XXX III REUNION
ABRIL 4 4
H ihe ,' .
SYM. iS -.
^ **.- A
n Breeding for
in Common Beane :
. ..' I
,.%&^ ^ 1
CENTRO INTERNATIONAL DE
INSTITUTE DE CIENCIA Y
XXXIII ANNUAL MEETING
SYMPOSIUM ON BREEDING FOR
HIGHER YIELD IN BEANS
March 31, 1987
Guatemala City, Guatemala
INSTITUTE DE CIENCIA Y
CENTRO INTERNATIONAL DE
BEAN COWPEA COLLABORATIVE
RESEARCH SUPPORT PROGRAM
MORPHO-PHYSIOLOGICAL CHARACTERIZATION OF EARLY GENOTYPES
IN COMMON BEANS (Phaseol us vuliaris L.)
Rafael Rail RodriQuez**
Salvador Miranda Colin**
Stephen E. Beebe***
The present study was carried out with the object of characterizing the
most important morpho-physiological traits and identifying archetypes
associated with higher yield in a group of early genotypes of common beans
(Phaseolus vulgaris L.) from Central America, Mexico, and Colombia.
Results obtained led to the conclusion that: (1) yield and physiological
maturity are highly and positively correlated with size parametaerc-
(2) early materials tend to be more efficient than late ones in their
utilization of leaf area; (3) to obtain higher yields, a strong, erect
stem, with few branches is desirable in early genotypes, as well as an
early-flowering and longer maturing genotype.
This work forms part of the thesis with which the main author obtained
his Master of Science degree, Centro de Genftica, Colegio de
Posgraduados, Chapingo, Mexico.
** Student and Professor-Scientist, respectively, Centro de Genetica,
Colegio de Posgraduados, Chapingo, 'Mixico.
*** Breeder, Bean Program, Breeding I, Centro Internacional de Agricultura
Tropical (CIAT), Call, Colombia.
Improvement of earliness, in common beans (Phas~e~_lu vulgaris L.), has
always been impaired by the fact that this characteristic has not been
sufficiently characterized quantitatively. There is some knowledge as to
its inheritance, but no numeric data are available on the characteristics
of possible early progenitors which both offer earliness as a mechanism to
escape adverse factors and at the same time sustain yields. Furthermore,
the dynamics of growth in early genotypes is not well known; this knowledge
could better orient a breeding program to recombine earliness and yield.
Due to risk of irregular rainfall distribution, many farmers prefer
short-cycle bean varieties, with which the crop is exposed in the field
during a shorter period, and therefore, has fewer risks.
However, earliness imposes certain limitations on the biological capacity
of the crop. A short cycle limits total photosynthesis and, to a certain
degree, yield potential. Additionally, a reduction in the total biomass of
the plant is observed, resulting in a smaller number of floral buds and
The present study was conducted with the object of characterizing a group
of early genotypes from Central America, Mexico and Colombia, that could
serve as outstanding progenitors in a breeding program for earliness and
high yield in common beans, as well as to identify some characteristics
related with higher yields.
The suggested working hypotheses were:
a. Smaller size parameters in early genotypes are related with lower
b. Early genotypes are more efficient than late ones, since they require
less structural material (smaller size);
c. A main strong, erect stem having few branches, is associated with
higher yield in early genotypes; and,
d. Late-flowering and early-maturing early genotypes have higher yields.
REVISION OF LITERATURE
Buttery and Buzzel (1972) point out that the search for high yields has
stimulated the interest in the physiological factors contributing to yield.
To increase yield in beans it has been suggested to identify the limiting
factors under favorable conditions and determine the characteristics
related with productivity (Tanaka and Fujita, 1979; Charles-Edwards, 1982),
as well as the most appropriate combination of variety, environment, and
agronomic practices (Yoshida, 1972); the genetic, physiological, and
environmental factors capable of influencing production must be considered
simultaneously (Wallace and Munger, 1966; Charles-Edwards, 1982).
Yoshida (1972) points out that the vegetative growth, the formation of
storage organs, and seed filling must be studied in order to obtain higher
yields. On the other hand Poey (1978) considers, with this same objective,
the need to study the vegetative cycle, plant architecture, and leaf area.
Mendoza and Ortiz (1973) point out that seed yield must not be the main
criterion of selection; criteria regarding seed production efficiency must
Laing (1977) and Bauer (1969) emphasize the need for early maturing bean
varieties with the object of avoiding dry periods.
Although it has been confirmed that late bean varieties yield more than
early varieties, the farmer prefers the latter in which case he has less
risk (Leiva, 1977; Garcia, 1985).
According to Donald (1968), to date two groups can be considered in terms
of the philosophy of the breeding programs to increase yield:
a. Where the purpose is to remedy some known defects in the crop or
"elimination of defects" and,
b. Where the basic objective is "selection by yield" without considering
the reason or the cause of the yield obtained. A valuable additional
option would be improvement by archetypes.
The improvement of archetypes implies the definition of both the
environment for production, and the morpho-physiological characteristics of
the archetype (Mock and Pearce, 1975); improvement of plant structure
(Adams, 1973); and changes in agronomic practices (Donald, 1968).
An erect, strong stem with few branches (2-4 branches forming upwards acute
angles with the stem), are attributes of a high-yielding bean archetype
(Laing, 1977; Davis and Evans, 1977; Stofella et al., 1979; Adams, 1973,
1981; Izquierdo and Hosfield, 1983).
MATERIALS AND METHODS
The present study was carried out at the CIAT-Palmira station of the Centro
International de Agricultura Tropical, CIAT, Colombia.
The Station is located at 3"30' latitude North, 7622' 22" longitude West,
and 965 masl; it has a mean annual temperature of 23.9C and a mean annual
rainfall of 938 mm.
Two experiments were conducted: the first (EA) was planted on June 28,
1985 and the second (EB) on September 19, 1985. The experimental plot was
composed of nine rows 6.0 long, with 0.6 m between rows. Subplots were
located within this plot where growth, yield, and its components were
measured. The eight samples for the growth analysis had 1.0 linear m (0.6
m2), adjusting the data later to 1.0 m2. For yield, the net plot had five
2.0 m rows, with 0.6 m between rows. Yield components were measured on 1.2
Sixteen bean (Phaseolus vulgaris L.) genotypes supplied by the CIAT Bean
Program were used, of which 12 presented diverse degrees of earliness; the
remaining four were used as late controls. The characteristics of the
materials are shown in Table 1.
Crop management recommendations of the Program were used.
The data collected was divided into three classes: data of growth
analysis, development data, and yield data and its components.
Eight samplings were done for the growth analysis. The first, which was
common to all genotypes, was done 12 days after planting; on the other
hand, the last sampling was made to coincide as much as possible with the
physiological maturity of each genotype. Samples were taken at random as
to their location within a total plot. Subsequently they were dried during
72 hours at 65C to arrive at a constant weight. Leaf area was measured
with an automatic Hayashi Denkoh apparatus, manufactured in Japan, model
AAC-200. The Functional Approximation was used.
Developmental data were taken from the yield subplot. To determine yield
and its components, each subplot was measured at random.
In all cases, results were subjected to analysis of variance and then to a
comparison of means by Duncan's method (5%); at the'same time linear
correlations were established among all the variables of interest, both
considering all genotypes (n = 16) and eliminating the late group (n = 12).
RESULTS AND DISCUSSION
Overall, in most of the variables studied, the tendencies were similar in
the two experiments, although their values were greater in experiment B
In general, it has been observed in everyday work, that reduced parameters
of size are found to be correlated with low yields in early materials.
Tables 3 and 4 give the basic information with respect to some variables
related with plant size.
In Table 5, size variables considered important have been grouped together,
based on preliminary analyses. In both planting periods these variables
are consistently associated with yield. Although correlations are positive
in all cases, there is absence of significance for total nodes to maturity
and maximum height. An analogous relationship exists between the size
variables and days to physiologic maturity. Considering the early group (n
= 12), tendencies seem to be less marked (Table 5). Therefore, results
suggest an overall close and positive association among parameters of size,
physiological maturity, and yield.
Within the group of early genotypes and for most of the size parameters,
there are two genotypes which always stand out among that group; these are
JU-84-7 and Rojo de Seda. Both consistently present the highest values for
the size parameters, and have intermediate earliness; these two genotypes
could be used as progenitors which deviate from the general tendencies
already mentioned (Tables 3, 4, and 8).
Efficiency, understood as the greatest gain at the lowest possible cost, is
a characteristic which is expected to be maximized in early genotypes, as
compensation for the lov size parameters of these
Tables 3 and 4 show the basic values for efficiency variables during the
two periods studied.
Considering dry matter production per unit of surface and time (Crop Growth
Rate, CGR), consistently the most outstanding genotypes within the early
group are Rabia de Gato, Rojo de Seda, and JU-84-7 for the two periods.
This is important because CGR is directly related to total dry weight,
which is associated with yield (Mesquita, 1973; Mosjidis, 1975) and has
also been suggested as an acceptable selection parameter for'high yield in
beans (Wallace, 1973; Masaya, 1985).
Table 4 shows that the highest CGR value was 16.4 g/m2/day corresponding to
JU-84-7, 30 days after emergence. This figure is very similar to the
highest CGR value obtained for soybean (Buttery, 1970), which was
17.2g/m2/day, and highlights the importance of JU-84-7 since this genotype
combined earliness and high CGR.
Efficiency of a plant in increasing its dry weight from its already
existing weight is important, especially during the first stages of growth.
Such efficiency, plus good initial agronomic management of the crop will
insure good growth during the first stages of the vegetative cycle, which
is important for plant establishment. Efficiency in this respect is
referred to as Relative Growth Rate (RGR).
In experiment A (EA), JU-84-7 was observed to be a very efficient genotype
in its RGR (Table 3), surpassed only by ICTA-Quetzal. In experiment B (EB)
JU-84-7 again shows the highest RGR value -- a marked difference with
respect to the total group of genotypes (Table 4).
CGR and RGR are positively associated with yield, either considering all
the materials or only the early ones, and especially in EA (Table 6).
It is likely that prevailing climatic conditions which were contrasting for
the two periods were the cause motivating a close association of these
rates with maturity in EA and a null association in EB; favorable
conditions resulted in exuberant growth of all genotypes in EB. Any
advantage that the late genotypes might have had was lost due to
inefficient use of excessive leaf area.
The efficiency of the leaf area of a plant can be described by the Net
Assimilation Rate (NAR), which is closely linked to the Leaf Area Index
(LAI). When the LAI passes a value of 4.0, considered by Laing et al.
(1983) as optimal, its efficiency declines due shading of leaves in the
lower part of the canopy. This reduced efficiency will be reflected in
lower values of NAR.
In EA, no genotype attained the optimal value of maximum LAI (MAXLAI),
while in EB many genotypes attained it and several exceeded it, especially
the later genotypes. This fact is manifested in lower values of NAR in the
late genotypes compared to EA (Table 4). NAR had a low and positive
association, although not significant, with maturity and yield in EA
(Table 6). But in EB such association became negative and highly
significant for maturity and of smaller magnitude for yield.
These observations on MAXLAI and NAR explain many of the differences
observed between EA and EB. For example, in EA there were significant
differences between genotypes for the parameters Maximum Total Dry Weight
and Yield/Day, but these disappeared in EB (Table 7). Apparently the late
genotypes lost much of their advantage in EB and the early.genotypes almost
equalled the late materials in these parameters. It appears that the late
genotypes arrived at a sort of "ceiling" in EB, since these did not
demonstrate a proportional improvement compared to the early materials in
passing from EA to EB. The MAXLAI no doubt is an important part of this
On the other hand, since NAR is a "net" rate, respiration is also involved
(Williams et al., 1965; Wallace and Munger, 1965); this should be taken
into account in interpreting fluctuations in the NAR as observed between EA
and EB, nevertheless, the relationship of respiration to LAI and other
parameters measured is not known.
Among the early group in EA, JU-84-7 was the most outstanding in terms of
yield and had a high NAR (Tables 3 and 8). In addition to JU-B4-7,
genotypes 6 3017 and Rabia de Gato show in EB a good combination of yield,
NAR and earliness (Tables 4 and 8). The contradictory NAR responses
obtained in the two periods do not permit firm conclusions as to a pattern
of response for early bean genotypes.
Yield/Day presented a significant association with maturity (r = 0.58 +)
only in EA (Table 6); in EB, this relationship is inverted (r2 = -0.48)
within the early group. It is likely that in EB. the greater leaf area
(LA) of the early genotypes may have influences the increases in yield and
that on the contrary, in the late genotypes, this greater LA was not used
as efficiently for yield increase as it was by the early genotypes.
Additionally, Yield/Day is a good parameter of efficiency to quantify final
yield, when comparing genotypes of different biological cycle. For
example, if yield per se is analyzed in experiment B, significant
statistical differences are found; but if Yield/Day is calculated, the
analysis shows that on this basis, all genotypes are statistically equal
(Table 7 and Figure 1).
Within the early group in EA the most outstanding genotype in terms of
Yield/Day is JU-84-7, and in EB, JU-84-7 and G 3017; the latter has the
advantage of being an earlier material although its yields are slightly
lower (Table 8).
In cereals, due to their low degree of defoliation, the harvest index (HI)
is a good yield selection criterion. Instead in legumes such as beans,
defoliation could affect the relationship between HI and yield. This
suggests that the poor environmental conditions (experiment A) could cause
rapid defoliation of the plant and thus an erratic HI, especially under the
conditions of the present study, where a broad range of maturities were
involved (Table 8). Under good environmental conditions (experiment B),
the HI seems to function better, and comparatively high HI were detected in
early genotypes (Table 8).
Overall, no consistency was found in the use of the HI as a criterion for
selection for high yield in early genotypes (Table 6). It is possible that
in considering total dry matter production, including that which could have
fallen during the biologic cycle (folioles, petioles, and roots), the HI
would be more effective in identifying the best-yielding early genotypes.
Archetype A. A plant type whose characteristics would be compatible with
early maturity was proposed. -It was hypothesized that a high-yielding
plant of reasonable earliness, would be that with a strong, erect primary
stem, and few branches. This coincides with most bean archetypes described
in the literature, although a more detailed description is not done in this
Data obtained for the variables under study appear in Tables 9, 10, and 11.
Results of the variance analysis are found in Table 12.
As shown in Tables 9 and 10, the highest-yielding early genotypes for the
two periods are JU-84-7, G 3017, and Rabia de Gato. These .three genotypes
are representative of the range of plant height of the early group
(Table 11). If the Maximum Length (MAXLEN) variable is considered, it is
found that the difference between MAXLEN and the height of G 3017 and Rabia
de Gato is not sufficient to result in a prostrate habit in the case of EA.
As to JU-84-7 in EA and the three genotypes in EB, the differences are more
marked. In any case, the excess length in the main stem did not induce
With this in mind, the distribution of seed dry weight in the main stem
(SDWS) and in the branches (SDWB) shows the same tendency in the three
cases mentioned, that is, that SDWS is twice SDWB (Table 11); this suggests
that moderate plant ramification is conducive to higher yields (Tanaka and
The three genotypes developed a strong, erect stem with few branches, and
thus it appears that the latter were not the determinants of greater seed
production. Limited ramification of these genotypes reduced the total
number of nodes. This could cause yield reduction due to fewer pods if
the nodes are considered to be potential productive points but such
reduction in pod set was not observed (Tables 9 and 10). Thus, total
number of nodes was not a limitation to yield in this case.
Information in the literature reviewed indicates that bean archetypes
should have 2-4 branches; this coincides with the limited ramification of
the three early genotypes and of the high yield levels found in this study.
This plant structure probably enables greater diffusion of light, air, and
CD2 within the strata of the leaf mass.
Archetype B. Assuming that the time elapsed until flowering is the period
during which plants establish the infrastructure on which fruits develop,
it was hypothesized that an ideal early, high-yielding plant can be
considered to be one that is late flowering, and early to physiological
maturity, even though this would obviously shorten the seed filling
process. Consequently, yield component compensation could be expected,
this expressed as a greater number of pods.
The genotypes which were closer to the model suggested seem to be Rojo
National, Rojo de Seda, and XAN 145, as shown in Tables 9 and 10 for the
two periods. From this same data it can be observed that these genotypes
do present a short seed filling period in both periods, as was expected.
However, they are not necessarily superior in Pods/Plant.
In EA these three genotypes present average to low yields (Table 9),
probably related to short seed filling periods. This situation is
maintained for EB, which had better conditions and therefore all genotypes
maximized their characteristics (Table 10).
Additional evidence is the fact that 6 3017 and Pata de Zope show
practically equal maturity values (DPM); the same is true for Rabia de Gato
and Huetar (Tables 9 and 10). In either of the two comparisons between
materials paired this way, when the number of days to flowering (FLOWER) is
greater, and pod filling period (PFP) is shorter, seed yield decreases.
The inconsistency in the compensation of components throughout the two
periods, in addition to the fact that in EB the mentioned genotypes did not
show any advantage in yield, suggests there is not enough support for the
hypothesis and the model of earliness proposed.
Rather, these results suggest a contrary hypothesis to be feasible that an
early high-yielding genotype should have an extended pod filling period,
even at the expense of days to flower. In fact, correlation of yield to
PFP was .74** and .61** in the two experiments.
According to results obtained in the present study, it can be concluded
1. Yield and physiological maturity are highly and positively correlated
with size parameters. Environmental factors influence these
2. Early materials tend to be more efficient than late ones as to leaf
3 Three genotypes were found corresponding to the overall description of
archetype A: strong, erect stem, with limited ramification; these are
JU-84-7, G 3017, and Rabia de Gato. However, the hypothesis holds
true only partially since the expected compensation was not observed
nor was there any increase in the components Seeds/Pod and Seed
Weight. This suggests that the number of productive nodes was not
limiting for yield.
4 There was not enough evidence to support the model: late
flowering-early maturity. Although, genotypes corresponding to this
model were observed, they presented average to low yields, probably as
a consequence of their short pod filling period. In general, this
latter parameter seems critical for final yield; therefore it is not
convenient to reduce it, and it may in fact be desirable to extend it.
ADAMS, M.W. 1973. Plant architecture and.physiological efficiency in the
field bean. In: Potential of field bean and other legumes in Latin
America. Centro Internacional de Agricultura Tropical, CIAT, Call,
Colombia. p. 266-278.
S1981. Update: New bean archetype. Michigan Dry Bean
BRAUER H., 0. 1969. Fitogen6tica aplicada. Limusa, Mexico. 422 pp.
BUTTERY, B.R. 1970. Effects of variation in leaf area index on growth of
maize and soybeans. Crop Sci. 10:9-13.
and R.I. BUZZELL. 1972. Some differences between soybean
cultivars observed by growth analysis. Can. J. Plant Sci. 52:13-20.
CHARLES-EDWARDS, D.A. 1982. Physiological determinants of crop yield by
breeding. Hortscience 15:244-247.
DAVIS, J.H.C., and A.M. EVANS. 1977. Selection indices using plant type
characteristics in navy beans (Phaseolus vulgaris L.). -J. Agric. Sci.
DONALD, C.M. 1968. The breeding of crop ideotypes. Euphytica 17:385-403,
GARCIA B., C.M. 1985. Interacci6n genotipo x ambiente en el frijol comdn
(Phaseolus vulgaris L.). Tesis de Maestria er Ciencias. Colegio de
Posgraduados. Chapingo. Mexico. 88 pp.
IZQUIERDO, J.A., and G.L. HOSFIELD. 1983. The relationship of seed
filling to yield among dry beans with differing architectural forms.
J. Amer. Hort. Sci. 108:106.
LAING, D.R. 1977. Implicaciones de la investigaci6n fisiol6gica para
mejoramiento de frijol comin (Phaseolus vulgaris L.). Centro
International de Agricultura Tropical, CIAT, Cali, Colombia. 11 pp.
P.J. KRETCHMER, S. ZULUAGA, and P.G. JONES. 1983.
Field bean. In: Smith, W.H. and S.J. Banta (eds.). Symposium on
Potential productivity of field crops under different environments.
Los Baios, Philippines, 1980 (Proceedings). Manila, International
Rice Research Institute. p. 227-248.
LEIVA R., O.R. 1977. Herencia y mejoramiento de la precocidad del frijol
(Phaseolus vulgaris L.) en el tr6pico. Tesis Mag. Sc. BogotA.
Universidad Nacional de Colombia, Instituto Colombiano Agropecuario
MASAYA, P. 1985. Selection for yield in early generations by total
weight. harvest index and yield/day. Report on the work done at CIAT
during 1985 as a CIAT fellow.
MENDOZA, B., E. 1973. Influencia de algunos components morfol6gicos en
el. rendimiento del frijol (Phaseolus vulQaris L.). Tesis de Maestria
en Ciencias. Colegio de Posgraduados. Chapingo, Mexico, T8 pp.
MOCK, J.J. and R.B. PEARCE. 1975. An ideotype of maize. Euphytica
MOSJIDIS CH., J. 1975. Distribuci6n de la material seca entire los 6rganos
aereos de various cultivares de frijol (Phaseolus vulqaris L.).
Fitotecnia Latinoamericana 11:29-33.
POEY, D.F. 1978. Los components de rendimiento y su aplicaci6n en la
investigaci6n de cultivos. ICTA, Guatemala. Boletin Tecnico No. 3
STOFELLA, P.J., R.F. SANSTED, R.W. ZOBEL, and W.L. HYMES. 1979. Root
characteristics of black beans. I. Relationship of root size to
lodging and seed yield. Crop Sci. 19:823-826.
TANAKA, A., and K. FUJITA. 1970. Growth, photosynthesis and yield
components in relation to grain yield of the field bean. 3. Faculty
of Agric. Hokkaido University 59:145-238.
WALLACE, D.H., and H.M. MUNGER. 1965. Studies of the physiological basis
of yield differences. I. Growth analysis of six dry bean varieties.
.Crop Sci. 5:343-348.
1966. Studies of the physiological
basis for yield differences. II. Variations in dry matter
distribution among aerial organs for several dry bean varieties. Crop
WALLACE, D.H. 1973. Commentary upon plant architecture and physiological
efficiency in the field bean. In: Potential of field beans and other
legumes in Latin America. Centro Internacional de Agricultura
Tropical, CIAT. Cali, Colombia pp 287-294.
WILLIAMS, W.A., R.S. LOOMIS, and C.R. LEPLEY. 1965. Vegetative growth of
corn as affected by population density. II. Components of growth net
assimilation rate and leaf area index. Crop Sci. 5:215-219.
YOSHIDA, D. 1972. Physiological aspects of grain yield. Ann. Rev. Plant
Table 1. Genetic materials used in the study,
Entry CIAT Other Seed color and
No. Identification Origin No. identification brilliance Habit Category"
1 BAT 41 Colombia Revoluci6n 79 (Nic.) Opaque red 2 !
2 JU-84-7 Guatemala G 18460 Opaque black 2 I
3 ICTA-Quetzal Guatemala DOR 41 Opaque black 2 I
4 Huetar Costa Rica RAO 29 Opaque red 2 I
5 G 3017 Guatemala Mezcla (Guatemala) Opaque mottled purple 2 L
6 A 321 Colombia Opaque cream 3 I
7 Negro Huasteco 81 M6xico DOR 60 D-145 (Guatemala) Opaque black 2 1
8 Rojo Nacional Nicaragua G 17661 Criollo (Nicaragua) Brilliant red 3 L
9 6 2858 M&xico Zacaticano (M6xico) Semi-brilliant 3 L
10 BAT 304 Colombia Brunca (Costa Rica) Opaque black 3 I
11 Rojo de Seda El Salvador G 4090 Criollo (El Salvador) Semi-brilliant red 3 L
12 Desarrural Honduras G 4477 C-0-1-63A (Honduras) Brilliant red 3 L
13 XAN 112 Colombia Opaque black 2 I
14 Rabia de Gato Guatemala G 2997 Opaque-black 2 L
15 Pata de Zope Guatemala G 17648 Opaque black 3 L
16 XAN 145 Colombia Opaque black 2 I
I = improved; L = Land race
Table 2. Meanings of acronyms used in Tables 3-13.
Days to flowering
Days to physiological maturity
Maximum total dry weight
Leaf area duration
Maximum length of stem
Maximum leaf area index
Crop growth rate 30 days after emergence
Net assimilation rate 30 days after emergence
Relative growth rate 30 days after emergence
Pod filling period
Pods per plant
Seeds per pod
Yield per day
Seed dry weight in the stem
Seed dry weight in the branches
Table 3. Parameters of size and efficiency of early and late genotypes. Average data of three replications.
Experiment A. CIAT-Palmira, 1985.
MAXTDW LAD NODES/m= MAXLEN MAXHEI MAXLAI CGR30 NAR30 RGR30
(g/m2) AT MAT. (cm) (cm) (m2/m2) (g!m2/d) (g/m2/d) (gig/d)
8 3017 136.4 34.2 310.8 51.4 43.7 1.413 4.85 3.95 0.0530
Pata de Zope 132.0 31.1 234.0 53.2 48.9 1.417 4.95 4.61 0.0641
Rabia de Sato. 232.4 47.4 262.5 67.6 50.2 1.907 7.49 4.99 0.0687
Huetar 157.2 42.3 292.4 41.5 37.4 1.663 4.61 3.65 0.0549
G 2858 253.2 53.7 612.2 94.0 34.9 1.511 6.63 4.36 0.0533
XAN 145 192.6 49.9 365.0 69.2 54.9 1.768 6.87 4.58 0.0667
Desarrural 214.1 64.6 548.6 96.0 42.3 2.101 5.62 3.13 0.0487
Rojo de Seda 263.9 74.0 369.5 95.4 48.3 2.445 8.56 4.46 0.0734
XAN 112 182.4 46.8 384.4 71.8 52.2 1.589 6.05 5.11 0.0698
Rojo Nacional 160.3 47.3 498.6 90.0 38.9 1.690 4.56 3.34 0.0541
BAT 41 193.8 53.4 420.6 64.0 45.7 1.732 5.60 3.65 0.0575
JU-84-7 322.3 70.1 387.4 114.1 52.9 2.114 7.96 5.20 0.0823
BAT 304 320.9 77.3 328.5 118.5 55.1 2.312 9.99 5.53 0.0808
Negro Huasteco 81 256.9 63.0 387.8 64.0 53.7 1.989 6.92 5.06 0.0779
ICTA-Quetzal 286.4 76.2 509.4 84.2 63.7 2.447 8.58 5.21 0.0858
A 321 397.4 108.3 547.1 184.6 49.8 3.164 9.14 4.01 0.0706
.x 231.4 58.7 403.7 84.8 48.3 1.954 6.77 4.43 0.0663
S.D. 74.9 19.5 110.9 34.3 7.51 0.467 1.72 0.74 0.01!8
Meanings of acronyms are in Table 2.
Table 4. Parameters of size and efficiency of early and late genotypes. Average data of three replications.
Experiment B. CIAT-Palmira, 1985.
MAXTDW LAD NODES/m2 MAXLEN MAXHEI MAXLAI CGR30 NAR30 RGR30
(g/m2) AT MAT. (cm) (cm) (o2/m2) (g/m2/d) (g/m2i d) (g/g/d)
8 3017 500.0 91.9 455.0 106.0 46.2 3.584 13.1 4.26 0.0618
Pata de Zope 431.0 99.3 303.0 105.2 59.2 3.785 14.5 4.26 0.0655
Rabia de Sato 483.0 101.9 419.0 94.7 55.9 3.773 16.0 4.91 0.0732
Huetar 383.0 98.5 425.9 66.5 57.2 3.797 13.8 4.13 0.0686
G 2858 493.0 101.0 708.9 114.3 38.6 3.215 14.4 4.80 0.0625
XAN 145 479.0 111.7 487.7 104.9 62.3 3.983 13.7 4.20 0.0677
Desarrural 481.0 141.8 538.8 117.4 45.9 4.599 13.7 3.64 0.0640
Rojo de Seda 491.0 161.6 459.5 135.9 53.6 5.281 15.2 3.57 0.0665
XAN 112 428.0 112.8 399.3 120.5 64.5 3.993 10.3 3.40 0.0520
Rojo Nacional 411.0 123.3 649.4 110.1 46.0 4.535 11.7 3.45 0.0649
BAT 41 474.0 126.9 499.3 103.8 61.2 4.779 13.6 3.70 0.0614
JU-84-7 545.0 143.7 408.7 152.8 60.1 4.941 16.4 4.79 0.0806
BAT 304 648.0 144.9 443.0 160.6 59.7 4.89.8 15.1 4.03 0.0665
Negro Huasteco 81 586.0 182.9 443.0 120.2 81.9 5.823 13.1 3.51 0.0669
ICTA-Quetzal 616.0 173.1 523.6 125.2 81.3 5.664 12.8 3.48 0.0687
A 321 653.0 196.8 596.5 191.4 61.2 5.475 12.8 3.02 0.0559
x 506.4 132.0 485.0 120.6 58.4 4.508 13.8 3.95 0.0654
S.D. 82.0 32.9 101.1 29.0 11.6 0.802 1.54 0.56 0.0065
Meanings of acronyms are in Table 2.
Table 5. Correlation coefficients among some characteristics of plant efficiency, maturity (DPM), and
-yield (YIELD), during two periods for n = 16 (upper part) and n = 12 (lower part). CIAT-Palmira, 1985.
Experiment A Experiment B
Characteristic DPM YIELD DPM YIELD
(days) (kg/ha) (days) (kg/ha)
Maximum total dry weight (g/m2) 0.79*** 0.87*** 0.66** 0.87***
0.60* 0.81** 0.11 0.81**
Leaf area duration 0.86*** 0.83*** 0.87*** 0.72**
0.76** 0.62* 0.69* 0.41
Total nodes at maturity (n/m2) 0.55* 0.21 0.43 0.21
0.59* 0.07 0.47 0.05
Maximum length of stem (cm) 0.68** 0.83*** 0.68** 0.81***
0.74** 0.59* 0.50 0.65*
Maximum height (cm) 0.52* 0.35 0.47 0.29
0.15 0.40 0.05 -0.12
Maximum leaf area index (m2/m2) 0.79*** 0.78*** 0.79*** 0.64**
0.54 0.51 0.59* 0.42
*, **, *** Significant at 0.05, 0.01 y 0.001, respectively.
Table 6. Correlation coefficients among some characteristics of plant efficiency, maturity (DPM), and
.yield (YIELD), during two periods for n = 16 (upper part) and n = 12 (lower part). CIAT-Palmira, 1985.
Experiment A Experiment B
Characteristic DPM YIELD DPM YIELD
(days) (kg/ha) (days) (kg/ha)
Yield/day (kg/ha/day) 0.58* -0.03
Harvest index (7.) 0.25 0.64t* -0.29 0.19
0.05 0.40 -0.28 0.39
Crop growth rate (g/m2/day) 0.67** 0.75*** -0.34 0.21
0.42 0.68* -0.31 0.57
Net assimilation rate (g/m2/day) 0.28 0.30 -0.71** -0.19
0.04 0.44 -0.56 0.40
Relative growth rate (g/g/day) 0.61* 0.50* -0.27 0.18
0.31 0.60* -0.23 0.65*
*, **, *** Significant at 0.05, 0.01 y 0.001, respectively.
Rates at 30 i~ys after emergence.
Table 7. Analysis of variance for growth parameters during two periods. CIAT-Palmira, 1985.
Experiment A Experiment B
Mean square C.V. (.) Mean square C.V. (7.)
Days to flowering 23.5388** 2.3 15.4819** 1.8
Days to physiological maturity 55.6569** 1.2 65.1555** 0.7
Maximum total dry weight 12340** 17.3 15745 17.5
Leaf area duration -854.87** 14.3 2339.0** 8.7
Total nodes at maturity 30213** 14.2 21658** 11.0
Maximum length of stem 2704.36** 10.2 1899.39?** 8.2
Maximum height 137.90** 6.9 314.38** 6.2
Maximum leaf area index 0.4871** 16.4 1.3843** 3.9
Crop growth rate 6.379 21.6 5.586 15.9
Relative growth rate 0.00031* 17.1 0.000099 12.7
Net assimilation rate 1.218 20.4 0.784* 15.1
Harvest index 52.29** 6.0 50.29* 7.3
Yield 53795** 17.8 316306* 14.0
Yield per day 89.966** 17.7 54.51 14.0
*, ** Significant at 0.05 and 0.01, respectively.
Rates at 30 days after emergence.
Table 8. Parameters of development and yield of early and late genotypes during two
three repetitions. CIAT-Palmira, 1985.
periods. Average data of
_Experiment A -Experiment 8__
FLOWER DPM HI YIELD YIELD/DAY FLOWER DPM HI YIELD YIELD/DAY
(days) (days) (7.) (kg/ha) (kg/ha/d) (days) (days) (7) (kg/ha) (kg/ha!d)
G 3017 29.2 57.3a 57.5 872 15.5 28.5 55.9a 66.4 2616b 47.2
Pata de Zope 30.5 57.5a 52.9 510 9.1 29.1 56.Oa 54.5 2364 42.1
Rabia de Gato 31.4 60.3b 55.8 990 16.2 30.7 58.0b 59.7 2509b 43.1
Huetar 33.8 60.4b 40.8 723 11.8 32.2 57.9b 62.3 2034 35.3
6 2858 31.8 62.9c 55.0 864 13.7 29.6 63.0c 56.2 2410b 38.2
XAN 145 37.7 63.2 55.9 983 15.6 33.9 64.9 56.7 2427b 37.5
Desarrural 33.5 63.9 52.7 848 13.2 31.7 65.1 64.8 2534b 38.9
Rojo de Seda 38.0 64.3 54.4 901 14.1 34.3 65.8 55.3 2449b 37.3
XAN 112 36.1 65.2 50.8 750 11.6 33.7 66.0 49.9 2064 31.3
Rojo Nacional 37.2 64.9 55.4 679' 10.5 34.7 65.8 63.8 2490b 37.8
BAT 41 35.6 64.5 49.3 823 12.8 33.3 65.2 57.5 2524b 38.7
JU-84-7 36.3 66.2 56.8 1555c 23.3 32.9 63.1 60.9 3128ab 49.3
BAT 304 35.9 68.1 61.0 1887b 27.7 32.3 67.5 61.4 2916ab 43.1
Negro Huasteco 81 39.2 70.0 54.1 1315c 18.9 37.2 69.7 51.9 2687b 38.7
ICTA-Quetzal 39.5 72.7 49.5 941 12.8 37.0 70.6 53.6 3048ab 43.3
A 321 38.2 73.9 61.6 2445a 33.1 35.5 74.0 60.8 3467a 46.9
x 35.2 64.7 53.9 1068 16.2 32.9 64.3 58.5 2604 40.6
S.D. 3.15 4.77 4.93 503 6.55 2.59 5.19 4.77 376 4.79
Values followed by the same letter are
Meanings of acronyms appear in Table 2.
statistically equal at 0.05.
Table 9. Parameters of development and yield and its components of early and late
repetitions. Experiment A. CIAT-Palmira, 1985.
genotypes. Average of three
Genotype FLOWER DPM PFF P/P S/P SW HI YIELD YIELD!DAY
(day(day) days) (days) (nm ) (.) (kg!ha) (ko!ha!d)
6 3017 29.2 57.3a 28.2 10.3 3.88 163.3 57.5 872 15.5
Pata de Zope 30.5 57.5a 26.9 8.3 3.72 157.8 52.9 510 9.1
Rabia de Gato 31.4 60.3b 28.9 11.1 4.20 177.2 55.8 990 16.2
Huetar 33.8 60.4b 26.6 9.0 4.18 151.6 40.8 723 11.8
6 2858 31.8 62.9c 31.1 8.8 2.92 300.0 55.0 864 13.7
XAN 145 37.7 63.2 26.5 10.1 3.73 147.5 55.9 983 15.6
Desarrural 33.5 63.9 30.4 7.6 3.62 215.4 52.7 848 13.2
Rojo de Seda 38.0 64.3 26.3 10.9 3.57 233.2 54,4 901 14.1
XAN 112 36.1 65.2 29.1 10.4 3.59 174.9 50.8 750 11.6
Rojo Nacional 37.2 64.9 27.7 9.6 3.48 159.9 55.4 679 10.5
BAT 41 35.6 64.5 28.8 8.1 3.70 138.2 49.3 823 12.8
JU-84-7 36.3 66.2 29.9 15.0 4.80 172.6 56.8 1555c 23.3
BAT 304 35.9 68.1 32.2 15.1 5.07 177.8 61.0 1887b 27.7
Negro Huasteco 81 39.2 70.0 30.9 11.7 4.91 177.9 54.1 1315c 18.9
ICTA-Duetzal 39.5 72.7 33.1 11.4 4.51 142.0 49.5 941 12.8
A 321 38.2 73.9 35.7 11.8 4.37 329.3 61.6 2445a 33.1
x 35.2 64.7 29.5 10.6 4.02 188.7 53.9 1068 16.2
S.D. 3.15 4.77 2.64 2.17 0.59 55.3 4.93 503 6.55
Values followed by the same letter are
Meanings of acronyms are in Table 2
statistically equal at 0.05.
Parameters development and yield and its components of early and late genotypes.
repetitions. Experiment B.
Genotype FLOWER DPM PFF P/P S/P SW HI YIELD YIELD/DAY
(days) (days) (days) (mg) (:) (kg/ha) (kg/ha/d)
8 3017 28.5 55.9a 27.4 17.6 4.69 218.5 66.4 2616b 47.2
Pata de Zope 29.1 56.0a 26.9 13.8 4.64 221.5 54.5 2364 42.1
Rabia de Gato 30.7 58.0b 27.2 16.1 4.25 226.3 59.7 2509b 43.1
Huetar 32.3 57.9b 25.7 15.4 4.74 190.1 62.3 2034 35.3
G 2858 29.6 63.Oc 33.4 13.9 3.10 385.4 56.2 2410b 38.2
XAN 145 33.9 64.9 30.9 18.3 4.57 193.7 56.7 2427b 37.5
Desarrural 31.7 65.1 33.3 15.2 4.13 284.3 64.8 2534b 38.9
Rojo de Seda 34.3 65.8 31.5 10.1 4.77 313.3 55.3 2449b 37.3
XAN 112 33.7 66.0 32.3 16.2 3.87 223.5 49.9 2064 31.3
Rojo Nacional 34.7 65.8 31.1 15.0 4.51 227.1 63.8 2490b 37.8
BAT 41 33.3 65.2 31.9 17.9 4.94 194.7 57.5 2524b 38.7
JU-84-7 32.9 63.1 30.2 17.9 5.39 223.9 60.9 3128ab 49.3
BAT 304 32.3 67.5 35.2 19.3 5.49 237.1. 61.4 2916ab 43.1
Negro Huasteco 81 37.2 69.7 34.5 12.0 6.21 229.5 51.9 2687b 38.7
ICTA-Quetzal 37.0 70.6 33.6 16.5 5.52. 211.9 53.6 3048ab 43.3
A 321 35.5 74.0 38.5 17.9 4.86 357.5 60.8 3467a 46.9
x 32.9 64.3 31.4 15.8 4.75 246.1 58.5 2604 40.6
S.D. 2.59 5.19 3.33 2.48 0.72 58.3 4.77 376 4.79
Values followed by the same
Meanings of acronyms are in
statistically equal at 0.05.
Average of three
Table 11. Maximum height (MAXHEI), maximum length of stem (MAXLEN), seed dry wright in the stem (SDWS), and in
the branches (SDWB) during two periods. Average of three replications. CIAT-Palmira, 1985.
Experiment A Exoeriment B
MAXHEI MAXLEN SDWS SDWB MAXHEI MAXLEN SDWS SDWB
(cm) (cm) (g/m2) (g/mb ) (cm) (cm) (g/m2) (g/2)
6 3017 43.7 51.4 46.2 106.01 168.0 85.5
Pata de Zope 48.9 53.2 59.2 105.2 183.0 9.7
Rabia de Gato 50.2 67.6 55.9 94.7 110.3 55.2
Huetar 37.4 41.5 57.2 66.5 52.8 35.7
G 2858 34.9 94.0 38.6 114.3 73.3 152.0
XAN 145 54.9 69.2 62.3 104.9 87.5 170.5
Desarrural 4 42.3 96.0 45.9 117.4 90.2 162.7
Rojo de Seda 48.3 95.4 -- 53.6 135.9 73.2 177.3
XAN 112 52.2 71.8 64.5 120.5 93.5 116.2
Rojo Nacional 38.9 90.0 46.0 110.1 53.2 166.7
BAT 41 45.7 64.0 61.2 103.8 94.0 178.5
JU-84-7 52.9 114.1 60.1 152.8 187.3 69.2
BAT 304 55.1 118.5 59.7 160.6 140.0 135.7
Negro Huasteco 81 53.7 64.0 81.9 120.2 159.2 89.3
ICTA-Quetzal 63.7 84.2 81.3 125.2 123.2 119.3
A 321 49.8 184.6 61.2 191.4 118.7 216.8
x 48.3 84.8 -58.4 120.6 113.2 121.3
S.D. 7.51 34.3 11.6 29.0 43.6 58.9
Table 12. Analysis of variance for development, yield, and growth data during two periods. CIAT-Palmira, 1985.
Experiment A Experiment B
Mean square C.V. (.) Mean square C.V. (7.)
Days to flowering
Days to physiological maturity
Pod filling period
Yield per day
Maximum length of stem
Seed dry weight in the stem
Seed dry weight in the branches
*, **, *** Significant at 0.05, 0.01 and 0.001, respectively.
Table 13. Correlation coefficients among yield (YIELD), its components, and harvest index (HI) during two periods
for n = 16 (upper part) and'n = 12 (lower part). CIAT-Palmira, 1985.
Experiment A Experiment B
S/P SW HI YIELD S/P SW HI YIELD
Pods/plant (P!P) 0.78*** 0.02 0.53* 0.70** 0.09 -0.32 0.38 0.38
0.67* -0.07 0.41 0.83*** 0.27 -0.63* 0.26 0.29
Seeds/pod (S/P) -0.22 0.20 0.63** -0.44 -0.07 0.51*
0.55 -0.06 0.62* -0.67* 0.29 0.47
Seed weight (SW) 0.44 0.51* 0.01 0.28
0.24 0.07 -0.14 0.05
Harvest index (HI) 0.64** 0.19
*, **, *** Significant at 0.05, 0.01 and 0.001, respectively.
Meanings of acronyms are in Table 2.
+ = LATE MATURING GENOTYPES
5 .4 L6
*;- C- -. '*
.5 0 0.
55 60 65 70
DAYS TO PHYSIOLOGICAL MATURITY
Fig. 1. Relationship between seed yield/day and number of days to maturity during two planting
seasons. Means from three replications. CIAT-Palmira. 1985.
Sr = 0.58*
r2 = 0.28
rl = -0.03
r = -nfl.
4+ LATE MATURING GENOTYPES
ri = -0.53*
r = -0.7g90
0.16 0.20 0.24 0.28 0.32 0.19 0.22 0.26 0.30 0.34
Fig. 2. Relationship among number of seeds/m2 and seed weight during two plantings. Continuous lines
rAnrnsnnt. constant vield. means of three reDlications. CIAT, Palmira, 1985.
170 g/m' 6
STRATEGIES AT CIAT FOR INCREASING YIELD POTENTIAL OF COMMON BEANS
--- FINISHING THE DOMESTICATION PROCESS
Jeffrey W. White**
Since the founding of the Bean Program at CIAT, there has existed interest
in finding ways to increase the-yield potential of beans (Adams, 1973;
Wallace, 1973). This interest has continued to the present day, and if
anything, has increased with successes in controlling other production
The present strategy of the Program contains two basic components. One
consists of continuing breeding efforts based on selection for yield,
considering that, to date, it has not been possible to identify reliable
morphological or physiological parameters for use as selection criteria for
yield. The other component is simply to continue the search for such
characteristics. It is this second component which provides the subtitle
for this presentation since the present philosophy of research on selection
criteria is that beans are still a very rustic crop, adapted to cropping
systems with low plant densities, infertile soils, strong weed competition,
and other all too familiar problems. In order to increase yields,
cultivars are needed which are adapted to agronomic management which
reduces such stresses.
In this presentation, basic aspects of the two strategies will be discussed
after a brief review of some strategies notable for their failure. It
should be appreciated that the intent is to describe the efforts of CIAT in
SReport presented at the XXXIII Meeting of the PCCMCA, Ciudad de
Guatemala. 30 April 1987.
Physiologist, CIAT, A.A. 6713, Cali, Colombia.
the context of the agro-ecological conditions of its experiment stations,
and not to present a universal formula for solving the yield potential
Suggestions on ways to increase yield potential of beans have been
numerous. Although CIAT is far from having investigated all the
possibilities, it has tried a great many.
Applying yield components in their simplest form, it is conceivable that
simply by intermating materials with different levels of yield components
and selecting recombinants with high values of certain components, new,
higher yielding materials may be developed. This strategy assumes that
each component is under independent genetic control. Unfortunately, as
noted by Adams (1967), yield components generally show strong interactions
such that an increase in one component tends to be associated with a
decrease in others. Adams called this phenomenon "yield component
compensation", and noted that its occurrence suggested that there is
competition for limited resources within a bean plant. Such compensation
has been detected in many studies at CIAT, both with bred lines (e.g. White
1981; Fig. 1) and in segregating populations (Nienhuis and Singh, 1985;
Given this reality, work at CIAT has steered away from this approach,
although admittedly it remains to be determined whether an optimum
combination of yield components can be defined for beans.
Biomass or Size:
In agreement with work on yield components, a consistent tendency for
larger plants to yield more has been noted, independent of whether size is
measured as biomass (dry weight), leaf area duration, or some other size
parameter (Table 2). It could be said that under conditions at CIAT, bean
yields are limited by supply, that is source strength, rather than demand,
or sink strength. This has suggested two "sub-strategies": search for
cultivars which grow more rapidly or for cultivars which, because they grow
for a longer period, have more growth.
For the first alternative, various approaches have been considered. The
simplest is simply to increase the rate of photosynthesis. However,
studies by Kueneman and co-workers (1979) indicated that selection for
increased leaf photosynthetic rates would be ineffective in beans. Similar
results in other crops (Gifford and Evans, 1980) have served to dissuade
CIAT from following this approach further. 'Nonetheless, it is clear that
photosynthesis in beans requires further study. Comparing leaf
photosynthetic rates reported over 20 years, one notes a tendency for rates
to increase with time (Table 3). Presumably these do not reflect genetic
changes, it being more likely that measurement techniques have improved.
It is also interesting to note that in work with Viona munoo, a
significant correlation was found between yield and photosynthetic rate
during podfill (r = 0.66, significant at the P=0.01 level, but not during
other growth phases (Chandra et al., 1985).
Another possibility for increasing growth rate would be to achieve more
efficient light interception. Among various alternatives, greater lodging
resistance seems the most crucial under conditions at CIAT. In a study on
effects of delayed maturity on yield (Masaya, personal communication),
there was no clear treatment effect with lodging, but with a trellis which
prevented lodging, a mean yield increase of 16% was obtained (Table 4). As
a further observation, it merits note that in almost all studies at CIAT on
yield potential, lodging appears to occur sufficiently early to affect crop
growth and yield.
Finally, one might attempt to increase biomass accumulation through
extension of the crop growth cycle. Work using long days to induce delayed
maturity has demonstrated the potential for increasing yields through late
maturity (Laing et al., 1984; Masaya, personal communication).
Nonetheless, notwithstanding 7 or 8 years of efforts to select late
maturing materials at CIAT, no materials combining late maturity with
higher yields have been obtained. The main problem appears to be that
selection for lateness results in unintentional selection for materials
highly sensitive to photoperiod, and in reality, such materials are late
because of their poor adaptation. This suggests that future selection for
late maturity should restrict selection to materials previously shown to
have good adaptation, and at the same time, use selection criteria which
minimize the chances of selecting materials with poor adaptation.
Besides trying to increase total biomass, one might increase the efficiency
of distribution of dry weight, the most obvious approach being to increase
harvest index. However, various lines of evidence suggest that this
strategy holds little promise for conditions at CIAT.
The first difficulty is that in contrast to crops such as wheat and rice,
where harvest indices of traditional and improved cultivars vary between 30
and 55% (Kertesz, 1984), the normal range of variation at CIAT is generally
on the order of 50 to 60%. This suggests that harvest index in beans is
already very high. Indeed, one might ask whether harvest indices are too
high if this increases tendency for lodging.
Lack of variability -on harvest index is also reflected directly in the
rarity of trials which show a significant correlation between yield and
harvest index (Table 2). This scarcity is actually worse than it seems
when one considers that the statistical artifact introduced by correlating
the two parameters should make the correlations artificially greater. The
value for biomass includes yield (biomass = vegetative weight + seed
yield), so one usually expects a large correlation although it is
impossible to determine its significance level. This effect is so strong,
that it has been shown that if pairs. of random numbers are used to
calculate hypothetical values of harvest index, a high correlation may be
obtained (Charles-Edwards, 1982). This point is seldom discussed by
advocates of harvest index, so we offer Figure 2 as an example of a "highly
significant" correlation generated with 20 pares of random numbers.
Perhaps the argument of greatest weight against use of harvest index is
that in comparing its effectiveness as a selection parameter, it has proven
inefficient compared to simpler alternatives (Table 6). Plant breeders
working with other crops appear to have had similar experiences. For
example, Kramer (1984), working with wheat, concluded that harvest index
served as a selection criterion only when one did not expect variation in
total biomass, and that in that case, selection directly for individual
plant yield is equally effective, while requiring less labor.
Other possibilities for increasing yield have also been considered: search
for an optimal combination of morphological characteristics (White, 1981),
increase the efficiency of pollination, increase nutrient availability
(especially nitrogen) through foliar fertilization, among others. Although
some of these approaches have not been discarded, they have not changed the
basic conclusion that, under conditions at CIAT, bean crops always seem to
show strong limitations of assimilate availability ("source"), while demand
("sink strength") is usually adequate.
Breeding for Increased Yield
Confronted with the absence of reliable selection criteria, efforts on
yield breeding at CIAT have evolved a strategy of selection directly for
yield. To optimize the selection process, studies on ways to identify
promising parents and optimal agronomic practices for selection have been
To date, the principle approach for identification of good parents has been
use of combining ability studies. In a study of 80 lines grown at Palmira
and Quilichao, a great range of variability in general and specific
combining ability was detected. While materials such as A 375 and BAT 477
showed high, positive general combing abilities (GCA's), some well known
cultivars such as Jamapa and Porrillo Sintetico had negative GCA's
suggesting little promise as parents (Table 7).
Studies on nursery management are still in preliminary stages, but it is
hoped to define optimal conditions for yield selection. One of the first-
aspects being investigated is efficiency of individual plant selection
under different plant populations.
Given the variability in GCA, a strategy which emphasizes selection among
populations is being pursued, with a deemphasis in individual plant
selection in early generations. It is hope that evaluating yields of
populations from F2 onwards will not only result in production of
populations with high frequencies of desirable genotypes, but
identification of parents with greatest GCA. In practical terms, yield
tests are being performed from the F2 to the F4, using a bulk-pedigree
method (Fig. 3). For each group of crosses (e.g. for a combination of
grain types and diseases resistances), populations are planted in lattice
designs with two replicates. Agronomic management is identical to that
used in yield trials of advanced lines, including populations of 200,000
plants/ha and bordered yield plots. To assure sufficient seed for
subsequent generations, bulks are formed by harvesting one pod per plant.
In the F5, a spaced planting is used to permit individual plant selection.
Seed of the F6 is again used in replicated yield trials, and in F7 final
yield tests in three locations (Palmira, Quilichao and Popayan) are
performed before distribution to national programs and CIrT uniform
To accelerate the accumulation of desirable genes, a recurrent selection
strategy is also being followed. It is similar to the preceding system,
but differs in specific details (Fig. 4). Among these are yield tests in
the FI, intercrossing among the most promising populations, individual
plant selection in F3, and return to intermating in the F6 generation.
Both breeding strategies require large quantities of Fl seed and imply
large numbers of yield plots. This implies a large expenditure of
resources of the Bean Program, but it was felt better to continue with a
difficult and costly approach which shows promise, rather than continue
with strategies which did not seem to produce the desired results.
In summary, it can be said that the main yield breeding strategy is based
on development of populations with high mean yields, the work being guided
by studies of combining ability. However, it is worth noting that this
approach has been criticized for ignoring the potential importance of
variablity within populations, and favoring parents with good competitive
ability (probably type III's). The need for continuing studies on
selection strategies is apparent.
The Physiology Strategy
Although it is hoped that yields can be increased through direct selection
for yield, a need is still seen for identifying more efficient selection
criteria. Identifying better criteria should facilitate the ongoing
selection process or help identify particularly promising parents. And
there is also no guarantee that seeking progressive increases in yield will
be as productive as redesigning the crop, such as has been done in other
crops such as rice (Chandler, 1968), wheat (Evans et al., 1975) and soybean
In work at CIAT an attempt has been made to investigate various possible
strategies for increasing yield potential, but at the present, the main
effort is centered upon the hypothesis that traditional bean cultivars are
adapted to cropping systems which imply sacrificing yield potential in
exchange for reduced risk, production costs or other problems. We can say
that traditional systems represent a state of semi-domestication where bean
plants are obliged to expend a high portion of resources simply in
surviving among problems of weed competition, low soil fertility, drought
stress, and attacks of insects and diseases. If we wish to increase yield
potential dramatically, we should not only search for new plant types, but
Provide then with adequate agronomic conditions where stresses are
minimized. This reality has lead physiological studies to pursue two
parallel lines of investigation, one of modification of agronomic practices
and the other of searching for genotypes which are adapted to such
The Agronomic Component
In order to define agronomic practices for a type of bean cultivar which
still may not exist, attention has been given to practices such as
increased density, reduced spacing between rows, and increased fertilizer
application, which have accompanied yield increases in the crops. In
commercial soybean crops planted at a row spacing of 0.17 m, populations of
700.000 plants.ha-' are now in use (Cooper, 1985). Data for cereal crops
are difficult to extend to beans, but the importance given to reduced row
spacings and increased populations is clear (Chandler, 1968; Duncan,
1968). Such changes represent a drastic shift from traditional practices
at CIAT with 0.6 m row spacings and maximum populations of 250.000
plants.ha-1. However, given that the practices being sought are for
genotypes which still may not exist in common bean, little importance is
being given to previous recommendations for practices with beans.
The first trials performed within the framework of this strategy compared
response of traditional and erect genotypes to row spacing and po-pulation.
In a trail with four row spacings, but conserving a constant popLlation of
200.000 plants.ha-1, crop dry weight at 56 days increased 20% in erect
genotypes and 14. in traditional materials, comparing the 0.6 m spacing
(traditional in CIAT) and 0.2 m spacing (Table 8). However, at harvest,
the 0.2m spacing gave the lowest yields, the 0.45 m spacing offering an 11%
increase in traditional lines, and only 3% increase with erect lines. This
suggested that the treatments were effective in producing an increase in
biomass, but that other factors during podfill or maturation reduced the
efficiency of conversion of dry matter to yield. The most probable
candidate was lodging, and this is being subjected to further study.
In any case, the data demonstrate the ability of traditional materials to
adapt to a wide range of agronomic practices. Fig. 5 also illustrates this
phenomenon for four early maturing lines where canopy cover at 18 days was
much greater with a 0.2 m spacing, but as the cultivars grew, differences
among row spacing treatments disappeared. For the moment, a row spacing of
0.3 m has been adopted as a provisional standard for intensive agronomy.
Assuming that this spacing is preferable to a 0.6 m spacing, it was
expected that plant populations could be increased since the 0.3 m spacing
permits a better spacial distribution of plants. In the first two trials
evaluating effects of increased population at different row spacings a
significant spacing by density interaction was detected in the first trial,
but not in the second, although yields were always higher at 0.3 m row
spacing (Table 9).
For the moment, 300.000 plants.ha-1 is being used with the 0.3 m spacing,
but it is recognized that this value is artificial in the absence of the
genotypes we are searching for.
As a third component of agronomic management, yield increases have been
sought through split nitrogen applications. Given that the majority of
studies at CIAT-Falmira have indicated that nitrogen is not limiting, and
that one of the supposed attractions of bean crops is their ability to fix
nitrogen, this sub-strategy requires further justification.
First, it was noted that in a 1 ton.ha-' yield increase represents an
increase of 220 to 230 kg of protein, which in turn implies making an
additional 15 to 16 kg.ha-1 of nitrogen available to the crop. Assuming
losses due to volatilization, leaching and nitrogen in crop residue, the
total additional N required could easily be 30 kg.ha-1
Studies in soybean and common bean suggest that the same high nitrogen
requirement is a key factor in determining the drop in efficiency of
photosynthesis during podfill (Sinclair and de Wit, 197 ; Tanaka and
Fujita, 197 ). Just when the crop would benefit most from a high
photosynthetic rate, nitrogen is remobilized out of leaves to supply
growing pods. This results in a very close relation between leaf
senescence and pod set. Comparing weights of leaves and pods at maturity
(relative to total plant weight) under various levels of pod removal, a
linear relation between leaf and pod weight was found (Fig. 6).
Although it would be preferable for a bean crop to provide its own nitrogen
through fixation, we have assumed for the moment that additional N will
have to be supplied through fertilizer, but with special emphasis on making
it available in a manner which will assure efficient use. Foliar
applications were discarded given doubts concerning their adverse affects
(Gray and Akin, 1984), and for the moment, emphasis is being given to
weekly applications of urea applied in solution to the soil. Comparing
treatments of 10 and 20 kg of N per week starting at onset of flowering, N
applications were associated with a 10% yield increase with 0.6m row
spacing, but gave a 10% yield reduction for the 0.3 m spacing (Fig. 7). A
possible explanation is that the N treatment or 0.3 m spacing function
separately to increase leaf growth, and thus yield, but when combined,
their effect is to produce a supra-optimal level of leaf development, or a
very vegetative plant with little lodging tolerance.
In all of the agronomic study, it has been difficult to interpret results
due to doubts about using in plant types which do not conform to our ideal.
To remedy this situation in future studies, an attempt will be made to
reduce lodging through artificial supports. In addition, growth regulators
such as CCC or B-9 will be used to produce plants of short stature and less
tendency to lodge.
The varietal Component:
The preceding discussions underline the urgent need.to find materials with
improved response to narrow row spacing and higher populations. Taking as
a reference point the 0.3 m row spacing with 30 plants m-2, materials from
.the VEF and breeders' nurseries are now being screened for such a response.
Since such nurseries are managed with traditional spacings, selection
criteria are limited to erect habit (type 2a), pod load and general
adaptation. No use is made of yield data because of the fear that this
will again lead to selection of traditional plant types adapted to wide
spacings and low densities. From the nurseries, materials are passed to
yield trials (lattices with 3 replicates) managed with narrow rows.
When agronomicc practices have been further refined, and a first group of
materials is identified for outstanding performance under these conditions,
breeding for this system will be started. The basic strategy could simply
be to follow the schemes outlined in Fig. 3 and 4, but employing the
intensive management practices. It would also be useful to select for
specific characters such as lodging resistance and nitrogen response. To
date, materials of growth habit type I have been ignored in our studies,
but given the success with determinate soybeans (Cooper, 1985), these also
might merit consideration.
A special problem for breeding efforts is possible adverse affects of plant
competition during the selection process. If one accepts that a desirable
characteristic for traditional cropping systems is outstanding competitive
ability, while one of the advantages of the proposed new materials would be
reduced competitiveness, it is easily imagined that a selection process
based on individual plant vigor or repeated use of bulks could prove
counterproductive. This problem has been well documented in rice (Jennings
and de Jesus, 1968), and Jennings (1968) suggested that it may be avoided
by thinning out the most vigorous plants with large leaves at the onset of
flowering. In CIAT, some researchers have argued that levels of
competition are not so extreme in beans, so the problem would be much less
severe. However, plant death at high densities is easily demonstrable, and
presumably reflects effects of excessive competition (Fig. 8).
While we are confident that ways can be found to increase bean yield
potentials, and that searching for erect architecture combined with
appropriate agronomic practices is a promising way to achieve this increase,
there is little doubt that the increases will not come easily. Since the
introduction of the wheat cv. Norin 10 as a source of genes for dwarf
architecture in 1946, 15 years passed before the release of the first
commercial dwarf wheat in the U.S.A. (Gale and Youssefian, 1985). In the
work of Cooper (1985) on dwarf, semi-determinate soybeans, it took 9 years
to produce cv Elf, the first commercial material. Extrapolating to beans,
it seems likely that yield increases will come slowly, and will reflect the
accumulation of effects of various modifications in cultivars and crop
management (Table 10).
Another worry is that the improved materials may only produce well in the
most favorable environments. This is illustrated by comparison of
hypothetical cultivars grown under optimal and stress conditions (Fig. 9).
Obviously, we still do not know if a given stress -will produce a relatively
greater or lesser yield reduction in comparison to traditional cultivars.
Adams, M.W. 1967. Basis of yield component compensation in crop plants
with special references to the field bean, Phaseolus vulraris. Crop
---- 1973. Plant architecture and physiological efficiency in the field
bean. p. 266-278. In Potential in field beans and other food legumes
in Latin America. CITA, Call, Colombia.
Chandler, R.F. 1968. Plant morphology and stand geometry in relation to
nitrogen. p. 265-289. In J.D. Eastin, et al. (eds.). Physiological
aspects of crop yield. ASA, CSSA; Madison, Wisconsin.
Chandra Babu, R., P.S. Srinivasan, N. Natarajaratnam and S.R. Srce
Rangasamy. 1985. Relationship between leaf photosynthetic rate and
yield in blackgram (Vipna mungo (L) Hepper) genotypes.
Charles-Edwards, D.A. 1982. Physiological determinants of crop growth.
Academic Press; Sydney. 161 pp.
Cooper, R.L. 1985. Breeding semi-dwarf soybeans. Plant Breeding Reviews
Donald, C.M. 1968. The breeding of crop ideotypes. Euphytica 17:385-403.
Duncan, W.G. 1968. Cultural manipulation for higher yield. p. 327-342.
In J.D. Eastin, F.A. Haskins, C.Y. Sullivan and C.H.M. Van Bavel
(eds.). Physiological aspects of crop yield. ASA,CSSA; Madison.
Evans, L.T., I.F. Wardlaw and R.A. Fischer. 1975. Wheat. p. 101-149. In
L.T. Evans (ed.). Crop physiology: Cambridge University Press;
Gale, M.D. and S. Youssefian. 1985. Dwarfing genes in wheat. p. 1-35.
In G.E. Russell (ed.). Progress in plant breeding. I. Butterworths;
Gifford, P.M. and L.T Evans. 1981. Photosynthesis, carbon partitioning
and yield. Ann. Rev. Plant Physiol. 32:485-509.
Gray, R.C. and G.W. Akin. 1984. Foliar fertilization. p. 579-584. In
R.D. Hauck (ed.). Nitrogen in crop production. ASA, CSSA, SSSA;
Jennings, P.R. 1968. Morphology, stand geometry, nitrogen-discussion, p.
286-287. In J.D. Eastin, F.A. Haskins, C.Y. Sullivan and C.H.M. Van
Bavel (eds.). Physiological aspects of crop yield. ASA, CSSA;
---- and J. de Jesus Jr. 1968. Studies on competition in rice. I.
Competition in mixtures of varieties. Evolution 22:119-124.
Kertesz, Z. 1984. Improvement of harvest index. p. 93-104. In W.
Lange, A.C. Zeren and N.G. Hogenboom (eds.). Efficiency in plant
breeding. PUDOC; Wageningen.
Kramer, T. 1984. Harvest index and grainyield as selection criteria in
plant selection, p. 109-112. In W. Lange, A.C. Zeren and N.6.
Hogenboom feds.). Efficiency in plant breeding. PUDOC; Wageningen.
Kueneman, E.A., D.H. Wallace and P.M. Ludfor. 1979. Photosynthesis
measurements of field-grown dry beans and their relation to selection
for yield. J. Amer. Soc. Hort. Sci. 104:480-482.
Laing, D.R.L., P.G. Jones and J.H.C. Davis. 1984. Common bean (Phaseoius
vulqaris L.). p. 305-351. In P.R. Goldsworthy and N.M. Fisher
(eds.). The physiology of tropical field crops. John Wiley & Sons;
Lugg, D.G. and T.R. Sinclair. 1981. Seasonal changes in photosynthesis of
field-grown soybean leaflets. 2. Relation to nitrogen conten.
Molina C, A. 1986. Efecto de la distancia entire surco sobre el rendimiento
en grano y otros parAmetros fisiol6gicos de cuatro cultivares de
frijol precoz (Phaseolus vulqaris L.). Ing. Agr. Thesis, Universidad
Nacional, Palmira, Colombia.
Nienhuis, J. and S.P. Singh. 1986. Combining ability analyses and
relationships among yield, yield components, and.architectural traits
in dry bean. Crop Sci. 26:21-27.
Sinclair, T.R. and C.T. de Wit. 1976. Analysis of the carbon and nitrogen
limitations to soybean yield. Agron. J. 68:319-324.
Tanaka, A. and K. Fujita. 1979. Growth photosynthesis and yield
components in relation to grain yield of the field bean. J. Fac. Agr.
Hokkaido Univ. 59(2):145-238.
Wallace, D.H. 1973. Commentary upon: plant architecture and
physiological efficiency in the field bean. p. 287-295. In
Potential of field beans and other food legumes in Latin America.
CIAT, Cali, Colombia.
White, J.W. 1981. A quantitative analysis of the growth and development
of bean plants (Phaseolus vul! aris L.). Ph.D. thesis. University of
Table 1. Expected direct gain (data in diagonal) and correlated gains under selection" (combined in
two locations) for yield and its components in beans.
Selection criteria Yield Pods/mn Seeds/pod Seed weight
Yield (g/m2) 5.77 -2.28 1.57 6.46
Pods/m2 -2.67 5.06 -6.54 -1.85
Seeds/pod 1.41 -5.04 10.01 -3.3
Seed weight (a/100) 6.84 1.67 -3.89 12.83
Mean performance 147.10 155.70 4.05 23.50
SPercent of mean of characteristic.
Direct gain = k 62, / 6,,p correlated gain = k Cova 12 hi02,2
Table 2. Correlations between yield and harvest index, biomass,
duration in various studies at CIAT.
days to maturity, and leaf area
Harvest Days to Foliar area
Trial index Biomass maturity duration Source
Habits 1 y 2 0.51 0.94** 0.75** 0.88** White, 1981
Habits 1 y 2 0.46** 0.87** 0.41** 0.42** White, 1981
12 trials 0.28 0.96** 0.85** 0.87** Laing et al., 1984
9 grain legume
species 0.50 0.91** 0.86** 0.93** Laing et al., 1984
*, significant at the p = 0.05 and p = 0.01 levels, respectively.
Table 3. Increase in leaf photosynthetic rates in beans reported since
Author' Year Rate
Gaastra 1962 22
Hesketh & Moss 1963 12
!zhar & Wallace 1967. 15
Austin & MacLean 1972 24
Fraser & Bidwell 1974 19
Louwerse & Zweerde 1977 22
Tanaka & Fujita 1979 35
El-Sharkaway et al. 1985 35
SFor complete citation of references not listed in the bibliography see
Table 4. Yields and days to maturity of 8 lines orown under natural (12.5
hours) and extend (13.5 hours) photoperiods, with or without
Source: Masaya, not published.
Photoperiod (hours) __
Days to Days to
Cultivar Yield maturity Yield maturity
Table 5. Comparison of photoperiod response in relation to yield and days
to maturity for promising late maturing lines. Palmira 1984b.
Number of lines with response toohtoperiod
Neutral Intermediate Sensitive
Chi2 = 10.7 p .005
Days to maturity
Chi2 = 7.1 p .05
Table 6. Mean yields from seed of F4 progenies derived from individual selections performed in the
F2 using various selection criteria: HI = harvest index, DW = plant dry weight; DY =daily
yield. Numbers in parenthesis indicate F2 plants in a given category.
Masaya, personal communication.
Parental yield 1 2 3 4 4 plus
Cross Female Male HI DW DY VS 1+2 1+3 2+3
A 429 x 268 256 266 312 283 270 287
XAN 112 (20) (10) (3) (62) (13)
A 429 x 268 259 258 e 286 269 258 317
Pata de ZoDe (16) (9) (23) (2) (4)
A 429 x 268 285 253 253 263 282
Pecho Amarillo (21) (9) (128) (9)
Table 7. Effects of general combining ability for yield in various
cultivars and advanced lines evaluated at two sites in Colombia.
Source: Nienhuis y Singh, 1986.
Cultivar Palmira Quilichao
Standard error of effects
Table 8. Crop dry weights (at 56 days) and yields of eight lines planted
at four distances between rows. All plots planted at 20
Distance between rows (cm)
60 45 30 20
S.E. for difference of means of:
S.E. for difference of means of:
Treatment x line =
Treatment x line =
Table 9. Response of yield to plantings at 0.3 m and 0.6 m between rows
with populations of 100.000 and 900.000 plants.ha-1. In
semester A,-data are for BAT 477, and in semester B, for means
of lines BAT 271, BAT 477 and BAT 881.
Significance level of effect:
Distance x population
Significance level of effect:
Distance x population
- ------- ---------
Table 10. A plan for a progressive increase in bean yields.
Increase (in .)
Plus effect of:
Distance between rows
Resistance to lodging
Response to No
& A yield increase of 1193 kg.ha-I represents 250 kg.ha-1 more protein,
which in turn represents 18 kg.ha-1 additional N in seed, not counting
losses in material not harvested.
Seed weight (g)
Example of compensation between yield components in 38 lines.
represent levels of constant yield. Source: White, 1981.
S' = 077**
S y 0.323 + 0.000128 X
R2 = 0.77**
1000 2000 3000
Figure 2. Illustration of the problem of artificial correlations
between yield and harvest index. Data were calculated
from 20 pairs of random numbers used to represent
vegetative and ,sed weight in 20 hypothetical cultivars.
Figure 3. Bulk-pedigree method used for yield breeding at CIAT.
Month Generation Activities
Parental and Fi
C = CIAT-Palmira, P = PopayAn, 0
Intermating between selected parents
Intermating between selected parents and crosses
Save F2 seed of selected individual crosses
Grow & save F2 seed of selected crosses
Yield test of replicated trials
Save bulk F3 (SP) of selected crosses
Yield test & save bulk F4 of selected crosses
Yield test & save bulk F5 of selected crosses
Space plant for selection of invidual plants
Progeny test in replicated yield trials
Save bulk F, of selected lines
Yield test & code selected lines for National
Programs & VEF, EP, IBYAN and other nurseries
_ ~I~ ___
= Santander de Quilichao
Figure 4. Recurrent selection method proposed for yield breeding in CIAT.
Month Generation Activities Locality
C~P Itermte slectd paentsfor reedng o
C = CIAT-Palmira, P
* If not needed, a
CoP Intermate selected parents for breeding of
CoFi Yield test and discard poor crosses
CoFI Intermate between selected crossis using
remnant seed (Fi)
CoFI Yield test and discard poor crosses
CoF2 Yield test.and discard poor crosses
CoF3 Space plant for. individual selection
CoF4 Yield test and discard poor families
CoFs Yield test and discard poor families
C1F, Intermate between families selected for
= PopayAn, 0 = Santander de Quilichao
new recurrent selection cycle could be initiated in F4
Distance between rows (cm)
Relation between cancpy cover and distance between
rows. Mean of four genotypes. Source: Molina, 1986.
0 *.* *
S 0 0 *
I B ~_0
Relative pod weight
Relation between relative leaf and pod weights (removed plus mat ire pods) in
individual plants of G 4523 (ICA Linea 17).
Y = 0.44 0.50 X
Nitrogen (kg.ha- .week-1)
Interaction of weekly applications of nitrogen (as urea,
starting 30 days after planting) and distance between rows
for mean of BAT 1481 and ICA Pijao.
1 I _j
No. of plants.ha-1
Loss of population at 0.3 m and 0.6 m between rows
for densities at thinning of 15 to 90 plants.m -2
0 -= Optimal management
E = Stress
Hypothetical responses of.traditional and bred
materials to optimal or stress management
The Role of Architecture, Crop Physiology and Recurrent Selection
in Ideotype Breeding for Yield in Dry Beans
M. W. Adams and J. D. Kelly.
Crop and Soil Sciences Department
Michigan State University
SEast Lansing, MI 48824 USA
According to Donald (1968), ideotype breeding conveys the thought
that the plant breeder will conceive of some set of architectural-
morphological characteristics in a given crop plant which, in a
particular agro-ecologic setting, would be expected to produce maximum
yields. The ideotype is an idealized plant model; it does not exist
initially except in the mind of the breeder.
In ideotype breeding, the breeder maybe likened to the architect-
builder. A design must first be conceived, a set of blueprints drawn,
and then the building constructed. Hopefully, the building will be both
attractive and functional, and appropriate for the site and community in
which it is located.
In present usage, breeding to a yield ideotype requires the breeder
to consider not only morphological-architectural characteristics but
crop physiological attributes as well, and since the breeder is also the
"builder," there is required an understanding of- the genetics of the
various characteristics, their interactions with each other and the
environment, and a strategy for bringing the various component parts
together into a single genotype. Our paper will follow this general
Critics may question this strategy, in preference to the more
traditional approach of "crossing the best with the best, and selecting
the best." While some progress in yield may result from this approach,
we feel that the procedure is more mechanical than creative since it
does not take advantage of what may be known of the morphological and
physiological determinants of yield. Thus the best yield achievable
will be less than that which is potential in the species as a whole were
morphological-physiological information to be exploited.
The fundamental architectural objectives in bean yield design were
previously identified (Adams, 1982) to be:
1. a leaf-stem canopy capable of intercepting all incident light,
by absorbing, scattering, and transmitting the
photosynthetically active radiation throughout the plant
2. to construct the canopy so that the profile consists of as many
source-sink units (leaf, internodal section, node, and auxiliary
raceme) as possible and feasible, and
3. to make each unit as functionally efficient as possible,
a. adjustment of the source to sink ratio to maximize pod and
seed number and single seed size relative to leaf size, and
b. the source leaves being displayed in such a way as to
intercept light sufficient for a high rate of
photosynthesis, without seriously occluding leaves located
lower in the plant or crop profile.
These objectives are general enough to cover plant ideotypes for
several agro-ecologic conditions. For the small-seeded navy (white)
- J -
bean to be grown as a monocrop under optimal soil and climatic
conditions in the mid-continental U.S., we have translated these
objectives into a specific set of architectural characteristics, which
we call an "architype," and have given the statistical data which
support the architype as a high yield ideotype (Adams, 1973, 1982). A
number of superior-yielding navy and black-seeded varieties (Table 1)
have been produced, based upon the architype model (Adams et al. 1986,
Kelly et al. 1984).
Architecture, in and of itself, however, does not guarantee high
yield. Morpho-architectural characteristics must be associated without
strong adverse relationships to yield, and moreover, must be combined
with yield-promoting crop physiological attributes, which were alluded
to in our earlier ideotypes, but now need to be specifically enunciated,
insofar as there is experimental evidence for their inclusion in a
It should be evident that seed yield is the integration of seed
growth rate per unit of land area over the time period of seed
filling. It might seem logical, therefore, to include both seed growth
rate (SGR) and seed filling duration (SFD), or effective filling period
(EFP) as ideotype traits, and to select for these traits in the breeding
program. But first we must ask to what extent in beans these traits are
correlated with yield or any of the components of yield, and how they
are correlted one with the other.
Paredes (1986), studying 12 cultivars of common beans varying in
plant type, seed type, maturity and yield, found significant differences
in both rate and duration of seed-filling. Mean filling rate, however,
was highly negatively correlated with number of pods/square meter,
number seeds/pod, seed yield, bioyield, and days to physiological
maturity. The linear filling period (major component of EFP), on the
contrary, was highly positively correlated with those same
characteristics. Clearly, SCR and SFD are inversely related to each
other in the data of Paredes (1986); in selecting for a high SGR, one
would select in a secondary way for a reduced duration of filling.
Izquierdo (1984) had previously shown this inverse relationship in the
two strains, NEP-2 and Black Turtle Soup. These observations were based
upon the performance of specially selected strains and released
varieties, and it cannot be assumed that the same relationships would
necessarily prevail in large unselected populations of lines.
Nevertheless, no exceptions to this relationship were found among F3 and
F4 families produced by single seed descent from the cross of NEP-2 by
Black Turtle Soup (Adams 1974, unpublished observation).
A recent paper by Salado-Navarro et al. (1986a) in soybeans
summarizes experiments which, in toto, present a somewhat inconsistent
picture as regards SGR and SFD in that crop. High yields may be
obtained in varieties with low SCRs, and varieties with different yields
could display similar SCRs. Varieties with similar yields exhibited
significant differences in SGR. Only small fractions of yield variation
among field-grown random soybean genotypes could be explained by linear
regression upon estimates of seed filling duration. Nevertheless, in
simulated genotypes, Salado-Navarro et al. (1986b) found that highest
yield was achieved by a genotype with the greatest biomass at the
beginning of seed growth, highest potential crop growth rate, and the
lowest dry matter allocation coefficient (DMAC, this being defined as
the rate of linear increase of the harvest index) which resulted in a
high SGR and the longest seed filling duration. DMAC, when high,
reflects a rapid withdrawal of C- and N-assimilates from vegetative
organs and transport to seeds, resulting in early senescence of leaves
and a strongly limited seed filling duration. When effective filling
period is determined as the final seed weight divided by the seed growth
rate, it is obvious that an inverse relationship between SCR and EFP
will be found. Sinclair (in Salado-Navarro, et al., personal
communication, 1986) believes that in unselected populations, where
there is genetic variability for both SGR and EFP, genotypes will occur
in which high levels of both parameters will be present.
In Izquierdo's (1981) data on nine bean cultivars, the length of
effective filling was positively and significantly correlated with
seeds/pod, number seeds/square meter, and yield. The linear seed
filling rate was significantly negatively correlated with seeds/pod, and
negatively but not significantly correlated with seeds/square meter, and
with yield. The top-yielding entries, however, were those with the
longest effective filling periods and with high, but not highest, linear
filling rates. These Type II architypes have since been released as
commercial cultivars in Michigan, where they continue to yield from 15
to 30% above standard determinate navy beans. These varieties and
subsequently released architypes not only yield well but display
superior yield stability over locations and years. (Table 2)
On the weight of simple logic and the accumulating experimental
evidence, it is clear that, in addition to morpho-architectural traits,
certain crop physiological characteristics should be considered in a
high yield ideotype. These characteristics include the following:
seed growth rate,
longer effective filling period,
biomass at the beginning of seed filling, and
a high partitioning coefficient or harvest index.
It must be recognized that these traits may not be genetically or
developmentally independent of other traits, including the architectural
ones. High partitioning without high initial biomass will not lead to
high yields unless photo-assimilate production during seed filling is
It should be apparent in all of the considerations concerning this
topic that there can be numerous alternative routes to high yield.
Highest yields, however, would only be possible when all components were
combined at their maximum levels. Since this may be impossible from a
practical standpoint for reasons of genetic and/or developmental inter-
dependency, we as breeders should probably strive for optimum levels of
combinations rather than maximal levels.
These crop physiological parameters, if they are to be used
successfully by bean breeders, must be shown to be heritable in the
additive genetic sense, and must be capable of being accurately
estimated in breeding populations. In beans, only limited data on the
extent of genetic control are available. In a diallel cross of six dry
bean genotypes differing in plant type, seed type, and the seed-filling
traits, Paredes (1986) showed general combining ability effects to be
significant and predominant for total filling duration, linear filling
duration, mean filling rate and seed weight. General combining ability
variance in linear filling rate failed to reach significance in his data
because of the large error associated with its estimation. The large-
seeded entry, Harris Great Northern, displayed the largest positive
general combining effect for rate of seed filling among the six parental
genotypes studied by Paredes.
On this limited evidence in beans, but consistent with evidence
from other grain crops, it may be tentatively concluded that additive
genetic control may be expected for these seed filling
characteristics. Biomass at flowering and harvest index are also known
to be moderately heritable.
The major problem related to their use in a selection program
derives from the ease--or lack of ease--of estimation of the metrical
values.of the traits. Suffice it to say, without arguing the point,
that reliable estimates will require data from replicated plots in
advanced generations of a cross. Non-replicated single plant estimates
will be essentially worthless for these characteristics.
In a breeding program designed to combine numerous traits with
differing genetic inheritance patterns from various genetic sources into
a single genotype, traditional breeding procedures for improvement of
self-pollinated crops are limiting. There exist certain obvious basic
defects in the system of conventional repetitive pedigree cycling of a
two parent cross commonly used in the breeding of self-pollinated
crops. The major limitation is that genetic variability and
recombination potential are low because of the initially restricted gene
pool and the selfing process which limits genetic recombination. If a
complete reassortment of parental genes is preferred, there is only a
limited opportunity for that to occur with self-fertilization. Since
the more common systems for improvement of self-pollinated crops, such
as mass, pedigree, and modified bulk (single seed descent) selection
procedures, do not permit adequate genetic recombination, Jensen (1970)
proposed the diallel selective mating (DSM) system. Because the DSM
system involves intermating F1 and F2 individuals without target trait
selection, Bos (1977) compared it to the effect of random intermating of
F2 plants which he showed was negative in the F3 generation. Pederson
(1974) showed that intermating within an F2 population from a cross of
two homozygous parents will only be of benefit'for two loci if the
alleles are linked initially in the repulsion phase. If loci are spread
over three or more chromosomes then the expected gain from intermating
is consistently small. Hanson (1959) suggested that recombination may
be promoted by an initial round of intermating commencing at the F2
generation and he concluded from a study of the breakup of linkage
blocks that a breeding program for a self-pollinated species should
include at least one and preferably three or four generations of
intermating if at all feasible. Other strategies to improve genetic
recombination in self-pollinated crops have been used in soybeans, where
Thorne and Fehr (1970) showed that genetic variance for yield in 3-way
populations was larger than in 2-way populations. The 3-way cross
produced a greater number of superior lines for yield and was a better
avenue for introduction of exotic germplasm into soybean cultivars.
Other systems, such as recurrent mass selection which should favor
genetic recombination, have been employed by Sullivan and Bliss (1983)
to simultaneously increase seed yield and seed protein in beans. The
seed protein was increased after two cycles of selection whereas mean
seed yields were not. Sullivan proposed the use of a selection index
for the seed yield trait but the study was restricted by the small size
of populations used and the limited number of cycles examined. Fehr and
Ortiz (1975) used recurrent selection for yield in soybean and showed S1
selection to be more efficient assuming SI testing for yield is based on
one cycle per year whereas S4 testing would require two years. Compton
(1968) proposed a.system of recurrent selection in self-pollinated crops
which would minimize crossing by advancing only descendants of a
different F1 rather than from different F2 plants within an Fl. The
limitation of many recurrent selection schemes for yield is the
requirement for elaborate data collection if selection indices are used
or when S1 selection is practiced for yield. The latter selection
procedure requires the use of numerous hill plots and hill plot planting
arrangements which are not easily mechanized. The time consuming nature
of planting and data collection reduces the size and numbers of
populations that can be handled effectively, and so.reduces the
efficiency of the system.
Evidence from the breeding program at MSU (Figure) indicates, for
small-seeded beans, that the architype offers the most favorable
combination of traits positively associated with yield, suggesting that
the obvious strategy for yield improvement in other seed types (Kelly et
al. 1987) would be the development of architypes in the different
commercial classes. Since the architectural component of the yield
complex is highly heritable and highly visible, its choice as the most
favorable trait for selection in a segregating population is obvious.
Thus a system of phenotypic recurrent selection (PRS) where recombinant
individuals can be easily identified in large populations, selected in
the S1 generation and intermated as S2 individuals, permitting a cycle
per year, offers the best opportunity for efficiency. The advantage of
- lU -
intermating S2 plants is that the alleles are not homozygous at all loci
and so permits more genetic recombination than intermating lines
homozygous at all loci. This has the disadvantage that the selected
single plant phenotype may not accurately reflect a superior genotype
for use in the intermating of S2 individuals for the establishment of
the next cycle. This is the case particularly for the crop
physiological traits, if the breeder were to attempt to select for them
independently and individually. In our architype selection program with
small-seeded genotypes, favorable combinations of the physiological
traits seemed to be associated with the desired architecture, with the
result that selection for architectural form achieved acceptable levels
of physiological characteristics simultaneously.
The success of the PRS system will depend on the ability to make
sufficient numbers of crosses with the reproductive potential of
producing large recombinant populations. It will be crucial that
desired genes, initially held in unfavorable linkages, be released
through recombination and re-assembled into genic associations that
promote the levels of phenotypic performance sought in the ideotype.
This process will be aided by intermating of S2 individuals but more
importantly it will depend upon the nature of the genetic homology
between the original germplasm sources. Singh and Gutierrez (1984) have
identified an apparent incompatibility between small and medium to large
seeded germplasm sources which is controlled by two complementary
dominant genes. The genes have been implicated in limiting free genetic
recombination between the two germplasm sources and would seriously
effect a breeding system designed to maximize genetic recombination.
Gepts and Bliss (1985) have shown that there exist multiple regions of
- 11 -
domestication of dry beans and he has catalogued the diverse regions
using the major bean storage protein, phaseolin, as a genetic marker.
Beans from the Mexican center display a Sanilac (S-type) banding pattern
while lines from the Andean center produce a Tendergreen (T-type)
banding pattern with disc gel electrophoresis. All reports in the
literature indicate that genetic lethality in bean hybrids occurs when
crosses are made between lines from the different centers displaying
either the S or T-type protein markers. The lack of complete
compatibility among parents from diverse centers of domestication could
adversely affect the outcome of a system where free genetic
recombination needs to be obtained. Nienhuis and Singh (1986) indicated
that many of the high yielding small-seeded Central American cultivars
have negative CCA (Ceneral Combining Ability) for yield in combinations
with large-seeded germplasm. In addition to genetic incompatability and
F1 hybrid weakness in many crosses between these classes, CIAT has
encountered difficulties in developing large seeded types with high-
yielding ability (Research Constraints at International Agricultural
Research Centers 1986). The origin and behavior of the two seed size
classes suggest the possibility that chromosomal structural differences
exist between the two types limiting the random association of useful
characteristics other than the narrow genetic exchange possible through
a rigid backcrossing program. Since a lack of homology between the
high-yielding small-seeded architypes and the large seeded classes
appears to result in unfavorable genetic recombinations, only an
organized cyclic improvement system offers the opportunity to permit
adequate random genetic interchanges.
In conclusion, a point is reached where it is desirable to merge
divergent gene pools in order to broaden the range of genetic
variability or introduce particular genetic effects into a self-
fertilizing crop species. There exists in such cases a strong
likelihood of encountering linkage barriers. Mather (1973) hypothesized
that long term assortment and recombination with selection within a
given gene pool will have given rise to specific adaptive and
functionally-integrated or "relationally-balanced" gene blocks.
Different gene pools, if they have been genetically isolated for a long
period of time, will form qualitatively different relationally-balanced
gene blocks. When such divergent gene pools are merged by crossing, the
breeder must expect initially to encounter a "linkage-freeze" upon free
and random recombination for effects regulated by genes in the
integrated linkage segments. With subsequent generations of
intercrossing and selection, the linked effects should reassemble into
new combinations, among which some should be favorable in a selection
sense. The experience at MSU (Kelly and Adams 1987) with the
architectural components associated with the type-II plant type and the
seed size factors associated with the type-III pinto gene pool is
consistent with this expectation. It is interesting that qualitative
factors like seed color and mottling genes recombined freely in the
first segregating generations, but the major genes affecting
architecture and seed size did not recombine until the third cycle of
A breeder rarely possesses a "read-out" of the genetic organization
of the targeted effects in divergent gene pools prior to undertaking the
crossing and selection process. It is particularly noteworthy that the
system of phenotypic recurrent selection as described in this case has,
over time, proved effective in breaking and re-assembling a small number
of functionally integrated linkage blocks. The medium-seeded pinto
architypes resulting from our program (Kelly and Adams 1987) can now be
used as a bridge for the improvement of architectural characteristics of
other medium seed-sized commercial classes, such as great northern,
pinks, red mexicans, canaries, and bayos. The highly heritable nature
of the architecture super-gene block should permit the easy recovery of
these characteristics without the associated problems of linkages with
small seed size.
- 14 -
Adams, M.W., 1973. Plant architecture and physiological efficiency.
In: Potentials of field beans and other food legumes in Latin
America. Centro Internacional de Agricultura Tropical, Cali,
Colombia. pp. 226-278.
Adams, M.W., 1982. Plant architecture and yield breeding in Phaseolus
vulgaris. Iowa State J. Res. 56:225-254.
Adams, M.W., A.W. Saettler, G.L. Hosfield, A. Ghaderi, J.D. Kelly, and
M.A. Uebersax, 1986. Registration of 'Swan Valley' and 'Neptune'
navy beans. Crop Sci. 26:1080-1081.
Bos, I., 1977. More arguments against intermating F2 plants of a self-
fertilizing crop. Euphytica 26:33-46.
Compton, W.A., 1968. Recurrent selection in self pollinated crops
without extensive crossing. Crop Sci. 8:773
Donald, C.M., 1968. The breeding of crop ideotypes. Euphytica 17:385-
Fehr, W.R. and L.B. Ortiz, 1975. Recurrent selection for yield in
soybeans. J. Agr. Univ. P;R. 9:222-232.
Gepts, P. and F.A. Bliss, 1985. F1 hybrid weakness in the common
bean. J. Heredity 76:447-450.
Hanson, W.D., 1959. The breakup of initial linkage blocks under
selected mating systems. Genetics 44:857-868.
Izquierdo, J.A. 1981. The effect of accumulation and remobilization of
carbon assimilate and nitrogen on abscission, seed development, and
yield of common bean (Phaseolus vulgaris L.) with differing
architectural forms. Ph.D. thesis. Michigan State Univ. 188 pg.
Jensen, N.F., 1970. A diallel selective mating system for cereal
breeding. Crop Sci. 10:629-635.
Kelly, J.D., M.W. Adams, A.W. Saettler, C.L. Hosfield and A. Chaderi,
1984. Registration of C-20 navy bean. Crop Sci. 24:822.
Kelly, J.D. and M.W. Adams, 1987. Phenotypic recurrent selection in
ideotype breeding of pinto beans. Euphytica 36(1): in press.
Kelly, J.D., M.W. Adams, A.W. Saettler, G.L. Hosfield, M.A. Uebersax and
A. Ghaderi, 1987. Registration of 'Domino' and 'Black Magic'
tropical black beans. Crop Sci. 27(2): in press.
Mather, K., 1973. (ed.) In Genetical Structure of Populations. Chapman
& Hall Publ. London (pp. 197).
Nienhuis, J. and S.P. Singh, 1986. Combining ability analysis and
relationships among yield, yield components, and architectural
traits in dry beans. Crop Sci. 26:21-27.
Paredes-Carcano, O.M. 1986. A study of seed filling and dry matter
partitioning characteristics and their combining ability effects and
relationship to yield among dry beans with differing growth habits,
architecture and maturities. M.S. thesis. Michigan State Univ.
Pederson, D.G., 1974. Arguments against intermating before selection in
a self-fertilizing species. Theor. Appl. Genet. 45:157-162.
Salado-Navarro, L.R., T.R. Sinclair and K. Hinson, 1986a. Yield and
reproductive growth of simulated and field-grown soybean. I. Seed
filling duration. Crop Sci. 26:966-970.
Salado-Navarro, L.R., T.R. Sinclair and K. Hinson, 1986b. Yield and
reproductive growth of simulated and field-grown soybean. II. Dry
matter allocation and seed growth rates. Crop Sci. 26:971-975.
- 16 -
Singh, S.P. and J.A. Gutierrez, 1984. Geographical distribution of the
DLl'and DL2 genes causing hybrid dwarfism in Phaseolus vulgaris L.,
their association with seed size, and their significance to
breeding. Euphytica 33:337-345.
Singh, S.P., 1982. A key for identification of different growth habits
of Phaseolus vulgaris L. Ann. Rept. Bean Impr. Coop. (New York).
Sullivan, J.G. and F.A. Bliss, 1983. Recurrent mass selection for
increased seed yield and seed protein percentage in the common bean
(Phaseolus vulgaris L.) using a selection index. J. Amer. Soc.
Hort. Sci. 108:42-46.
Thorne, J.C. and W.R. Feher, 1970. Exotic germplasm for yield
improvement in 2-way and 3-way soybean crosses. Crop Sci. 10:677-
- 17 -
Table 1. Seed color and size characteristics, growth habit and
center of domestication of twenty-eight dry bean cultivars
representing seven commercial seed classes.
Commercial Seed Hundred Growth' Center
seed coat seed habit of
Cultivar class color weight domestication
tGrowth Habit I = determinate, II = indeterminate upright short
vine, and III = indeterminate prostrate long vine, Singh (1982).
- io -
Table 2. Days to maturity, mean
coefficient (b) and standard error 4Sb),
and coefficient of determination (r~) of
seed yield, regression
deviation mean square (S2)
28 dry bean cultivars from
42 trials conducted in Michigan during 1980-1985.
Cultivar to yield b s sb r
*,** Significantly different
from b=1.0, at the
0.05 and 0.01
- 19 -
S STA.. LITF' PA :^ 'A M1 ETECR
COMPARISON OF 28 BEAN GENOTYPES
SEED SIZE & PLANT HABIT
1 YIELD KG/HA = EMS'100
SS-I = Small-seeded, Type-I;
AB-II = Aurora, Bunsi, Type II;
LS-I = Large-seeded, Type I; ARC-II = Architype, II;
Type III = Medium-seeded, Type III.
SLS-I A-1 AE- I
SS-i LS--! ARC-I! AE--I TYPE-Il1
Heri tabi cities and FPhnotypi .:: Correlations for Seed Yield and
Yield Compornents. of Bean Popul] at ons Derived from Crosses
bet.wr'(:e D-etermi late anid Indetermi nate Genotypes
I'. Mateo Solano, J.S. Beaver, and F. Saladin Garcia 1/
INTRODU CTI ON
In spite of the f ac.:t tliat. climatic: and edaphic conditions di f er
r!-:at1ly f or b eainls gro own in the Domiinic an Republ]ic, -farmers tend to
-:. the same groa.upf of c.u:i t vars 4or all plantings. 1 here ore, in
-der toC be succ:essf u], bean cultivars need to be capable of
.oducingi predictable yields over a wide range of environmental
--nlditions. Result ts from -field trials conducted on small farms and
:r periment stations in the Dominican Republic found determinate,
-rd-rmottled bean cultivars to have lower and less predictable
fields than indeterminate, small-seeded genotypes ( 2 ). In an
-ttempt to improve the yield level. and stability of beans in the
o-.miniican Republ ic, a plant breeding program was initi-atec to
ovelop erect, indeterminate bean genotypes with seed size and
--_ed characteristics acceptable to the Dominican.consumer. Crosses
.--tween large-seeded, determinate and small-seeded, indeterminate
_. e*an g.enotypes can result in genetic dwarfs (5,10) which, perhaps,
:\'e discouraged bean breeders from making wide crosses. Nienhaus
.7d1 Singh (8) note that the design of effective selection
--- ocedCures to simultaneously increase yield and seed size depends
Former graduate r research assistant and associate
*-,ofessor, respectively, D ep. of Agronomy and Soils, Univ. of
.errto Rico, Mayaguez Campus, Mayaguez, Puerto Rico (00708 and
-incipal investigator Bean/Cowpea CRFSF Project, Secretaria de
:.t.ado de Agricultura, Santo Domingo, Dominican Republic
on the knowledge of the inheritance of yield and its components
and also on an understanding of the strength and stability of the
relationships among these traits.' The objectives of the research
were : 1) determine the yield potential and describe the yield
components of indeterminate F. lines derived from crosses between
mcdi num to large-seeded determinate and small-seeded
i ndetermi nate bean genotypes, 2) determine the effect of
productivity of environment on the expression of yield
components, .) determine the strength and stability of the
relationship of yield components with yield, and 4) estimate the
narrow sense heritabilities of seed yield and its components.
MATERIALS AND METHODS
Six bean populations derived from crosses between medium to
large-seeded determinate and small-seeded indeterminate bean
genotypes were used in the study. The medium to large-seeded,
determinate genotypes used in the study were Pompaduor checa,.Jose
Beta, 8241-168A, and Borinquen. The small-seeded, indeterminate
genotypes used in the study were H-376, La Vega, and PAI 92. The
F, generation of the populations was planted in blocks at the
Fortuna Agricultural Research and Development Center (ARDC), Juana
Diaz, Puerto Rico on 20 October 1984. Rows were spaced 60 cm apart
and-the spacing among plants within a row was approximately 10 cm.
At harvest maturity, 50 plants were selected at random from each
block. Seed yield and its components were measured. Seed yield
was measured as seed weight per plant. Seed size was measured as
the weight of 100 seeds. Number of pods was measured on a per
plant basis. Number of seed per pod was estimated by dividing the
number of seed per plant by the number of pods per plant.
A total '50 F_ plant rows of each population were.planted on 22
Feburary 1905 at the Fortuna ARDC and on 22 March 1985 on a small
farm in the El Rio parish, Constanza, Dominican Republic. A
replications within blocks design was utilized (6). The experiment
consisted of 5 blocks with 3 replications within each block. Each
replication of a block consisted of 10 F. lines selected at random
from a population and the two genotypes used as the parents to
create the population. The experimental units consisted of ten
seeds planted in 1 m row lengths spaced 0.6 m apart. Two bordered
F, plants having the same growth habit as the plant from which
they were derived were selected from each row. Seed yield and its
components were measured utilizing the previously described
Seed yield and yield component means of the indeterminate F3
lines and their parents were determined for each population and
location. Means of the F3 lines and their parents were compared
using an approximate t test. Fhenotypic correlations between seed
yield and the yield components were.calculated for each population
and location using the means of the indeterminate F_ lines. Narrow
sense heritabilities were estimated for each population _and
location by regression of means of F, lines onto individual F,
plant measurements. Spearmnan rank correlations between locations
were calculated using F. means.
RESULTS AND DISCUSSION
Seed yield per plant of the indeterminate F, lines were
generally significantly less than the seed yield per plant of
their indeterminate parents (Table 1). The differences between
the means of the F, lines and the means of the indeterminate
parents were particularly large in the low yield environment,
Constanza. These results indicate that the evaluation of large
popu atiorns may be necessary in order- to identify F: lines with
yield potential similar to the indeterminate genotypes used as
parents in this study. Moreover, it may not be possible to
identify indeterminate F, lines with adequate yield potential from
the first cycle of selection. Kelly (7) found recurrent selection
to be effective in the development of erect, indeterminate pinto
beans. The moderate to low narrow sense heritability values for
" seed yield indicate that individual plant selection for seed yield
in early generations would not be effective (Table 2). The
magnitude of the narrow sense heritabil cities are in general
agreement with previous research (4,8). These results suggest that
it would be more effective for breeders to select simply or highly
heritable traits such as -seed type, growth habit, relative
maturity, and resistance- to specific diseases in the early
generations. Selection for seed yield would be more effective in
replicated trials in advanced generations.
Number of pods per plant of the F. lines was significantly
less than their indeterminate parents for four of the six
populations at Constanza, the low yield environment (Table 1).
Narrow sense heritability estimates for number of pods per plant
were similar in magnitude to the narrow sense heritabilities for
seed yield (Table 2). As was Ifound in previous research (3,4),
number of pods per plant was highly correlated with seed yield at
both locations (Table 3).
Hu ndred seed weight of the F_ lines was greater at Constanza
than at Juana Diaz (Table 4). This is possibly due to a lack of
rainfall during flowering and pod set which reduced pod set per
plant. The plants, however were able to compensate for more
favorable environmental conditions during the latter part of the
growing season by producing larger seeds (1). Hundred seed weight
of the indeterminate F, lines was significantly less than their
determinate parents. The results again point to the need to
evaluate large populations in order to identify indeterminate F3
lines with seed sizes that are similar to their determinate
parents. Kelly (7) has noted in his effort_ to develop erect,
indeterminate pinto beans that desirable plant types are often
associated with small seed size. Narrow sense heritabilities for
100 seed weight tended to be low to intermediate in magnitude at
both locations (Table 5). Since it is such a simple trait to
measure, selection for greater seed size could begin in the early
generations. F'henotypic correlations between seed size and seed
yield were generally positive or non-significant (Table 3). The
lack of negative phenotypic correlations between seed yield and
100 seed weight indicates that the simultaneous selection for
greater seed yield and seed size might be possible.
Number of seed per pod of the ::, lines tended to be intermediate
between their determinate and indeterminate parents (Table 4).
Little difference was observed between locations for number of
se-eds per pod. Narrow sense heri ability estimates for number of
seeds per pod were low to intermediate (Table 5). Only two of the
six populations had significant phenotypic correlations between
sec:d yield and number of seed per pod at Juana Diaz (Table 3). In
Cor is tanrza, however, five of the six populations had significant
rcorrelations between number of seed per pod and seed yield. A
grea ciL.e number of seed per pod might have provided an opportunity
for plants in Constanza to partially compensate for a reduced
number of pods per plant (1).
Spearman rank correlations between the Fortuna and Juana Diaz
locations for seed yield and its components varied among
populations (Table 6). These results indicate that a large number
of. different populations should be developed in order to identify
populations which produce lines that perform well in contrasting
environments. Nienhaus and Singh (8) identified bean populations
which had positive general combining ability for yield in more
than one environment. These results also illustrate the
desirability of multi-location testing of advanced lines in
environments which are representative of the bean growing regions
of _the Dominican Republic. Hundred seed weight was the only
characteristic with significant rank correlations- for all
populations. This indicates that selection for seed size at one
location would be effective for a wide range of environmental
condi ti ons.
1. Adams, M.W. 1967. Basis of yield component compensation with
special reference to the field bean (Phaseolus vulgaris L.).
Crop Sci. 7:505-510.
2. Beaver, J.S., C.V. Paniagua, D.P. Coyne, and G.F. Freytag. 1985.
Yield stability of dry bean genotypes in the Dominican Republic.
Crop Sci. 25:923-926.
3. Camacho, L.H., C. Cardona, and S.H. Orozco. 1966. Genotypic and
phenotypic correlation of component of yield in kidney beans.
Bean Improvement Cooperative 7:8-9.
4. Coyne, D.PF. 1968. Correlation, heritability, and selection of
yield components in field bean. (haseolus'vulgaris L.). Crop Sci.
5. Gepts, F. and F.A. Bliss. 1985. F hybrid weakness in the
common bean. Journal of Heredity 76:447-450.
6. Hallauer, A.R. and J.B. Miranda.1981. Quantitative genetics in
maize breeding. Iowa State Univ. Press.Ames, Iowa.
7. Kelly, J.D. and M.W. Adams. 1986. Utilization of phenotypic
recurrent selection in bean breeding. Bean Improvement
8. Nienhaus, J. and S.P. Singh. 1986. Combining ability analysis
and relationships -among yield, yield components, and
architectural traits in dry bean. Crop Sci. 26:21-27.
9. Quinones, F.A. 1965. Correlation of characters in dry beans.
Proc. Amer. Soc. Hort Sci. 86:368-372.
10.Singh, S.P. and J.A. Guiterrez. 1984. Geographical distribution
of the DL1 and DL, genes causing hybrid dwarfism in Fhaselous
vulgaris L., and their association with seed size, and their
significance to breeding. Euphytica 33:337-345.
lable I. IlLan seed yield and number of pods per plant of parents and
indcAtermi nate F- populations of si:: bean populations grown at
Constanza, Domih'iican FRepublic arid Juana Diaz, FPuerto Rico.
Number of pods
N~imber of po~ds
per- p3 ant
Geereration Constanz a Juana Diaz
Constanza Juana Diaz
Pomp. checa X F / 9 16 8.0 17.1
H-376 F, 18 23 2().9 27.7
F, 12t. 17 1 1.94 19.7*.
Jose Ieta X F 10 16 7.2 12.9
H-376 p'. 19 21 22.3 24.6
F, 12: 17*: 11.9* 18. 0*
La Vega X F" 9 17 7.5 16.9
Pomp. checa F:, 16 19 14.8 22.0
La Vega X
PA I 92
1 0. 2 *
1/ F'P = determinate parent and P, = indeterminate parent
3 Significantly different from tHe indeterminate parent at the 0.05
Table 2. Narrow sense heritabilities for seed yield and number
of pods per plant. Heritabil.ities were estimated using parent-
ofCspring regressions of indeterminate F. and F. plants
of si: bean populations grown at Constan a, Dominican Republic
and Juana Di a Puerto Rico.
Seed yield Number of pods per plant
F'opul] action Constanza Juana Diaz Constanza Juana Di a
F'onip. c-h -ca > 30 O C. 57 0. 44
IJose BcDta X 0.37 0.60 o 0.63 -0. 13
La Vega X 0.39 .0.34. 0.33 0.51
La Vega X ( 0.18 0.22 O0 0.47
8241-168A X 0.59 0 C. 19 0.28
PAI 92 X 0 0 0 C. 19
able 3. Fhenotypic correl-ations between number of pods per plant,
hundred seed weight, number of seed per pod and seed yield.
Correlations were estimated using F, plants of six bean pop-
ulations grown at Constanza, Dominican Republic and Juana Diaz
Number of pods Hundred seed Number of seed
per plant weight per pod
Population Const. J. Diaz Const. J. Diaz Const. J. Diaz
F. checa X 0.82 1/ 0.'77*1 0.07 0.23 0.28 0.46 4
Jose Beta X 0.69*. 0.81*3 o. 14 0. 11 0O17 0.23
La Vega X 0.82** 0.85 t 0.29 -0.05 0.47** 19
La Vega X 0.63.: 0.79* 4 0.50t*C -0.01 0.49t* 0.01
82'11-168A X O. 0** 0.85** 0.24 0.53t* 0.50*:. 0.41**
PFA1l 92 X 0.81 0.86*. 0.39* 0.11 0.42** 0.07
1/ *,**:-Significant at the 0.05 and 0.01 probability levels,
able 4. Mean hundred seed weight and number of seed per pod of parents
and indetermi nate F. populations of six bean populations grown at
Cohstanza, Dominican Republic and Juana Diaz, Puerto Rico.
Hundred seed Number of seed
weight per pod
F'opulation Generation Constan::a Juana Diaz Constanza Juana Diaz
Pomp. checa X P 1/ 5. 9 29. 2 3. 1 3. 2
H-376 PF 24.3 19.6 3.8 4.3
F. 26. It 22.2#t 4.0 3.9t
Jose Beta X F' 48.4 41.0 2.9 3.1
H-376 P7, 20.9 19.8 4.6 4.3
F3 29.7# 26.5# 3.6* 3.7
La Vega X FP 35.4 30.0 3.2 3.3
, Fomp. checa F', 23.2 20.2 4.7 4.2
F. 28.5# 24.5# 3.5* 3.4*
La Vega X P 49.9 45.5 2.6 3.1
Jose Beta 'P 22.0 20.1 4.1 4.4
F- 27.9# 25.5# 4.0 3.9*
82-41-168A X FP 37.1 32.4 3.4 3.8
PAI 92 PF2 23.7 21.7 4.1 4.3
F3 27.0# 24.6# 3.5* 3.44
PAI 92 X P .29.2 29.7 3.2 4.2
Borinquen P'- 24.1 22.3 4.1 4.6
F3 25.9# 23. 7# 3.8 4.1*
I/ PF = determinate parent and PF = indeterminate parent
# Significantly different from tNe determinate parent at the 0.05
* Significantly different from the indeterminate parent at the 0.05
Table 5. Narrow sense heritabilities for hundred seed weight and
and number of seed per pod. Heritabilities were estimated using
parent-offspring regressions of indeterminate F. and F,
plants of six bean populations grown at Constanza, Dominican
Republic and Juana Diaz,. Puerto Rico.
Hundred seed weight Number of seed per pod
--- -- --- -- --- -- -- --~--- -- ------- --'-- --------
Populati on Constanz Juana Diaz Constanza Juana Di az
FPonmp. ceca X 0.25 0.22 0. 11 0.40
Jlose Beta X 0.37 0.45 0.23 0.24
La Vega X 0. 33 23 0.15 0.19
La Vega X 0.51 0.44 0.25 0.30
8241-168A X 0.40 0.40 0. 19 0. 10
PFA I 92
PAI 92 X 0. 33 0.20 0 0.64
or i nquen
Table 6. Spearman ranl correlations between Constanza, Dominican
Republic and Juana Diaz Puerto Rico for F_ family means for
seed yield, number of pods per plant, hundred seed weight, and
number of seed per pod.
Seed Number of pods Hundred seed .Number of
Population yield per plant weight seed per plant
Pomp. checa X 03 0.17 0.54* U. 14
Jose IOic a X 0..32* 42** 0.53 :* 0.60* :
La Vega X 0.01 0.38 0. C. 74*t 0.20
La Vega X 0., 5 0.706. 0. 68** 0.12
8241-168A X 0.29* 0. 10 .37*t 0.33*
FPA 92 X 0.31* 0.07 0.37** 0.32**
------------------------------------------------- 7 ------ 7------------
1/ *,** Significant at the 0.05 and 0.01 probability levels,
EARLY GENERATION SELECTION FOR HIGH YIELD IN BEANS USING PHYSIOLOGICAL
COMPONENTS OF YIELD AS SELECTION CRITERIA
P.N. MASAYA *, J.W. WHITE *, D.H. WALLACE --:' AND R. RODRIGUEZ .
Superior yield is an important, if not the ultimate goal of any
plant breeder. The several breeding methods are meant to identify and
perpetuate those genotypes with the potential for superior yield. Yield
being the result of all genes in the plant, is difficult to see as a -
trait at the time of individual or mass selection.
When making individual selections the breeder must keep in mind
that yield is produced by a community of plants growing together on a
given area of land in a given period of time. Such a plant community
must have the capability for the efficient use of the sunlight radiation,
C02 water, and soil nutrients during a growth cycle that is fixed by a
agricultural or climatological characteristics. Donald (3) has emphasized
the differences in viewing the yield per plant and the yield per unit of
land. Wallace (12), and Wallace et al (14) pointed the features required
in a cultivar for accomplishing the efficient use of time and environmental
resources for seed yield. The breeder, then, should visualize the plant
being selected as growing in a homogeneous or heterogeneous plant popular
tion, as monoculture or as a mixture of crops, but in any case, being
part of a stand.
The environment modulates plant growth, by its factors: water avai-
lability, soil fertility, temperature, photoperiod and light quality. The
bean plant responds to the environment by developing varying numbers of
branches, nodes on branches, roots and rootlets, and later, of pods, de
pending on the intensity or quality of environment factors. For example,
when the temperature is warmer, in the range from 190C through 290C(mean
temperature), the internodes are longer, and leaves are more numerous (7)
(6). Long day, promote bigger leaves, more numerous nodes and a longer
period of growth, in comparison to short days, specially in photoperiod
sensitive cultivars (5) (6) (11) (13).
* Bean Breeders. Institute de Ciencia y Tecnologla Agrlcolas -ICTA-
** Bean physiologist. Bean Program. Centro Internacional de Agricult.u
ra Tropical -CIAT- Call, Colombia.
-* Profesor. Department of plant Breeding and Biometry. Cornell Univ.
The neighboring plants have a strong effect on the growth of the
bean plant. This is specially intense in an F2 population where each
plant has a different genome. Plants with a more aggressive growth
reduce the growth and the seed yield of adjacent, less aggressive plants.
Since plant phenotypes are affected by the environment there is the
need for a testing of progenies or lines in early or advanced generations.
In fact, all breeding methods include some means of testing the result of
selection. The breeder uses then her or his experience and knowledge of
the crop and a proper methodology for reducing the amount of lines to be
:ested, since the line testing, is the most expensive part of the bree--
It is considered that selection in early generations for yielding
ability is non productive because of low heritability of yield charac-
ters (14) (10) (8). Is is futile to attempt identifying by visual ins-
pection the plants which produce progenies with superior yield. The bree
der can discardthe plants with obvious defects or plants with poor ar-
chitecture. On the other hand, an empirical selection seems unavoidable
(4), followed by progenie or line testing, of yield.
In many bean breeding programs, a multiple objective scheme is ne-
cessary as when breeding for resistance to several diseases, as in CIAT
bean program. In such situations, it is highly desirable to reduce by
some early selection the number of lines to carry through advanced gene
rations and through on farm trials. Recurrent selection and crossing is
becoming more used in bean breeding. An early identification of superior
crosses and lines for new combinations is essential for efficient recurrent
It has been advocated before the need for an integrative selection
that should be able to select those plants that have the more complete
combination of genes for yield (13, 14). Since all genes in the plant
affect adversely or favorably the seed yield, it is difficult to identi
fy or predict which combination of genes will give the highest yield,
and which plant has the very combination of genes for superior yield.
Because of that reasoning it is considered more practical to measure the
outputs of the yield process more closely related to the seed yield (12).
They are: a) the total plant weight, which measures the ability of the
plant for accumulating dry matter; b) the efficiency in partitioning,
that is, the ability of the plant for diverting the accumulated dry ma-
tter toward the seeds. Such efficiency is measured by the harvest index,
indicated by Wallace for bean breeding (12). The harvest index is defi
ned as the ratio of seed weight divided by the total plant weight.
The ideal, high yielding genotype, coupled to the ideal crop management,
should combine the largest total weight and a large value of harvest in
dex, that is, the largest proportion of the total plant weight put into
The use of both, total plant weight and harvest index as selection
criteria requires facilities for storing, drying and weighing whole plants
(excluding leaves and petioles) besides the weighing of seeds. In beans,