Details of inflorescence and flowers.

a.-b. Branchlet of an inflorescence with terminal female flowers and laterally inserted male flowers.
c. Three male flowers.
d.-e. Female flowers, d) just before anthesis, e) expanded.
f. Male flower partially dissected to show central column and the two series of stamens.
g. Female flower partially dissected. (After Heusser, 1919).
Fig. 4 (lower half). Mature fruits of Hevea, the upper one intact, the lower one dehisced sho­
wing the sections of the capsule with prominent woody endocarp and the three seeds
with characteristic testa pattern. (Photo Firestone Company, Liberia).

434
 RUBBER

interplant transfer. This has, amongst other things far-reaching implications for
biclonal seed orchards, which are supposed mainly to produce the cross between the
two clones involved but, in fact, may produce a large proportion of selfed seed unless
one of the partners is male sterile (c.f. p. 451).
The occurrence of male sterility in Hevea was reported for the first time by Ramaer
(1935), who ascribed it to irregularities in the meiosis of pollen mother cells. Majumdar
(1967), on the contrary, claims that normal tetrades and even apparently functional
microspores are observed in the male-sterile Hevea clones investigated by him. To all
appearances, degeneration of the microspores takes place during the pollen maturation
stages, just as in other more intensively investigated male sterile plants like tomato
and onion.
In practical breeding work various degrees of male sterility have been observed in
different clones ranging from reduced pollen production to complete absence of pol­
len (Dijkman, 1951). For that reason, some valuable clones can be used only as female
parents in crosses.
When stored without special precautions, Hevea pollen quickly loses its viability.
Dijkman (1938) succeeded in maintaining reasonable pollen viability for 17 days by
storing anther columns in a relative humidity of 67-80 % at a temperature of ± 6 °C.
Longevity studies conducted by Majumdar (1966) showed that a fair degree of viabili­
ty and in vitro germination percentages of not less than 20% could be maintained
for a week if pollen was stored at a temperature between 5 and 0°C and a relative
humidity of 75-81 %. Pollen stored this way was capable of bringing about fertiliza­
tion.

DEVELOPMENT OF FRUIT AND SEED

The Hevea fruit takes about five months to develop to maturity. During the develop­
ment period a large proportion of fruitlets are shed, especially during the first two
months after flowering. Ultimately, not more than 5 % of the initial number of female
flowers develop into a mature fruit in the case of artificial pollination. Under natural
conditions, the fruit-setting percentages are as a rule much lower. Premature fruit-
shedding in Hevea has many parallels with that in the pome fruits of temperate regions
(George a.o., 1967; Majumdar, unpublished data).
The Hevea fruit is a large distinctively three-lobed capsule which contains three
walnut-sized seeds surrounded by a thick testa having a characteristic pattern which is
different for every tree and is sometimes used as identification criterion for clones.
Seeds stored without special precautions will lose their germinative power within a
few days. If stored in tightly closed containers with damp charcoal powder of a hu­
midity in equilibrium with that within the seeds, viability may be maintained for up
to a month.

435
F. P. FERWERDA

TECHNIQUE OF ARTIFICIAL POLLINATION

The technique of artificial pollination was worked out by Maas (1919) and later
improved by other research workers (Morris, 1929; 's Jacob, 1931; Ehret, 1948). For
a detailed description of the pollination technique reference is made to Dijkman
(1951).
The method can be briefly described as follows :
In accordance with the time sequence in the development of the two sexes, inflorescences
of the male parent have to be collected early in the morning and kept fresh in a
container with some moist moss or fresh leaves.
The inflorescences to be pollinated should be selected carefully, those borne on
stunted and sparsely foliated branchlets being avoided. All flowers that have already
opened up or are still too young are trimmed away to ensure that not more than six
to eight female flowers are kept on each inflorescence as rarely more than two
fruits develop on an inflorescence. Practical experience (Dijkman, 1951) and statistical
analysis (Ross, 1960) have shown this to be approximately the optimal number of the
flowers to be pollinated. Only in exceptional cases more than two fruits develop from
one inflorescence.
The best stage for pollinating the pistillate flowers is when the perianth is bright
yellow and the spirally arranged tips of the perianth lobes are just parting.
In carrying out the pollination, the perianth lobes are gently pushed apart and the
staminal column of a male flower of the selected parent is inserted horizontally by
means of a pair of fine tweezers. It is kept in position by a small plug of cotton wool
which is fastened by folding the perianth tips over it like a cross band and sealing
them together with a drop of latex. The artificial pollination of rubber, which entails
standing on tall bamboo ladders or on often, rickety bamboo scaffoldings built around
the trees, demands, in addition to the necessary technical skill, a good deal of physical
agility.
Three or four months after pollination the developing fruits are surrounded by
baskets of plaited bamboo or of chicken wire in order to prevent the loss of seeds
when the capsules dehisce.
The definite fruit-setting percentages obtained after artificial cross-pollination in
Indonesia and Malaysia average 5 %, the actual figure depending on the female parent
used and the prevailing external conditions (Dijkman, 1951; Baptist, 1953). Conside­
rably higher setting percentages of about 15%, probably due to more favourable cli­
matic and edaphic conditions, have been reported for Viet Nam (Ehret, 1948). Oidium
leaf disease, in particular, should be kept in check by regular fungicidal treatments if
good seed-setting is to be ensured.

436
 RUBBER

Improvement by breeding

VARIABILITY OF INITIAL MATERIAL

Hevea is one of those tropical crop plants which have found their widest distribu­
tion and attained their greatest economic importance outside their native habitat.
Around 1875 Wickham somewhat clandestinely collected a few thousand seeds of
well-developed Hevea brasiliensis trees in the Amazon area. These seeds were ger­
minated in Kew Gardens and the seedlings were transported to Ceylon and Malaya,
where they grew into adult trees. The seed of these trees was successively distributed
over the future rubber-producing areas in South-East Asia: Malaysia, Indonesia,
Indo-China and Ceylon. Other batches of Hevea seeds were introduced from Brazil
into the Far East on a few subsequent occasions but the results obtained with them
generally less satisfactory than those obtained with the Wickham material (Dijkman,
1951 ; Bouychou, 1956). The economic prospects being favourable, rubber cultivation
in South East Asia experienced a considerable expansion after 1910. This entailed,
at short notice, the large-scale propagation of material of a relatively limited prove­
nance.
At that time, foresighted people had already realized that seed from certain trees
excelling in vigour or yielding capacity was vastly preferable to any seed taken at
random. This principle was often acted on during the laying-out of seedling plantations
on the East Coast of Sumatra from 1910-1921.
The results obtained with mother tree seed collected according to different degrees
of precision have been compiled in table 2.

Table 2 Yield of dry rubber (in kg/ha) of various seedling areas on the East Coast of Sumatra
(according to Maas, 1948).

material year of planting yield (kg/ha)

Unselected seed before 1917 496

Mother tree seedlings 1917-1918 639
Seedlings grown from seed of
critically selected mother trees 1919-1921 704

Chittenden (1950) reported similar results for Malaysia. The progress obtained even
by this very rough mass selection indicates the presence of a considerable breeding
potential in this genotypically quite heterogeneous material. Full advantage of these
possibilities was not taken until after 1917, when methods of vegetative multiplication
had been developed by means of bud grafting. This mode of propagation was discovered
almost simultaneously in Indonesia by van Helten, Bodde and Tas in 1916 (Dijk-

437
F. P. FERWERDA

log. number of trees

2000
1500

500
400

300

200

100+
80T
60 '

10 12 14 J.16 18 20 22 24 26 28 3Σ 32 34 36 38 40 42 44_46 48

x ?X 3X
yield in cc. latex per day

Fig. 5 Frequency distribution of individual yields of 5000 unthinned, unselected rubber seedlings all
tapped at the same height and with the same length of tapping cut. x, 2 x and 3 x indicate the
classes of 1, 2 or 3 times the average yield. (Adapted from Maas, 1934).

man, 1951). In Malaysia Gough gave considerable stimulus to budgrafting.

The availability of an adequate method of vegetative propagation made it possible
to develop clones from certain prominent mother trees. This was one of the first suc­
cesses thanks to the joint efforts of researchers of the experimental stations and the
planters in the field. If we confine ourselves to Indonesia and Malaysia, the available
initial material covered an area of several hundreds of thousands of hectares almost
exclusively planted with unselected seedlings. Rubber seedlings show a great variabili­
ty which finds expression not only in morphological characteristics such as stem shape,
mode of branching, crown form, leaf colour and leaf shape and the colour pattern on
the seed coat, but also in yielding capacity, the most important factor from a practical
point of view. It is possible to form an idea of the variability of this factor by taking
the individual yield records of a large number of seedlings, which reveal that tree

438
 RUBBER

yields may range from practically nil to over 100 g of dry rubber per tapping. The
results of such an experiment have been compiled in fig. 5.
It may be concluded from this figure that the best 8 % of the trees accounted for
nearly 24 % of the total yield and 75 % of this represented the capacity of 49 % of the
trees. In other words, on average the highest yielding category of trees produced well
over four times as much as the lowest yielding category. This variability may be
attributed partly to genotypical differences and partly to external conditions. A
comparison of the co-efficients of variability within seedling populations with those
within clones grown under comparable conditions provides us with enough evidence
to show that at least part of the variability in seedlings is due to genotypical differen­
ces. A clone is a group of individuals having the same genotype. Hence, the variability
within it may be attributed to :
a. the environmental conditions, and
b. the influence of the rootstock
Schmöle (1937) reported for some illegitimate clonal seedling progenies - i.e.
material that has already undergone some selection - variability co-efficients of 40 %
to 50% as against 16% - 22% for two clones. These figures are fairly well in accor­
dance with those mentioned by other research workers ; in the case of certain legitimate
seedling families coefficients of variability up to 70 or 80% have been reported (Ross,
1966). It is tempting to explain this discrepancy between seedlings and buddings
by assuming a large genotypic variability in the case of the former. Of course, more da­
ta and an appropriate biometrical analysis would be required to determine with greater
exactitude which part of the variability is genotypically founded and, consequently,
available for selection.

OUTLINE OF THE BREEDING PROCEDURE

The method followed in rubber breeding can be briefly characterized as a system in

which clonal selection and generative breeding alternate in regular succession. The
seedling progenies resulting from the latter treatment provide the initial material
from which the next generation of clones is developed (fig. 6). Mother trees selected
from the highly variable basic populations gave rise to an elementary group of clones
indicated as primary clones. The outstanding members of this group served a dual
purpose. They were used as planting material to establish commercial groves and, in
addition, were deliberately intercrossed in as many ways as possible with the aim of
obtaining seedling material of a considerably higher intrinsic value than the original
unselected seedlings.
These improved seedling families can be arbitrarily used either to establish commer­
cial groves or to serve as initial material for a new generation of clones indicated as
secondary clones. The latter material, in its turn, has by intercrossing produced a
further improved generation of seedling families with which the breeding procedure

439
F. P. FERWERDA

N i »
X
1 /
• • • • • -® • • • • •

• ® - • •
• ; • ®! • • •
IV
• • • v® • ; ®- • • • • • •

• • '• • • • / I &• • • ' •

1 • * • * •(

Fig. 6 Diagrammatic representation of one cycle of breeding and clonal selection in Hevea.
I. Initial seedling material from which prospective mother trees are selected (O) and others rejected
(/)•
II. Testing of primary clones (O selected, / rejected).
III. Performing crosses between outstanding clones.
IV. Evaluation of the Ft progenies and selection of secondary mother trees.
V. A. Testing of secondary clones.
B. Raising seed of the prominent cross D x E in quantity by means of an isolated bi-clonal seed
garden.
The breeding cycle is continued by repeating the procedures mentioned under III and IV.

can be repeated as indicated above and continued as long as there are prospects for
further progress.
In the most advanced breeding programmes three full breeding cycles have been
completed over a period of 40 to 50 years.
The successive stages of the breeding procedure are described in the following sec­
tions.

SELECTION OF MOTHER TREES AND PRIMARY CLONES

The chances of success for a clonal selection programme are smaller than one would
suppose. By no means all outstanding mother trees yield superior clones, a discrepan­
cy which may be ascribed to various factors.
Undoubtedly mother trees differ in the ability to bear the process of vegetative

440
 RUBBER

propagation. Furthermore, the root stocks on which the buddings are made constitute
an important but largely incalculable factor.
The following example may serve as an illustration of the chances of success in a
large-scale clonal selection programme (Maas, 1934; Ferwerda, 1940). After a critical
survey of an estate seedling area in Indonesia comprising approximately 15,000
hectares, a group of 3750 trees was singled out to serve as prospective mother trees. In
the selection of these prospective mother trees yield capacity (particularly the yield
ratio in comparison to the tapping task in which they stood) was an important but not
the governing criterion. In addition, such traits as tree habit, absence of a mode of
branching indicative of liability to wind damage, disease resistance, thickness, quality
and renewal rate of the bark carried considerable weight.
After continued observations over a period of three to five years, a large number of
the pre-selected trees were discarded, so that finally 260 remained with a yield ratio
of 470 in relation to their tapping task. These were raised to the rank of mother trees
and considered worth being evaluated as clones. For this purpose each of them was
propagated vegetatively by means of budgrafting. The clonal material so obtained
was tested in simple avenue test gardens with two replications, each clone being re­
presented by two tree rows of 20 individuals each. For five or more years the clones
were test-tapped and their yields carefully recorded. At the same time, critical obser­
vations of other important properties, such as general vigour, girth increase, incidence
of wind damage, bark thickness and rate of bark renewal were made. When at the
end of this observation period, the score was added up, only five out of the 260 clones
originally put under observation proved to have fully stood the test. Later, on the
strength of more comprehensive tests in larger-scale trials, two out of these five clones
were dropped, so that, ultimately, only three were maintained that found acceptance in
commercial practice.
Hence, roughly one in a thousand pre-selected mother trees and one in a hundred
finally chosen mother trees gave an approved clone. For estate - chosen candidate
mother trees the score is still lower; here the ratio of success seldom exceeds 1:5000
(Dijkman, 1951). These examples illustrate that clonal selection is not a short cut, as
has been presumed by some persons in the early days of rubber breeding. The outstand­
ing performance of a seedling mother tree offers no guarantee as to the behaviour
of the clone derived from it. This can only be determined by careful and time-consum­
ing tests. The development of a clone takes 10-12 years from the moment when the
mother tree is finally selected.
For a period of several years experimental stations and progressive enthusiastic
estate managers joined forces in developing clones directly derived from mother trees
in the original unselected seedling plantings and, consequently, indicated as primary
clones. These efforts resulted in a limited number of outstanding primary clones which,
one after another, found acceptance in practice.
The increase in yield obtained by using these primary clones was considerable.
Under the estate conditions prevailing in Indonesia and Malaysia, they reached a

441
F. P. FERWERDA

yield level two to three times as high as that of the old unselected seedlings, i.e. 1200
to 1500 kg of dry rubber per hectare per year at maturity (Maas, 1948). By the middle
thirties breeders in Indonesia and Malaysia had come to realize that the continued
search for primary clones might yield a few more clones of the same capacity as those
already existing but that material of such an origin was hardly likely to surpass the
level already reached.
It is interesting to quote one of the leading research workers in Indonesia who
stated that the old seedling material had already ceded all it possessed. In order to
reach a higher level it was necessary to create improved populations excelling over the
old material in genetic constitution. For this purpose crosses were made between the
best primary clones.

PRIMARY SEEDLING FAMILIES AND THE SECONDARY CLONES DERIVED FROM THEM

The idea of producing and studying the generative progenies of outstanding mother
trees or clones actually dates from much earlier, practically from the moment when
the first clones were being tested. Thus, the first phases of generative breeding and
clonal selection ran parallel for a number of years.
In Indonesia the first series of rubber crosses were performed as early as 1919 and
1920 by the A.V.R.O.S. Experimental Station, Medan (Sumatra). In Java this work
was initiated inl927, in Malaya in 1928, in Viet Nam in 1933 and in Ceylon in 1939
(Dijkman, 1951 ; Ehret, 1948). On page 436 a brief description of the crossing techni­
que has been given. The crosses were performed in a rather empirical way, particularly
in the early stages. Later on, when distinct clones had shown their merits as breeding
parents, a more systematic pattern was followed.
The F! progenies so obtained, mostly indicated as seedling families, were to serve a
dual purpose :
a. as initial material for developing secondary clones ;
b. as improved seedling material for practical purposes, in the case of those families
found to be really outstanding.
In order to assess their value, the various seedling families were planted in carefully
designed trials. The varying number of seeds obtained in the test crosses often permit­
ted no replications and the size of the test plots sometimes had to be adjusted to the
number of plants available. Whenever possible, an appropriate statistical design was
followed and, in any case, one or two wellknown, reliable clones were included as
checks. In addition to the legitimate seedling families obtained by hand-pollination,
the open-pollinated progenies (known as 'illegitimate' seedlings) of several outstand­
ing clones were also included in various trial gardens. After a few years of test tap­
ping and careful yield recording, it was possible to make an evaluation of the yielding
capacity of the various categories of seedlings.
The general impression was that notably the hand-pollinated seedling families
but some of the illegitimate clonal seedlings as well showed a good performance,

442
 RUBBER

equalling or even surpassing the best clones available at that time. The establishment
of this fact led to an increased demand for improved seedling material during the
period 1935-1942. In Indonesia this need was met at short notice by performing de­
sirable crosses by means of large-scale hand-pollination. The long-term needs of cer­
tain cross combinations were supplied by laying out biclonal seed gardens, which were
isolated either spacially or by laying them amidst other crop plants such as oil palms
or coffee. When it proved necessary to lay out a seed orchard near other rubber groves,
the section destined for seed production, consisting of alternating or staggered rows
of A and B clones, was surrounded by a protective belt of several rows of trees deep,
planted with buddings of one of the clonal partners, as described on p. 450.
However important this improved seedling material may have been in practice, its
primary function remained that of providing material for secundary clones.
It is possible to adopt one of the two following procedures in working with the thus
obtained cross material.

a) The traditional method of progeny testing

In the initial stages of breeding work the cross families were usually planted in sim­
ple blocks or rows. Later on, statistically designed experiments were preferred, pro­
vided that the quantity of available seed permitted this. One or two well-established
clones were also included as checks. When the tappable stage has been reached,
the seedling families plus the check clones are test-tapped in order to discover their
value.
By referring to individual yield records and carefully assessing other valuable pro­
perties such as growth vigour, mode of branching, bark thickness, bark renewal and
the properties of the latex, a limited number of outstanding trees are singled out to be
raised to the rank of mother trees and to be vegetatively propagated into clones. Hen­
ce, a comparatively small number of clones is prepared from the highest yielding and
otherwise desirable trees. These clones are preferably planted out in a statistically
designed trial and submitted to test tapping as soon as the trees achieve the arbitrary
size for tapping. Two to three years of test tapping suffice to give an idea as to which
of the clones are promising. These are increased vegetatively and planted in a statisti­
cally designed experiment on a rather larger scale more closely approximating the
conditions found in practice. Here the final evaluation is carried out, after which
planting on a practical scale may be considered.
This is a fairly long procedure. Counting from the moment the crosses and selfings
have been established, it takes about 16 years before it is possible to pass a provisional
judgement upon the clones derived. This period can be shortened by three to four
years if the individual yield recording is restricted to the first normal tapping year
and if, on the strength of these records, the 10-20% best producing seedlings are
selected and multiplied into clones. Most of the potential high yielders are found in
this group as has been shown by Dijkman and Ostendorf (1941). These investigators

443
F. P. FERWERDA

derived their conclusion mainly from the significant and rather high correlation co­
efficients-ranging from 0.75 to 0.85 - between the yield in the first year and the
accumulated yield over the first five tapping years.
The method just described consists in making a choice between mother trees in
anticipation that the best among them will yield good clones too. Experience in this
respect, just as in the selection of primary clones, has been that such an assumption is
not invariably valid. This fact has been confirmed and emphasized by the slight corre­
lation in yield that different investigators (Brookson, 1959) have found between seed­
lings and the clones derived from them. This method is not only time-consuming but
it has also the disadvantage that only part of the initial trees that are potential pro­
ducers of good clones can be traced.

b) The accelerated method of progeny testing

In order to overcome these objections and to speed up the process, the following
alternative method has been applied making use of the circumstance that a Hevea
plant can be multiplied by budgrafting already at an early age.
It is easy to remove 6-10 bud patches from an approximately one-year-old rubber
seedling and bud them on to root stocks of the proper age and size. In this way a
limited number of buddings (5-8) are obtained from each seedling, which, after having
been stumped, can be transplanted and grown in the ordinary way.
The small clones so established, together with the seedlings from which they are
derived, are planted out in a trial field according to an appropriate statistical design.
This procedure requires a much larger area, five to six times as large as would have
been required for planting only the seedlings, but it has the advantage that the seedling
families and their derivative bud grafts are tested concurrently, so that the information
about the two categories of material is obtained at the same time.

The data on the clones however have only a provisional character because they are
based on observations made on a small number of individuals. For that reason the
clones that, after one of two years or yield recording, look promising in this provisional
test are increased vegetatively and included in a larger-scale experiment designed
statistically and including the necessary check clones. The data thus obtained will
allow the experts to form a more definite opinion about these clones.
This alternative method differs essentially from the conventional method in that
it distinguishes between clones rather than between seedling mother trees. The gain
in time resulting from vegetative multiplication at an early age is considerable. It
is possible to make a preliminary evaluation of the secondary clones within 10 years
of carrying out a crossing programme as against 16 years if the traditional procedure is
followed.
Malaysia and Indonesia were the first countries where different institutions adopted
this accelerated testing programme. Unfortunately the conditions of World War II led

444
 RUBBER

to the loss of much material or delayed the compilation of data from the experiments
that were not destroyed.
Some remarks have to be made on the budgrafts which arose from stem buds taken
from young seedlings. As soon as the first groups of buddings of this type grew up it
was found that the conical shape of their stems was far more reminiscent of seedlings
than of budgrafts, since the latter are characterised by a cylindrical stem (Ferwerda,
1940, 1953; Paardekooper, 1956). The results of Mclndoe's (1958) investigations have
shown that the place on the seedlings from which the buds are taken determines the
outward appearance of the budgrafts obtained from them. Budpatches cut from the
lower part of the stems extending from the root collar to a height of about one metre
yielded buddings with a conical stem; those taken from the upper part of the stem
(above one metre) or from an orthotropic shoot developing on a stem resulted in
buddings with a cylindrical stem. Buddings showing characteristics similar to those of
their seedling mother trees were termed 'juvenile type' or J. T. buddings ; those ex­
hibiting the more nearly cylindrical shape of the normal budded tree were named
'mature type' or M. T. buddings.
Mc Indoe (1958) is correct in his criticism that these terms do not adequately indi­
cate the origin of the two types. They suggest juvenility and maturity, developmental
phases, in fact, as the causal agent, whereas Mc Indoe's trials are indicative of topo-
physis. No conclusive evidence has been obtained in this respect. Until more is known
about the anatomical or physiological causes of this phenomenon, it would be pre­
ferable in the opinion of the present author to adhere to the terms J. T. and M. T. bud­
dings which have found common acceptance.
As has been stated above, those among the secondary J. T. clones that show any
promise are increased vegetatively so that they can be retested on a larger scale.
Budwood cut from the crown of the tree is almost invariably used for this purpose,
the buddings obtained from them proving to be of the M. T. type. It had been queried
whether a clone which performs well as a J. T. budding will also yield good results as
an M. T. This doubt was removed by experiments carried out at the Rubber Research
Institute of Malaya (Brookson, 1959). Here, rather high and generally significant cor­
relations were found between the yield of J. T. buddings and that of the M. T. bud­
dings derived from them. Hence, a fair degree of correspondence may be expected to
exist between the selections based on a small-scale trial as 'juvenile' buddings and the
more reliable selections reached after large-scale tests as 'mature' buddings.
By the outbreak of World War II the data collected enabled a provisional judge­
ment to be made about some secondary clones. The results of the earliest uninter­
rupted series of breeding experiments conducted at the Medan A.V.R.O.S. Experi­
mental Station (East Coast of Sumatra) are here mentioned primarily to illustrate the
progress made (fig. 7). It is possible to conclude from this histogram that the mean
yield level of the primary clone A.V. 49, chosen to serve as a representative of the older
Sumatra clones, is more than twice that of the original unselected seedlings.
The yield of seedling families obtained by crosses between primary clones exceeded

445
F. P. FERWERDA

1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

Year of tapping

Fig. 7 Progress attained in two breeding cycles (Sumatra).

A = Unselected seedlings, B = primary clone A.V. 49,
C = average of primary seedling families, D = secondary clone A.V. 352.
—= average yield over 7 tapping years. (Adapted from: d'Angremond, 1935 and van Hell, 1950).

that of the unselected seedlings by 150% and clone A.V. 49 by about 25%. The yield
of A.V. 352, the representative of the secondary clones, surpassed that of the unselect­
ed seedlings by well over 400 % and that of the primary clone A.V. 49 by more than
100%. The above data were derived from trial plots in experimental gardens. An ana­
lysis of yield records taken from large commercial areas (Maas, 1948) showed roughly
the same progressive tendency for the different categories of planting material
(see p. 452).
The spread of World War II to the Pacific meant the suspension of breeding activi­
ties in the South-East Asian area. After the cessation of hostilities the work gradually
got under way again. Many experimental gardens proved to have been badly damaged
or had deteriorated owing to neglect. Malaysia suffered relatively the least damage, so
that work could be resumed in that country comparatively soon after the war. There­
fore, it is better to illustrate further progress in breeding by reference to results ob­
tained by the Rubber Research Institute of Malaya and shown in fig. 8. The average
yield in the first five tapping years of the clone Pilm. B 84, here chosen to represent the

446
 rubber

C D

E10- 500 •

1
j—. % 11
I1
6 - -<>

1
1
|1| 1
.E 4-
Fi 1
2 -
i 1 $
I
i1
12 3 4 5 12 3 4 5 1 2 3 4 5 1 2 3 u
Year of tapping

Fig. 8 Progress achieved in two breeding cycles (Malaysia).

A = unselected seedlings,
B = primary clone Pilm. B 84,
C = Legitimate seedling families from crosses between primary clones.
D = Average of secondary clones (R.R.I.M. 600-series).
— = Average yield over 5 tapping years. (After Baptist, 1953).

primary clones, was about twice as high as that of unselected seedlings under compara­
ble conditions. A series of seedling families obtained by intercrossing the best primary
clones reached a yield index of well over 300 in relation to the unselected seedlings.
The secondary clones derived from these crosses are referred to as the 'R.R.I.M. 600
series'. The 31 best clones from among these, the residue of a series of 3148 clones
originally under observation, reached a yield level that was more than four times as
high as that of the unselected seedlings and nearly twice as high as that of the good
primary clone Pilm. B 84. Continued testing resulted in the removal of some less
satisfactory clones from this 600 series, either because the further yield trend did not
quite come up to expectations or because the clones suffered from diseases of the tap­
ping panel, experienced unsatisfactory bark renewal or were subject to wind damage.
After ten tapping years only half a dozen clones had maintained their front-rank posi­
tion.

In many small trial plots scattered over a large number of estates the average yield
of this limited group of clones for the first seven years of tapping was approximately
1500 kg of dry rubber per ha per year (1340 lbs/acre per annum). This is over 30%
more than the average of the two renowned primary clones (PB 86 and PR 107) in­
cluded in this trial as standards. During the eight and ninth tapping year the yields
reached the incredibly high level of 2500-3000 kg ha, i.e. five to six times as much as
those of the unselected seedlings. From the tenth year tapping was continued on renew­
ed bark. Since that time the yields have fallen off markedly but, in some cases, seem

447
F. P. FERWERDA

to have resumed an upward trend from the eleventh tapping year onwards (R.R.I.M.,
PI. B. 1967,1968).
In evaluating these figures, it should be borne in mind that they have been collected
from several experimental plots, which, although they are distributed over a large
number of estates, are not quite representative of the conditions of large-scale planting.
Yields obtained in commercial practice and compiled from data collected on a large
number of estates in Malaysia are now available. During the first five tapping years
the lead of the prominent R.R.I.M. clones over the long established primary clone,
PR 107, ranged from 21 % to 62% with an average of 40%, this figure being in fairly
good agreement with that mentioned for the experimental plots (see page 447).
In terms of absolute figures the production of this outstanding group of secondary
clones has amounted to well over 1600 kg/ha over the first five tapping years. This is
about 20 % less than in the case of planting in small experimental plots but is nothing
out of the ordinary. Thus, the results of the second breeding cycle were certainly satis­
factory. Nevertheless, the hard-to-please breeders desired further progress. This en­
tailed the initiation of a third breeding cycle in which the best clones of the first and
second cycles served as parental material.

SECONDARY SEEDLING FAMILIES AND THE TERTIAIRY CLONES DERIVED FROM THEM

At the start of the third breeding cycle the best secondary clones were crossed with
each other and with certain primary clones which had emerged as particularly good
parent material from the first series of crossings. Critical analyses of the performance
of the cross combinations obtained in the second breeding cycle, as have been carried
out by Paardekooper (1956) for the crosses made in Java, and by Ross (1965) and
Ross and Brookson (1966) for the Malaysian hand-pollinated seedling families en­
abled a better evaluation of the breeding capacity of the various parents and a more
judicious choice in planning future breeding programmes.
Yield data for the resultant seedling families are so far available only for a limited
number of years (varying from two to five years). As far as can be judged from these
restricted data, the average yields are at about the same level as those obtained from
the crossing families in the second cycle.
It is rather difficult to compare the two cycles owing to the different tapping systems
adopted in either case.
There are some outstanding seedling families the yield of which is considerably
superior (25% to 50%) to the average for the crossing group to which they belong.
A striking feature is the fact that variability coefficients in relation to yield are still
fairly high, varying from 37 % to 88 % for the various families, the average being 61 %.
Stem girth and bark thickness display a remarkably small variability of 22 % and 18 %,
respectively (Ross, 1965; Ross and Brookson, 1966). The fairly high variability coef­
ficient for yield, which is not much less than that found in illegitimate mother tree
seedlings, demonstrates that a certain amount of residual genotypic variability, i.e.

448
 rubber

potential for further improvement, still exists. However, it still remains to be seen
whether this potential can be utilised for selection purposes.
The yield figures for two years are now available for the tertiary clones (known as
R.R.I.M. 700 clones) derived from the seedling families of the third cycle. The initial
impression is that the best of these new clones are of the same production level as the
foremost of the 600 series. However, no final assessment can be made until these
tertiary clones have been in production a little longer (R.R.I.M., Pl. B. 1967).
Despite the high variability coefficient in relation to yield found in the secondary
seedling families, there are various indications to show that the limit has now been
reached for material of this origin. Various families subjected to the S2/d2 100%
tapping system, the system used hitherto, contained fairly high percentages of dry
trees, particularly in those tapping years during which the panel was changed. This led
to the introduction of a less intensive tapping system (S2/d3 67%), which certainly
gave some reduction in the percentage of dry trees but by no means entirely eliminated
the problem. In the families which from the outset had been tapped according to the
less intensive tapping system (S2/d3 67 %) the percentage of dry trees remained within
reasonable bounds.
The later crosses, for which a more efficient parental selection has been applied
based on previous experience, may provide a more encouraging picture. These younger
crossing families have either not yet been tapped or have been in tap for such a short
period that it will be some years before any judgement can be made of their merits.
The incidence of dry trees amongst the secondary seedling families already in tap is
a symptom which cannot be ignored, since it is evidence of the fact that as far as
yielding capacity is concerned the limit has now been reached with this material. It
may perhaps be possible to obtain a further yield improvement by importing for
breeding purposes new parent material from the vast gene reservoir in the Amazon
area. As far as the secondary properties such as bark thickness, bark renewal, suscep­
tibility to wind damage are concerned, the material now available presumably still
offers sufficient potential for further improvement.

Relative merits of clones and improved seedling families

In this discussion of the course of the improvement of the rubber tree by breed­
ing the main emphasis has been on the clonal material. The principal reason for this is
that the most spectacular advances have been made in the field of clone selection not
only in terms of yield potential but also in respect of such properties as bark thickness,
bark renewal, vigour of growth and resistance to wind damage. Further important
advantages have been the uniformity, the absolute reproducibility and the identifia-
bility of the clonal material. This made it perfectly clear to the planter what was to be
expected if certain clones were used. The combined effect of these factors was a marked
preference for clonal material both in large estates and in progressive small holdings.
In consequence, seedling material, which has also been raised to a considerably higher

449
F. P. FERWERDA

O X O X O • + 0 m Y 0 •
O X O X O 0 • • + 0 m
O X O X O m 0 • • + 0
O X O X O + 0 m y} 0 A •
O X O X O • • + 0 m 0
O X O X O 0 • • + 0 •
O X O X O O u 0 • • +

Fig. 9 Layout of two types of isolated seed garden.

(A) Bi-clonal garden. (B) Multi-clonal garden, initially containing 7 clones of which the ones indicated
by a / are supposed to be discarded on the basis of continued test-cross progeny testing.
Wavy line = pollen barrier planted with the same rubber clones as in the actual seed garden, or with
an other tall growing crop plant like oil palm. A belt of forest may also serve as a screen against in­
truder pollen.

quality level as a result of breeding, has to some extent been neglected. Nevertheless,
the progress achieved in the improvement of seedling strains is so striking that it
certainly deserves to be mentioned.
As has already been stated earlier, it had been noticed in the very first stages of
systematic breeding work that certain cross-combinations and also the illegitimate
progeny of certain clones resulting from open pollination performed very well. On
the basis of this experience, various land development societies and small holders, in
addition to budgrafted material, have also planted considerable areas over the years
with selected seedlings. A great deal of the seed required for this purpose was obtained
from mixed clone plantations or was gathered from the border area between two ad­
jacent monoclone blocks containing different clones. As only the mother of this seed
was known in most cases, it was termed illegitimate.
Legitimate or near-legitimate seed was obtained from isolated seed orchards, where
two clones which, according to the result of test crossings produce an outstanding
Fj, were planted on a mixed basis.
Fig. 9A illustrates the lay-out of such a seed orchard. A second type of seed orchard
(Fig. 9B) contains a mixture of several clones, the composition of which has been pre­
ferably based on the results of test crosses but which may have been subsequently
corrected on the basis of further progeny testing by the elimination of those clones
which have later proved to be poor parent material.
Strictly speaking, completely legitimate crossing seed can only be produced in bi-

450
 RUBBER

clonal seed orchards where one of the partners is male sterile or by means of hand-
pollination. In most biclonal seed orchards a certain percentage of seed resulting
from selfing will be produced in addition to the desired combinations A x B or B X A.
This percentage varies according to the particular type of clone, the flowering fre­
quency, the weather conditions and the type of insect visiting (see p. 433). For safety's
sake, therefore, it is best to refer to seed from biclonal orchards as near-legitimate.
There is an interval of at least five years between the planting and the coming-into-
bearing of the seed orchard. In order to be able to supply legitimate seeds on a limited
scale during the waiting period, however, certain outstanding cross - combinations
have, in fact, been produced by means of artificial pollination. With an efficient or­
ganisation and an adequate supply of skilled labour this will certainly be feasible. To
ensure that this expensive seed is utilised as efficiently as possible, methods have been
developed for splitting the young seedlings into two, thus, theoretically, obtaining two
plants from one seed (Dijkman, 1951; Loomis, 1942). In practice, no more than 150
successful plants are generally obtained from 100 split seedlings.

Comparison of the results obtained in test plantations with those obtained in practice

The data collected from various test plantations in Java, Sumatra and Malaysia
have adequately demonstrated the value of the different categories of improved seed­
lings. The impressions based on observations made in relatively small test plantations
have been largely confirmed by the experience obtained in practice on large estates.
On the basis of an analysis of the yield data relating to large estate areas planted
with various categories of seedlings and clones on the east coast of Sumatra, Maas
(1948) came to the following conclusions, which are represented graphically in fig. 10.
From this it can be seen that the plantings of open-pollinated seed (d) attained in their
eighth tapping year yields of between 1200 and 1300 kg/ha, which is about the same
as those of the old A.V.R.O.S. clones (c) under comparable conditions, and appro­
ximately three times as high as those of the non-selected seedlings (a). Since insuffi­
cient practical data were yet available at that time with regard to the outstanding
legitimate crossing families, the expected production trend of this category has been
represented by a line of dashes (e). The data collected in Java (Greven and Yollema,
1950; Paardekooper 1956) in respect of the seedling material planted there give rise
to approximately the same conclusions.
Generally speaking, it is true to say that the results obtained in practice substantiate
the conclusions drawn from the results of the experiments conducted in the trial
gardens even if, as can be appreciated, the overall yield level in practice is somewhat
lower than that of the test plantations. Practical data have also been collected in Ma­
laya which illustrate that selected seedlings are capable of achieving a high performance.
Seedlings grown from seed obtained from a carefully composed polyclone seed or­
chard, when planted on a practical scale on a large number of estates distributed through­
out Malaya, gave yields averaging 25 % more for the first five tapping years than those

451
F. P. FERWERDA

Fig. 10 Yield trend of various categories of

seedlings in estate areas on the east coast of
Sumatra as compared with the average perfor­
mance of the oldest tested clones (c).
(a) Non-selected seedlings.
(b) Seedlings grown from seed of carefully select­
ed mother trees.
(c) Average of the oldest tested clones.
(d) Seedlings grown from illegitimate seed of the
older A.V.R.O.S. clones.
(e) Estimated yield trend of recommended seed­
ling families and tested clonal seedlings.
(f) Newer clones,
Year of tapping After Maas, 1948.

of the good primary clone P.R. 107 which had been used as a standard for comparison
purposes. Expressed in absolute terms, the average yield for this period was in the
region of 1200 kg/ha, which is only 10 % below the level reached by the best secondary
clones of the R.R.I.M. 600 series under practical conditions.
Selected seedlings are therefore certainly worth the attention devoted to them. The
esteem in which this modern seedling material is held is seen reflected in the percentage

Table 3 Percentage of clonal seedlings in the total rubber-planted area of Malaysia for the years
1946/1965 (taken from R.R.I.M. Planters' Bulletin 88, 1967).

1946/1961 1962/1965 1946/1965

(mature) (immature) (total)

clones 69.7% 80.7% 72.4%

clonal seedlings 30.3% 19.3% 27.6%

100.0% 100.0% 100.0%

452
 RUBBER

of new seedlings included in recently planted areas. The figures listed in table 3 il­
lustrate this.
The manner in which the relationship between seedling and budgraft areas will
develop in the future depends on so many factors that it is impossible to make any
firmly based prediction. One factor should not however be overlooked. Seedlings, by
virtue of their uncomplicated harmonic structure, their robustness, their variability,
which provides them with greater adaptability and reduces the risk of mass attack
by disease or pests, have positive advantages over buddings. Added to this is the fact
that by the skilful application of selective thinning the yield level of a seedling or­
chard can be raised considerably, a possibility present to a much smaller extent in
clonal complexes. All these factors in combination are such as to ensure that seedlings
continue to remain receiving attention.

BREEDING FOR DISEASE RESISTANCE

The Hevea rubber planted in large estate areas in South-East Asia and also in
Africa has remained comparatively free from serious diseases and pests. The various
forms of root fungi which can cause serious trouble must be regarded as a legacy of
the jungle that had to be removed to make way for the rubber plantations. Diseases
of the tapping panel can for the most part be regarded as wound infections and these
can be controlled by regular medication of the affected area. The only leaf disease
which gives real trouble in the rubber-producing areas of South-East Asia is powdery
mildew (Oidium heveae Steinm.), which can cause serious damage in humid districts
particularly during and immediately after wintering, but which, so far, can be ade­
quately kept under control by systematically spraying the trees with sulphur powder
or organic sulphur compounds. In recent years Phytophthora leaf fall is a matter of
increasing concern.
In seedling orchards trees are occasionally encountered which remain free from
Oidium even when no protective measures are applied, and in some cases the clones
derived from such trees likewise possess this property. With regard to one such clone,
L.C.B. 870, it was initially assumed that its ability to resist Hevea mildew stemmed
from the protracted nature of its leaf-shedding phase, as a result of which the clone is
never completely leafless and, in consequence, is also never completely occupied by
young susceptible leaf. This was subsequently found not to be the chief cause of
immunity, such immunity being rather due to the ability of this clone rapidly to form
a thick new cuticle after wintering which prevents the Odium spores penetrating.
These and other mildew - resistant or - tolerant clones are being used in Ceylon,
where Oidium does a lot of harm in certain humid districts, as parents in a breeding
programme aimed at developing resistance to this disease (Wyenwantha, 1965).
The South American leaf blight endemic in South and Central America (abbreviated
to S.A.L.B.), which is caused by Dothidella ulei P. Henn. has wrought enormous de­
vastation in that part of the world and Brazilian and American research workers

453
F. P. FERWERDA

have been combating it, particularly during and after World War II. S.A.L.B.-resis­
tant, though not very productive, clones were developed as a result of breeding work
conducted by the Instituto Agronomico do Norte in Brazil in co-operation with the
U.S. Department of Agriculture. The main source of resistance was Hevea bentham-
iana (Rands and Polhamus, 1955; Townsend, 1960). According to the last-mention­
ed author, more than 12,000 leaf-blight-resistant clones were selected out of more
than 133,000 cross-pollination progenies in the period 1942-1956.
For years now it has been realized in the rubber-producing countries of the eastern
hemisphere that the possibility of the eventual penetration of their part of the world
by S.A.L.B, is a very real one. In Malaysia extensive precautions have been taken.
Should S.A.L.B. break out, a plan will come into operation for the eradication of
the disease by defoliating all the rubber trees by means of the aerial application of
2,4,5-T herbicide (Hutchison, 1958).
In various countries, including Liberia (only five hours' flying time from Brazil!)
but also Ceylon and Malaysia, long term breeding programmes have been initiated,
based on crosses between susceptible Eastern clones and resistant South American
clones, followed by repeated backcrosses with the highly valuable non-resistant re­
current parent, accompanied by continuous screening of the progenies thus obtained
for the occurrence of resistant individuals. These latter are likewise required to possess
the production capacities of the high-producing Eastern parent.
Such a backcrossing programme covering a period of many years, which was
carried out by the Firestone Plantations Company in Liberia, is described in detail by
Bos and Mclndoe (1965). In the course of the programme numerous individual
difficulties arose, each of which had to be overcome in its turn.
To start with, a number of S.A.L.B.-resistant South American clones were trans­
ported to Liberia after adequate quarantine measures had been taken. These clones
were already the result of repeated backcrosses between resistant Brazilian material
and susceptible Eastern clones.
Using these resistant clones, a further backcrossing programme was carried out. In
view of the fact that the number of seedlings obtained by hand-pollination is notorious­
ly low, most of the crossed seed was obtained from isolated seed orchards having a
lay-out such as that illustrated in fig. 11. In each seed plot one Eastern clone and three
S.A.L.B.-resistant clones are planted, two of which are removed in the fourth or
fifth year on the strength of observations made of the growth and flowering habits.
Seeds are harvested from the non-resistant Eastern parent. These seeds must have
arisen partly from selfing and partly from crossing. This gives rise to no difficulties,
since the plants grown from self-pollinated seeds have no resistance and are therefore
automatically eliminated in the screening test for resistance.
This screening for resistance should be carried out in a country where the disease
is endemic; in the case under discussion, the country was Guatemala. To that end, the
seedlings which were to be studied were raised in nurseries in Liberia until they were
about one year old and then stumped. The upper parts of the seedlings were despatch­

454
 RUBBER

Shelter Belt

• • • • • •
• • • • • •
O o o O o o
A A A A A A
• • • • • •
• • • • • •
O O O O O O
A A A A A A
• • • • • •
• • • • • •
O o O o o o Fig. 11 Layout of an isolated seed garden for
large scale production of the cross between
A A A A A A
S.A.L.B. resistant clones (•, O and A) and one
high yielding, susceptible Eastern clone (•).
(After Bos and Mclndoe, 1965).

ed by air to a Firestone estate in Guatemala, where they were budded out in a screening
nursery, in which many susceptible seedlings provided an abundance of spores.
The result of the screening test was reported to the Liberia station, where the small
number of seedlings that had provided a resistant clone were further raised and used
for the continuation of the backcrossing programme.
Later, facilities were provided for despatching seed to Guatemala for screening and
for transferring the resistant clones back to Liberia through the intermediary of a
quarantine station in Florida. In this way much larger numbers of individuals could
be handled. As a result of this rather elaborate procedure only 129 (1.7 %) out of 7542
clones tested exhibited sufficient resistance. At least ten years are required to complete
one back-crossing cycle and to obtain a preliminary impression of the yield capacity
of the clones thus obtained. This example illustrates how slow and time-consuming
a process it is in the case of a tree crop like Hevea to breed clones which combine high
yield with disease resistance.
New complications have arisen as a result of the advent of new extremely virulent
physiological races of Dothidella ulei to which breeding parents hitherto considered
resistant have proved to be susceptible. Therefore, the basis of resistance will have
to be broadened by also incorporating in the breeding programmes H. pauciflora and
H. brasiliensis material from the Madré de Dios region which has been recognized as
having a high degree of Dothidella resistance (Seibert, 1947). A long road has yet to
be covered before these programmes attain their goal.

455
F. P. FERWERDA

What has been achieved and future developments

During the fifty years of work aimed at the systematic improvement of the rubber
plant remarkable results have been achieved. The best clonal material now available,
when planted on a large scale, gives yields which are three to four times higher than
those of the original non-selected seedling material. In addition properties closely as­
sociated with the yield capacity such as bark thickness, bark renewal and stem girth
have been considerably improved. The best of the improved seedling families are
hardly inferior to the clones planted on a commercial scale.
Improvement of the composition of the latex and of the chemical and physical
properties of the rubber obtained from it may become an important factor in the futu­
re. Sufficient potential for such improvement certainly appears to exist. In this respect,
clonal selection offers the most possibilities.
The problem of the rootstock has received relatively little attention in rubber
cultivation. It is generally recognised that the rootstock-scion interaction is considera­
ble. Certain seedling families have been found to have distinct advantages as root-
stocks for a number of clones. Particularly striking results, giving yield increases of up
to 30 % were obtained in experiments with hybrid spruceana rootstocks. This problem,
particularly the last-mentioned facet of it, deserves further study and more extensive
trials. Exploratory tests covering the vegetative multiplication of rootstocks with the
object of obtaining rootstock clones have provided promising results which, however,
have not yet materialized into rootstock experiments using this vegetatively propagated
material.
The future course to be followed by breeding programmes in general will be in­
fluenced to a large extent by developments in the synthetic rubber industry.

References

BALDWIN, J. T. jr. 1947. Hevea, a first interpretation. Journ. of Heredity, 38: 54-64.
BAPTIST, E. D. C., 1952. Recent progress in Malaya in the breeding and selection of clones of Hevea
brasiliensis. Report of the thirteenth International Horticultural Congress, London: 1100-1121.
Bos, H. and MAC INDOE K. G., 1965. Breeding of Hevea for resistance against Dothidella ulei P. Henn.
Journ. R. R. I. Malaya, 19: 98-107.
BOUHARMONT, J., 1960. Recherches taxonomiques et caryologiques chez quelques espèces du genre
Hevea. Sér. Sei. INEAC, no. 85.
BOUHARMONT, J., 1962. Fécondation de l'ovule et développement de la graine après croisement et
autopollinisation chez Hevea brasiliensis Muell. Arg., Cellule, 62: 119-130.
BOUYCHOU, J. G., 1956. The origins of the Far-East rubber tree I. Introduction and distribution. Rev.
Gén. du Caoutchouc, 33: 730-735.
BROOKSON, C. W., 1959. Seedlings and clones of Hevea brasiliensis. Proc. C.I.T.A. World Congress of
Agricultural Research, Rome: 699-703.
CHITTENDEN, R. J., 1950. The Prang Besar (Malaya) experiments in the selection and propagation of
Hevea. Empire Journal of Experimental Agriculture, 18: 105-111.
DUCKE, A. D., 1935. Revision of the genus Hevea mainly the Brazilian species. Arch. Inst, de Biologia
Vegetal, 2: 217-346, (revised ed. 1939).

456
 RUBBER

DIJKMAN, M. J., 1938. Preliminary data on storage of Hevea pollen (In Dutch). Arch. v. d. Rubber-
cult., 22: 239-259.
DIJKMAN, M. J., 1951. Hevea, thirty years of research in the Far East. Univ. of Miami Press.
DIJKMAN, M. J. and OSTENDORF, F. W., 1941. Seedling test garden Pangkalan 1929. Arch. v. d.
Rubbercult., 25: 435-467 (In Dutch).
EDGAR, A. T., 1960. Manual of Rubber planting. Ed. Incorp. Soc. of Planters, Kuala Lumpur.
EHRET, M., 1948. Étude sur la sélection de l'Hevea en Indochine. Cahiers I.R.C.I., 3: 13.
FERWERDA, F. P., 1940. Data, up to 1939, on clonal and seedling test gardens planted 1929-1932.
(Dutch, Engl, summary). Arch. v. d. Rubbercult., 24: 353-394.
FERWERDA, F. P., 1953., A possible explanation of the divergence between juvenile type budgrafts and
their seedling mother trees in Hevea. Euphytica, 2: 15-24.
GEORGE, P. J., a.o. 1967. Observations on the floral biology of, and fruit set in Hevea brasiliensis Müll.
Arg.. Rubber Board Bulletin (India), 9: 18-27.
GREVEN, M. and VOLLEMA, J. S., 1950. Results of selected seedlings in comparison to recommended
clones in rubber cultivation (In Dutch). Bergcultures, 19: 159.
HUTCHISON, F. W., 1958. Defoliation of Hevea brasiliensis by aerial spraying. Journ. R. R. I. Malaya,
15: 241.
JACOB, J. C. 's, 1931. Experiments on artificial self and cross pollination in Hevea brasiliensis. (In
Dutch) Arch. v. d. Rubbercult., 15: 261.
LA RUE, C. D., 1926. The Hevea Rubber in the Amazon Valley. U.S.D.A. Bull. No. 1422.
LASSCHUIT, J. A. and VOLLEMA, J., 1953. The mildew resistant clone LCB 870. Arch. v. d. Rubbercult.
Conference of Rubber Research Institutes in the Far East, Bogor, 30: 132-135.
LOOMIS, H. F., 1942. Methods of splitting Hevea seedlings. J. Agric. Res., 65: 97-124.
MAAS, J. G. J. A., 1919. Floral biology of Hevea brasiliensis (In Dutch). Arch. v. d. Rubbercult., 3:
280-312.
MAAS, J. G. J. A., 1934. Breeding of Hevea brasiliensis at the Government Agricultural Enterprises
(In Dutch). Arch. v. d. Rubbercult., 18: 58-81.
MAAS, J. G. J. A., 1948. Selection of Hevea and its effect upon the yield of estate plantations. Arch. v.
d. Rubbercult., 26: 108-114.
MAJUMDAR, S. K., 1966. Pollen longevity studies in Hevea brasiliensis. Transactions Kentucky Acad,
of Sc., 27: 16-18.
MAJUMDAR, S., 1967. Male sterile clones in Hevea brasiliensis. Canadian Journ. of Botany, 45: 145—
146.
Mc INDOE, K. G., 1958. The development of clonal rootstock in Hevea. Rubber Research Institute
of Ceylon Combined Quarterly Circulars, 34, (3-4): 1-19.
MORRIS, L. E., 1929. Field observations and experiments on the pollination of Hevea brasiliensis.
Journ. R. R. I. Malaya, I: 41-49.
PAARDEKOOPER, E. C., 1956. The occurrence, properties and possible application of Juvenile Type
buddings. Arch. v. d. Rubbercult., 33: 141-158.
PAARDEKOOPER, E. C., 1956. Results of the testing of Hevea seedlings obtained by hand pollination
during 1927-1944. Arch. v. d. Rubbercult., 33: 61-140.
PAX, F. and HOFFMANN, K., 1931. Euphorbiaceae in: Engler und Prantl, Die natürlichen Pflanzen­
familien., 2nd ed. 1931, vol. 19c.
P. W. J. Annual Report, West Java Experiment Station, Bogor, Indonesia. 1939.
RAMAER, H. 1935. Cytology in Hevea. Genetica, 17: 193-236.
RANDS, R. D. and POLHAMUS, L. G., 1955. Progress report on the co-operative rubber development
programme in Latin America. Circ. U.S. Dep. Agric. No. 976.
Ross, J. M., 1960, Observations on the 1959 hand pollination programme at the Rubber Research
Institute of Malaya. Proc. Nat. Rubber Res. Conf. Kuala Lumpur: 392-408.
Ross, J. M., 1965. Progress of breeding investigations with Hevea brasiliensis IV. The crosses made in

457
F. P. FERWERDA

the years 1947-1958. Doc. Res. Arch., Rubber Research Institute of Malaya, 28: 1-32.
Ross, J. M. and BROOKSON, C. W., 1966. Progress of breeding investigations with Hevea brasiliensis
III. Further data on the crosses made in the years 1937-1941. Journ. R.R.I.M., 19: 158-172.
R.R.I.M., 1939. Annual Report, Rubber Research Institute of Malaya, Kuala Lumpur, Malaysia.
R.R.I.M., 1967-1968. Pl. B. Rubber Research Institute of Malaya, Planters' Bulletin, 88 and 94.
SCHMÖLE, J. F., 1937. Results of test tapping in some older A.V.R.O.S. clones. Arch. v. d. Rubber-
cult., 21:10-16 (In Dutch).
SCHMÖLE, J. F., 1940. The influence of the rootstock on the yield of buddings (In Dutch). Arch. v.d.
Rubbercult., 24: 305-314.
SCHMÖLE, J. F., 1941. Hevea brasiliensis and Hevea spruceana hybrids as stock forbudgrafts. Arch. v.d.
Rubbercult., 25: 159-167.
SEIBERT, R. J., 1947. A Study of Hevea in the Republic of Peru. Ann. Missouri Bot. Garden, 34:261-
352.
TINLEY, G. H., 1960. Vegetative propagation of clones of Hevea brasiliensis by cuttings. Proc. Nat.
Rubber Res. Conf. Kuala Lumpur: 409-418.
TOLLENAAR, D., 1941. The importance of the rootstock problem in rubber cultivation. Bergcultures,
15: 1014-1019 (In Dutch).
TOWNSEND, C. H. T. Jr., 1960. Progress in developing superior Hevea clones in Brazil. Econ. Bot., 14:
189-96.
WARMKE, H. E., 1951. Studies on pollination of Hevea brasiliensis in Puerto Rico. Science, 113: 646-
48.
WARMKE, H. E., 1952. Studies on natural pollination of Hevea brasiliensis in Brazil. Science, 116:
474-475.
WIERSUM, L. K., 1955. Observations on the rooting of Hevea cuttings. Arch. v. d. Rubbercult., 32:
213-243.
WYENWANTHA, R. T., 1965. Some breeding problems in Hevea brasiliensis. Quarterly Journal of the
Rubber Research Institute of Ceylon, 41:12-22.

458
TEA

Camellia sinensis (L.) O. Kuntze

T. VISSER

Institute of Horticultural Plant Breeding, Wageningen, The Netherlands

Taxonomy

CLASSIFICATION

LINNAEUS (1752) was the first to typify the tea plant as Thea sinensis. Later on, two
distinct taxa obtained recognition. The small-leaved 'China' bush usually with the
epithet 'sinensis', and the large leaved 'Assam' tea plant. The latter was discovered
in Assam in the eighteen thirties (Wallach, 1935; Griffith, 1938) and described by
Masters (1844) as Thea assamica. The typical China variety is a shrub with more or
less virgate stems arising near the ground; it is 1.0-3.0 m high and has relatively small,
hard, dark green, 3-6 cm long, leaves with a dull (matt) surface. The typical Assam
variety is a small, 10-15 m high, many-branched tree, sometimes with a trunk one-
third its height, with supple light green, 15-20 cm long, leaves, with a glossy surface.
Originally, the genus Thea existed separately from the genus Camellia, but the tea
plant is nowadays classified as Camellia on account of the close resemblance between
the species included in the two genera. Sealy noted in 1937 that the correct name
for the tea plant should be Camellia sinensis (L.) O. Kuntze.
Until recently the onlyintra-specificforms of C.sinensis (L.) generally recognized were
those described by Kitamura (1950) as C. sinensis var. sinensis (L.) and C. sinensis var.
assamica (Masters). However, extensive investigations by Wight, Barua, Roberts and
Wood - research workers associated with the Tocklai Experimental Station - have
shown that a concept which includes only these two kinds of tea is inadequate to ac­
count for the différent variants found among the tea plants cultivated in India and
probably elsewhere.
Wight (1962)-see also Barua (1963b; 1965) - accorded a specific rank to the
'China' and 'Assam' kinds of tea. He retained the name Camellia sinensis (L.) for the
former and renamed C. sinensis var. assamica (Masters) C. assamica (Masters). A
third form of tea referred to as the 'Cambodia race' by Kingdom Ward (1950) and
termed 'Southern form of tea' by Roberts et al. (1958) could be equated to Planchon's
Thea lasiocalyx and has been named C. assamica ssp. lasiocalix (Planch-MS). It
was treated as a subspecies of C. assamica because of its close relation to this species.
Barua (1965) gave a specific rank to a fourth form - known as 'Wilson's Camellia'

459
T. VISSER

at Tocklai - which has been described and named as C. irrawadiensis P. K. Barua.

Another kind of tea allied to C. irrawadiensis is 'Forest's Camellia', referred to by
Sealy (1948) as C. taliensis. However, on account of certain anatomical and chemical
properties, the latter appears to be a hybrid of C. sinensis and C. irrawadiensis and
its separate taxonomie status is therefore probably undeserved (Barua, 1958; Roberts
et al., 1958; Wood and Barua, 1958).
While the so-called variant C. taliensis is likely to make acceptable tea, C. irrawa­
diensis makes a product which visually looks like tea, but is spurious as an infusion.
Hybrids between the latter species and C. assamica did however, produce tea, even
if not of the best quality (Wight and Barua, 1957). It is probable that certain valuable
characteristics of Darjeeling tea have been added to it by C. irrawadiensis (Wood,
1958; Wight and Gilchrist, 1961). It has been suggested by Wood and Barua (1958)
that this latter species exists in its pure form in the Shan States and that uncontrolled
crossing has occurred between this taxon and C. sinensis. This species is, however,
not completely compatible with tea proper. Crosses and back-crosses with tea as a
seed-bearing parent have been known to succeed, but neither the reciprocal nor the
back-cross to C. irrawadiensis has been successful (Barua, 1958).
C. assamica subsp. lasiocalyx is not generally cultivated in Assam or North-East
India (Wight, 1962). On account of its odour it is of value for the breeding of new kinds
of tea, although these may not always be acceptable. According to Wight (1962), its
thoughtless introduction may significantly and perhaps adversely alter the character­
istics and, particularly, the liquor strength of Assam tea. Under the different manufac­
turing conditions and requirements existing in Ceylon, several outstanding clones - in
which a strong element of this variant occurs (Green, 1958) - were found to produce
excellent teas. The original mother bushes (seed bearers) were, however, discarded at
Tocklai (Assam), because they were lacking in typical Assam qualities.

DISTRIBUTION AND ORIGIN

According to Sealy (1958), the genus Camellia includes as many as 82 named spe­
cies. Its principal area of distribution is in the highlands of South-East Asia above
the 30th Parallel; its range extends from Nepal north-eastward to Formosa, the
Liu-Kiu Islands and southern Japan, while it is absent from southern Siam and the
Malay Peninsula.
Tea is one of the most widespread Camellias and also the most important one
commercially. The tea-drinking habit was earliest known in China, its origin dating
back to more than 2000 years ago but the tea leaf has probably been in use as a méde­
cine much longer than that.
Since tea has been cultivated and dispersed by man over such a long period, it is not
easy to determine whether it still exists or ever existed in the wild state, and, if so,
where and in what form it originated. An added difficulty is the fact that tea is ex­
tremely variable, as it is mainly a cross-pollinator; in all likelihood, not only most

460
 TEA

cultivated teas but also the 'wild' teas are hybrids. The indigenous tea plants discover­
ed, e.g. in Upper Assam and Indo-China, in comparatively recent times are probably
only semi-wild. Quite probably their occurrence and the dispersal of tea in South-
East Asia in general, are facts associated with the exploitation and subsequent aban­
donment of perennial tea plantations by the migratory tea-drinking peoples of this
area.
As to the origin of the main races, Kingdom Ward (1950) holds the view that the
China and Assam races, being so different from each other in habit and appearance,
must have had separate origins. This author reasons that the primary centre of dis­
persal is likely to have been in Central Asia, possibly as far north as the 60th parallel.
The China race may have come from the north around the Pacific seaboard, while
the Assam race and the related Cambodia race took the more direct route southwards.
In this connection, the region of the sources of the Irrawady may have been a seconda­
ry centre of dispersal.

Cytology

In its normal form tea is diploid (2n = 30 chromosomes). Numerous studies have
been carried out, especially in Japan, on the breeding and utilisation of polyploid
varieties of tea (Harada et al., 1957 ; Simura et al., 1952,1953; Simura, 1957). Natural
triploids have been found among Japanese 'varieties' as well as among the large-leaved
China tea called 'macrophylla'. Tetraploids have been obtained from the progenies
of triploid plants and by treating the growing points of diploid seedlings with colchi­
cine.
The viability of the pollen and the fertility of triploids are usually poor. Tetraploids
appear to be more fertile that triploids, but less so than diploids. Crosses between
female tetraploids and male diploids were found to succeed well, whereas the reci­
procal crosses almost all failed (Toyao, 1965). Polyploid plants generally have thicker
leaves with larger stomata, although the number of stomata per unit area is smaller.
Triploids grow more vigorously, have larger leaves and are hardier than diploids
and may, therefore, be of considerable use, e.g. as clones, in colder climates. The in­
formation on tetraploids appears scanty ; their use is probably restricted to breeding.

Growth and development

The tea plant in its natural habitat (forest) - or as a seed bearer - grows into a
shrub or small tree, which starts to flower after about six years or so, sometimes earlier,
sometimes later, depending on the variety and the conditions. The environment and
growth habit of the tea plant as a plantation crop are very different. Estate conditions
require the close planting of some 10,000 plants per ha, which are shaped into low
bushes and subjected to various treatments to ensure that a maximum crop of young
shoots is maintained. Tea fields may be established from seed or from cuttings (clones).

461