You are on page 1of 14

Euphytica 102: 343–356, 1998.

343

c 1998 Kluwer Academic Publishers. Printed in the Netherlands.

Breeding for yield in dry bean (Phaseolus vulgaris L.)

James D. Kelly, Judith M. Kolkman & Kristin Schneider


Crop and Soil Sciences, Michigan State University, East Lansing, MI 48824, U.S.A.

Received 3 June 1997; accepted 13 February 1998

Key words: breeding pyramid, combining ability, gene pools, ideotype, partitioning, wild species

Summary

Strategies employed by dry bean breeders to improve yield include early generation testing, ideotype breeding,
selection for physiological efficiency, and selection based on genotypic performance and combining ability across
gene pools of Phaseolus vulgaris. Ideotype breeding has been successfully deployed to improve yield in navy, pinto
and great northern seed types. The ideotype method is based on an ideal plant architecture to which breeders target
their selection. Breeding for physiological efficiency is important in combining increased biomass, high growth rates
and efficient partitioning. Genotypic performance and combining ability are also critical for yield improvement,
since crosses between gene pools can exhibit negative combining ability and problems with lethality, whereas
interracial crosses within the same gene pool exhibit the greatest potential. Breeders must work within specific
constraints for growth habit, maturity, seed quality and disease resistance. A three-tiered pyramidal breeding
strategy is proposed to facilitate yield improvement in dry bean. Breeding of elite, agronomically acceptable
germplasm within the same market class is restricted to the apex of the pyramid. The intermediate level has fewer
constraints and greater access to diverse germplasm. Interracial crosses within the same gene pool are utilized to
exploit genetic differences within adapted material. Extracting genetic diversity from unadapted sources, including
wild germplasm and other Phaseolus species, is conducted at the base of the pyramid. The objective of this breeding
strategy is the movement of improved germplasm towards the apex, using different breeding procedures to optimize
improvement at each tier of the breeding pyramid.

Abbreviations: AYT – advanced yield trial; BLUP – best linear unbiased predictions; CP – coefficient of parentage;
EGT – early generation testing; GD – genetic distance; G  E – genotype by environment interaction; GCA –
general combining ability; HI – harvest index; HOPE – hierarchical open-ended population enrichment; MAS –
marker assisted selection; MPV – mean parent value; PI – plant introduction; PYT – preliminary yield trial; QTL
– quantitative trait loci; RIPE – recurrent introgression population enrichment; SSD – single seed descent; YSA –
yield system analysis

Introduction ing nor responsive cultivars have generally been use-


ful as selection criteria for improving yield of grain
Progress in breeding for high yield in dry bean, Phaseo- legumes. In North America, soybean (Glycine max) is
lus vulgaris L., has been modest (Singh, 1991). Unlike the major grain legume, bred almost exclusively for
self-pollinated cereals, where improved partitioning yield by both the private and public sectors. Breed-
has resulted in higher grain yields, similar improve- ing has resulted in a steady gain in yield of 0.6% per
ments have not been forthcoming in grain legumes. year, which can be attributed solely to genetic improve-
Cereals, in general, are more responsive to increased ment (Fehr, 1987). One of the constraints recognized
inputs, such as nitrogen fertilization, which accounts by soybean breeders is the narrow genetic base avail-
for improved cultivar performance through selection able within cultivated soybean. Given this constraint,
for responsive genotypes. Neither improved partition- breeders attempt to choose parents that are genetical-

MENNEN/typeset: Pips Nr.:163806; Ordernr.:233475-cw BIO2KAP


euph4504.tex; 8/06/1998; 13:24; v.7; p.1
344

ly diverse based on measures such as the coefficient of lar approaches using molecular markers, linkage maps
parentage (CP). Limitations to CP are recognized since and wild species, need to be evaluated in P. vulgaris as
the effect of selection is not included in this measure- an alternative to the conventional breeding strategies
ment and CP is not well correlated with genetic dis- currently being deployed.
tance (GD) determined by other methods (Beebe et al.,
1995). Genetic distance based on molecular markers
is becoming increasingly important as an independent Literature review
method to aid in the selection of genetically diverse par-
ents. Studies comparing GD, based on CP or molecular Breeding strategies to improve seed yield of dry bean
markers, with genetic variance for yield, have shown have followed approaches similar to those applied
the comparative advantage of the information generat- in other crops. These include early generation test-
ed by the markers (Kisha et al., 1997). The major strat- ing (EGT) and selection based on yield components.
egy employed by soybean breeders in North America is Adoption of either approach has had limited success,
to cross genetically-distinct, elite, high-yielding lines because the distinct market classes in dry bean differ
that exhibit agronomic similarity for seed size, growth greatly in seed size and pod number, which restricts
habit, and maturity. Expected gain from a cross can improvement due to yield component compensation.
be estimated as mean parent value (MPV) for yield In addition, the need to select for disease resistance fac-
or using the best linear unbiased predictions (BLUP), tors and quality traits compete in efficiency with EGT
which provide higher rank correlations and lower stan- for yield, particularly when wide crosses are made.
dard errors. Using the BLUP method, breeders have Yield selection based on an individual component has
identified higher percentages of superior crosses than not been a successful approach (Nienhuis & Singh,
with MPV (Panter & Allen, 1995). The use of plant 1988). This result is not unexpected since dry bean was
introductions (PI) as novel sources of variability for used as the model to explain the theory of yield com-
yield in soybean has had limited success. Ininda et ponent compensation in crop plants (Adams, 1967).
al. (1996) demonstrated that the use of PIs to increase In soybean, Cooper (1985) demonstrated the value of
genetic gain for yield was not effective based on three EGT to identify F2 -derived lines with the highest yield
cycles of recurrent selection. The increased variabili- potential. This method was successful in developing
ty anticipated in the PIs has not been demonstrated in semidwarf soybean cultivars suitable for high input
marker studies. Many of the PIs show no greater vari- production systems in the American Midwest. The
ability than that observed among cultivated materials greatest impact of semidwarf soybean cultivars is on
and their use as parents has not resulted in major gains more productive soils, where lodging continues to be a
in yield. If variability in this germplasm is limited, yield barrier to taller indeterminate cultivars (Cooper,
as suggested by the marker studies, then the poten- 1985). F2 -derived EGT schemes have not been wide-
tial value of PIs in yield improvement will be similar- ly accepted by soybean breeders except for specific
ly restricted. Progress may be achieved by using the management situations (semidwarf soybean) or for the
wild species (G. soja) in crosses with the cultivated introgression of determinacy into indeterminate culti-
species. Since G. soja has a weedy, low-yielding phe- vars (Cooper, 1985).
notype, single crosses to G. max exhibit unacceptable In dry bean, visual selection for yield in the F2
variability. Three-way crosses and backcrosses offer and F3 generations has been shown to be ineffective
greater opportunity to exploit useful genetic variabil- (Patino & Singh, 1989), whereas selection directly
ity from G. soja in the agronomic background of G. for yield was the best selection criterion among the
max (B. Diers, personal communication). To better yield components tested in F3 to F7 generation mate-
utilize the wild species, Tanksley & McCouch (1997) rials (Singh & Gutierrez, 1990). EGT was effective
suggest that breeders need to shift the paradigm from for the selection of yield across locations, but no com-
phenotypic selection to selection based on advanced parison of efficiency with other breeding systems was
backcross-quantitative trait loci (QTL) analysis of wild demonstrated (Singh et al., 1990). Selection of other
species in search of transgressive segregation of nov- key traits measured in this study would be conducted
el/unique recombinants that might contribute to yield in in later generations, but the value of EGT is doubtful in
the cultivated background. Support for their strategy is breeding programs where resources for yield evalua-
based on studies on fruit size in tomato (Lycopersicon tion are limiting. Most breeders recognize yield testing
esculentum), and yield in rice (Oryza sativa). Simi- as the most costly and labor intensive operation they

euph4504.tex; 8/06/1998; 13:24; v.7; p.2


345

perform, so financial responsibility delays yield test- cal regions, selecting for specific morphological traits
ing until numbers can be reduced based on other traits. like plant height, node and leaf number, and branch-
However, Singh & Urrea (1995) suggested that EGT ing characteristics has not proven to be a successful
could be combined with other breeding systems (pedi- strategy to increase yield of beans (Nienhuis & Singh,
gree, single seed descent (SSD), gamete selection) to 1985), although such traits have been shown to be posi-
evaluate F2 and F3 populations across contrasting loca- tively correlated with yield in growth types I, II and III.
tions. Gamete selection for the simultaneous selection Recognizing the important role that growth habit plays
of multiple traits in common bean has been proposed in yield, adaptation, maturity, and harvest mechaniza-
(Singh, 1994). Results, to date, suggest that gamete tion, it is interesting to note that, in North America,
selection permits the rapid identification of lines with most breeders continue to work within the constraints
a combination of disease and plant architecture traits of the growth habit of current commercial seed types.
(Singh et al., 1996). A larger number of elite lines con- Based on the success of ideotype breeding in small-
taining the desired traits, however, were generated by seeded Mesoamerican race beans, Kelly & Adams
SSD than through gamete selection (Singh, 1997). (1987) used interracial recurrent selection to develop
In addition to traditional empirical approaches to Durango race architypes. Currently, the type II growth
select for yield in dry bean, other breeding methods habit from tropical black beans has been successfully
have been proposed and implemented. These are clas- introduced into commercial pinto and great northern
sified under three separate approaches to breeding for: seed types. Pinto cultivars Sierra (Kelly et al., 1990),
an ideotype (Adams, 1973), a specific combination of and great northern cultivars Alpine (Kelly et al., 1992a)
physiological traits (Wallace et al., 1993), or combin- and Starlight (Coyne et al., 1991) are among the first
ing ability based on specific genotypic combinations Durango race cultivars to possess a type II growth habit
(Singh, 1992). originating from race Mesoamerica. With the adoption
of these improved Durango and Mesoamerican race
cultivars by growers, current commercial bean yields
Ideotype breeding in Michigan for the 1990–1995 period are 25% high-
er than any previous five-year period. Similar cyclic
The ideotype concept was first proposed in cereals breeding programs have been used to develop high-
by Donald (1968), and was later applied to common yielding Andean beans for the tropics. Beaver & Kelly
bean by Adams (1973, 1982). Breeding for a small- (1994) used S4 recurrent selection to develop indeter-
seeded ideotype resulted in the development of high- minate, large-seeded Caribbean red beans that yielded
yielding navy bean architypes with longer maturity and 30% more than the determinate check cultivars, but
a growth habit different from previous commercial cul- were 10 days longer in maturity. Inter-gene pool cross-
tivars. Kelly et al. (1987) showed that type II cultivars es between Nueva Granada and Mesoamerican race
exhibited the highest yield and yield stability across beans have been used to successfully develop upright
both rainfed and irrigated environments in Michigan. type II large seeded beans with high yield.
Type III indeterminate cultivars exhibited high yield The use of ideotype breeding in other crops has met
but were less stable because of disease problems with with only limited success. In barley (Hordeum vul-
white mold (causal organism; Sclerotinia sclerotiorum gare L.), for example, Rasmusson (1987) concluded
(Lib.) de Bary), and harvest risks associated with pros- that the requirement for symmetry among plant parts,
trate growth habit. In general, the highest yielding yield component compensation, pleiotropy, and differ-
determinate cultivars yielded less and were less sta- ences in genetic background, are major factors which
ble across a broad range of environments than indeter- limit progress in ideotype breeding. He recommend-
minate types. Low yielding, small-seeded determinate ed that ideotype breeding be used to augment, but not
cultivars, however, tended to be more stable. It was not replace, traditional yield breeding. Similar limitations
until the development of small-seeded cultivars with can occur with dry bean, given major differences in
high yield potential that this instability was observed. growth habit and seed size, and may contribute to the
However, data from the Cooperative Dry Bean Nurs- lack of acceptance by breeders working with different
ery indicated the overall superior performance of type seed types. Brothers & Kelly (1993) showed that differ-
III genotypes across North America (Stewart-Williams ent ideotypes need to be developed for market classes
& Myers, 1995) despite potential production problems which differ widely in seed size. The pinto bean ideo-
with this growth habit in the humid Midwest. In tropi- type differed from the small-seeded navy bean ideo-

euph4504.tex; 8/06/1998; 13:24; v.7; p.3


346

type in having fewer pods per plant and fewer seeds indirect effects from exclusive selection of individual
per pod. In the semi-arid regions of the U.S., where components imply that superior effective performance
a more rigid upright stem offers no agronomic advan- based on selection for yield and its three physiolog-
tage, Myers (personal communication) believes that, in ical components can be achieved. Many breeders do
comparison with the type III plants, the type II growth not collect data on biomass. Without this data or an
habit may actually result in lower-yielding plants. The indirect measure of biomass, yields will stagnate as
type II plant may partition more energy into building breeders eliminate larger, less efficient genotypes from
a rigid, lodging-resistant, thick stem that competes for their nurseries. Yield in Sierra pinto is a direct result
the photosynthate available for yield. Breeders need to of an increase in biomass with an extension of the seed
develop specific ideotypes for each major market class filling period which lowered HI. Further progress in
and production zone. breeding for yield in North America could be achieved
by combining the high HI, early season type III cul-
tivars from the west with the high biomass and full
Breeding for physiological efficiency season type II cultivars from the Midwest. New pin-
to bean cultivars Aztec (Kelly et al., 1992b), Chase
Wallace et al. (1993) suggested that indirectly selecting (Coyne et al., 1994), Maverick, and Hatton (Grafton et
for the three major physiological components of yield, al., 1997) have resulted from this strategy. Similar suc-
namely biomass, harvest index (HI), and days to matu- cess has been achieved in the Mexican highlands by
rity, should result in improved yield. Simultaneous combining tropically adapted type II Mesoamerican
selection is required because genetically established race genotypes with the type III photoperiod-sensitive
interrelationships among the three components exist. Durango race beans adapted to the highland conditions
The correlation between HI and maturity is negative, (Flor de Mayo M38; Acosta-Gallegos et al., 1995a).
as is the correlation between HI and biomass, where- However, Myers (personal communication) argues
as maturity and biomass are positively correlated. All that it is doubtful if progress towards improving yield
three physiological components and the correlations can be achieved simply by crossing high biomass types
among them can be quantified by a yield system analy- with the high HI types, as suggested by the authors.
sis (YSA) of yield trial data. Breeders need to combine genotypes with high overall
Long term direct solution for yield will gradually growth rates with those exhibiting complete partition-
exploit all useful genetic variability for HI and maturi- ing to make major advances in yield. He bases his
ty. Ultimately, it will lead towards an excessively high argument on the YSA approach to breeding for yield
HI because a smaller vegetative structure must sup- in common bean proposed by Wallace et al. (1995) and
port larger yield without causing lodging and this will summarized as follows:
reduce canopy photosynthesis or result in an increase Adaptation is the major constraint to breeding for
in specific diseases. Lower planting densities of type III yield in common bean. A cultivar must fit the environ-
medium-seed size cultivars in the western U.S. where ment in which it will be grown so days to maturity is
disease pressure is minimum, offset this disadvantage. the most important physiological component affecting
Indirect selection for higher biomass can only occur that outcome.
if genes for superior biomass are introduced. Wallace All bean plants have a specific growth rate.
et al. (1993) argued that a YSA approach is needed Through genetic manipulation, this growth rate can be
to insure that biomass is measured since long term increased. Increasing growth rate does not necessarily
benefits will not be accrued if this trait is ignored. translate into higher seed yield unless greater amounts
Selection for biomass within the constraints of matu- of biomass are partitioned to the seed.
rity is equivalent to actual selection for a higher rate Different genotypes have different degrees of parti-
of biomass accumulation. Selection for yield per day tioning. Indeterminate navy (race Mesoamerican) and
should help compensate for the tendency of higher pinto (race Durango) beans have similar growth rates
biomass to result in longer duration of growth. Selec- (Stewart-Williams & Myers, 1995), and exhibit similar
tion solely for greater biomass will result in longer biomass accumulation. When genotypes switch from
maturity, lower HI and higher yield, whereas selection vegetative to reproductive growth, navy beans are less
solely for HI results in early maturity with an accom- complete in the transition than are pinto beans. As a
panied reduction in yield. Selection for early maturity result, navy beans continue to add vegetative growth
will reduce both yield and biomass. These negative (biomass) and the rate of seed fill is relatively slow,

euph4504.tex; 8/06/1998; 13:24; v.7; p.4


347

resulting in longer days to maturity (incomplete par- 1986; Singh et al., 1989a, 1992a). Clearly, knowl-
titioning types). Since pinto beans make a rapid and edge of parental performance and GCA among and
complete transition to reproductive growth, seed fill is within races, and selection for yield per se are essen-
rapid, resulting in faster progression to maturity (com- tial to breed for higher yield (Singh, 1992). Using this
plete partitioning types). Because the transition is more information, high-yielding lines from interracial pop-
complete in pinto beans, a higher HI is achieved. The ulations of dry bean have been developed within gene
rapid and complete transition from vegetative to repro- pools (Voysest et al., 1994; Singh, 1995a; Singh et al.,
ductive growth is essential to achieve high yields in 1993; Singh et al., 1989a) and from interracial pop-
duration-limiting environments. Without an efficient ulations between gene pools (Beaver & Kelly, 1994;
method to collect biomass data, vital information crit- Paredes & Gepts, 1995; Singh et al., 1992b). With few
ical for the YSA, is lost. Breeders need to be con- exceptions, breeders in North America have under-
vinced that the gain in information is worth the effort utilized other gene pools or races to affect change
to collect the biomass data. Finally, breeders need to in yield potential of dry bean despite the information
select genotypes that exhibit an efficient partitioning available on the yield potential of specific races and the
switch from vegetative to reproductive growth for spe- favorable combining ability for yield present in specific
cific local environments. interracial crosses.
Yield relative to a production system is critical.
Breeding dry bean for yield can only be successful-
Breeding for yield based on genotypic ly accomplished within the framework of very specific
performance and combining ability constraints. Work conducted at the International Center
for Tropical Agriculture (CIAT) suggested that breed-
A survey of North American dry bean cultivars based ing for increased yield in dry bean must consider the
on CP revealed a narrow genetic base of all major mar- interrelated effects of growth habit, seed size, maturity
ket classes (McClean et al., 1993). The lack of vari- and gene pool (Kornegay et al., 1992). The authors
ability may be one explanation for the limited progress showed that improved yield in segregating populations
to breed for yield in dry bean. Progeny generated from was associated with increased expression of climb-
crosses within the small-seeded Middle American gene ing bean growth habit, which involves traits like vine
pool and between gene pools did not outyield the best length, climbing ability, node number on the main
check cultivars. Only in crosses between small- and stem, and plant height. Genetic progress within one
medium-seeded genotypes from the Middle Ameri- growth habit type does not ensure progress in another
can gene pool did lines out-yield the best parent and type, nor does it apply universally across seed types.
check cultivars (Singh et al., 1989a). In general, limited High yields reported in the production of small-seeded,
gain in yield was achieved among crosses from with- long-season, highland, type IV climbing beans has lit-
in races and gene pools, due to insufficient levels of tle or no application to the improvement of determinate
genetic variability (Singh et al., 1989b). Only progeny short season beans needed for lowland production in
from Middle American interracial crosses out-yielded the Caribbean. Support for individual physiological
the best parent, whereas none of the progeny of the processes that combine differentially to generate yield
intraracial crosses out-yielded the best parent, regard- in genotypes from the different gene pools is based
less of planting density (Singh & Gutierrez, 1990). on data from interracial crosses between gene pools
The mean of interracial F1 hybrids was found to be (Welsh et al., 1995). Yield in dry bean is achieved
higher than that of intraracial hybrids (Singh & Urrea, in a fashion similar to that observed in landrace cul-
1995). In another study, substantial yield gains could tivars from the two domestication centers. Yield in
be expected from interracial crosses within the Mid- early maturing genotypes with few pods and larger
dle American gene pool, whereas selection for higher seeds is best combined in determinate, bush Andean
yield in Andean  Middle American crosses would be beans in the tropics, whereas yield in late maturing
problematic (Singh, 1995b). Progress has been limited genotypes is best expressed in the indeterminate Mid-
also by the occurrence of F1 hybrid weakness between dle American small- and medium-sized genotypes that
members of the two gene pools (Gepts & Bliss, 1985). have more pods and greater biomass. Differences in
Large differences in general combining ability (GCA) yield potential between gene pools may also be a func-
were found among dry bean cultivars belonging to tion of adaptation to specific environments (J. Myers,
different races and gene pools (Nienhuis & Singh, personal communication). Large-seeded Andean geno-

euph4504.tex; 8/06/1998; 13:24; v.7; p.5


348

types appear to possess a co-adapted complex of adap- size and early maturity, and the inherent inability of
tation genes for a more benign environment, such the determinate plant to recover from stress, are con-
as mild temperatures, less intense sun, and plentiful tributing factors to lower yields observed in type I cul-
moisture. In contrast, Durango race genotypes pre- tivars. Determinate beans were found to be inherently
fer more intense sunlight, and tolerate more extremes less stable and lower yielding in trials conducted in the
in temperature and moisture. Myers suggests using an American Midwest (Kelly et al., 1987). In other studies
inbred-backcross system to transfer genes for seed size conducted across nine locations in Puerto Rico and the
alone into Durango race background as an approach to Dominican Republic, yield and stability were higher
develop high-yielding, large-seeded types with toler- among indeterminate lines, as compared to determinate
ance to more extreme environments. Clearly, many lines (Beaver et al., 1985, 1996). In an interracial cross
of the physiological processes that control yield differ (Mesoamerican/Jalisco), a growth habit modification
between diverse genotypes, so genetic gains in yield towards semi-climbing was suggested to be the major
are not always directly transferable. Rather than apply reason for increased yields in this type of cross (Korne-
yield improvement to specific genotypes or seed types, gay et al, 1992), whereas White et al. (1992) believe
the strategy for yield improvement needs to be evaluat- that the optimum ideotype for large-seeded indeter-
ed in view of successes obtained in other grain legumes. minate beans has not yet been elucidated. The data
Again, improvement in cereals, which differ greatly in suggest that the indeterminate growth habit permits a
physiological basis for yield from legumes, are unlike- greater expression of yield potential than the deter-
ly to contribute to this strategy. minate growth habit. Major modifications of growth
Breeding for high yield in dry bean must be con- habit, if undertaken to increase yield, may not be read-
ducted within the major constraints of growth habit, ily accepted by growers.
maturity (adaptation), and seed size (quality) prefer-
ences as well as disease resistance factors. Breeding Maturity. This must be regarded as a major constraint
for yield outside these major constraints will result in to yield since it restricts yield potential of germplasm
germplasm with either little or no value in terms of selected for climatic extremes. In temperate and high-
commercial utility. Breeding programs must be struc- land zones, the growing season is shortened by early
tured to allow for the utilization and improvement of killing frosts, whereas cropping systems and rainfall
germplasm with potential to improve yield while not patterns similarly restrict the growing season in tropi-
being limited by the major constraints discussed below. cal zones. Selection for early maturity resulted in lower
yields – yield potential was decreased by 72 kg ha 1
for each day less in maturity (White & Singh, 1991).
Constraints In a cross of Redkloud/MAM 4, an increase in num-
ber of days to maturity resulted in higher yields, but
Growth habit. If high yield was chosen as the sole smaller seed size (Welsh et al., 1995). However, the
basis for selection in dry bean and growth habit was negative association with seed size and maturity sug-
ignored, type IV growth habit would indirectly be gests that large-seeded Andean beans achieve maxi-
selected. In one study, determinate genotypes were mum yields through different physiological phenom-
lower yielding than indeterminate genotypes, where- enon than small- or medium-seeded Middle American
as another group of isogenic lines differing in growth beans. This implies that breeders will not be success-
habit showed no negative effect on yield due to growth ful in exploiting yield potential from inter gene pool
habit (White et al., 1992). The authors suggested that crosses differing in genetic background and maturity.
a mere change in stem type from determinate to inde- As shown in the YSA, maturity must always be fac-
terminate did not increase yield of large-seeded near- tored into a breeding program for yield (Wallace et al.,
isogenic lines. Other traits, such as node and branch 1993, 1995).
number, need to be increased to affect a change in
yield. In an inter-gene pool cross between genotypes Quality constraints in beans. These are a major lim-
from the Nueva Granada and Durango races, lines with itation, given the range of seed types differing in size,
an indeterminate growth habit had singificantly high- color, shape, and cooking quality. These quality traits
er seed yield than lines with determinate growth habit restrict germplasm available to the breeder, or restrict
(Welsh et al., 1995). In reality, the pleiotropic effects the liberal use of wide germplasm, and dramatically
of the fin gene expressed in genotypes with larger seed limit the number of progeny advanced into yield trials.

euph4504.tex; 8/06/1998; 13:24; v.7; p.6


349

Certain color combinations are inherently difficult to new cultivars. Vigorous high-yielding type III Duran-
breed, limiting progress within these seed types. Par- go race pinto beans in the humid American Midwest
ents from the Middle American gene pool possess neg- are very susceptible to white mold, which dramatical-
ative combining ability for seed size, which restricts ly reduces yield. High-yielding genotypes may help
their value as parents for improving the yield of large- promote specific diseases like white mold because of
seeded Andean beans (Singh et al., 1992a). Similar their prolific foliage (biomass). Morphological avoid-
reports by Welsh et al. (1995) confirmed an inverse ance mechanisms need to be incorporated which may
relationship between seed size and yield in crosses prevent the optimization of yield potential. Disease
between gene pools. In fact, White & Gonzales (1990) resistance does not directly increase yield, but in the
demonstrated a 280 kg ha 1 yield reduction for every absence of adequate levels of resistance to diseases
100 mg increase in seed weight. Breeders working on like bean common mosaic virus (BCMV), common
the large Nueva Granada type beans may have to find bacterial blight, bean golden mosaic virus (BGMV),
alternative genetic sources for increasing yield rather and anthracnose, bean yields will fall below optimum.
than through inter-gene pool crosses. Since consumers Adequate levels of disease resistance to a number of
expect a specific combination of color, size, and shape, pathogens are needed to help stabilize dry bean yield.
high-yielding dark-colored red beans have had limited However, breeding for resistance to BGMV has been
acceptance by growers in Central America, partially shown to reduce genetic variability in Central Ameri-
negating the impact of high-yielding germplasm (J. can beans (Beebe et al., 1995). Despite evidence to the
Beaver, personal communication). contrary, some breeders believe that certain resistance
Similar experiences have been observed in North genes carry a physiological load which results in lower
America, where high-yielding kidney bean culti- yield. Near-isogenic lines carrying the different alleles
vars (Isabella, Charlevoix) were not accepted by the of the Are/are anthracnose resistance gene showed no
processing industry. Therefore, farmers must grow difference in yield and/or maturity (Park & Tu, 1987).
cultivars with the desired quality trait but with low- Breeders need to recognize the disease constraints
er yield potential. A new kidney bean genotype with within their production zones and restrict resistance
a 20% yield increase (selected based on YSA) was breeding to these specific pathogens. In the semi-arid
abandoned since it produced an unacceptable canned western production states of North America, BCMV
product (Wallace et al., 1993). The expenditure of and root rots are the major pathogens. Incorporating
resources by North American breeders to select for resistance to a wide array of additional pathogens will
processing quality is large relative to other traits. This limit progress towards improving yield and offer little
limits progress in breeding for improved yield. A recent advantage to growers within these regions. Similarly in
consumer survey, conducted by scientists in Mexico the humid Midwestern U.S., where a broader array of
(Castellanos et al., 1997), determined that the con- pathogens limit productivity, certain seed-borne bac-
sumer used cooking time as the basis for choice when terial pathogens can be effectively controlled through
selecting bean types for consumption. This informa- preventative measures such as seed certification pro-
tion has forced bean breeders in Mexico to evaluate all grams. An over-emphasis on resistance breeding may
breeding material entering yield trials for water uptake hinder breeding for yield by restricting variability and
and cooking time. High yielding cultivars such as Pinto limiting resources, but it may be essential in areas
Villa (Acosta-Gallegos et al., 1995b) and Bayo Victo- where disease management programs are inadequate.
ria, have a dramatically longer cooking time than the
higher priced Flor de Mayo Bajio cultivar. Additional
selection for shorter cooking time will certainly have a Materials and methods
negative effect on breeding for high yield since a por-
tion of the germplasm derived from the high-yielding Improving yield while maintaining or increasing genet-
Pinto Villa parent will have to be discarded on the basis ic diversity has been a common concern in many
of cooking time. Similar quality constraints confront crop breeding programs. A Hierarchical Open-ended
bean breeders in East Africa (Elia et al., 1997). Population Enrichment (HOPE) breeding system was
designed by Kannenberg (1981) as a method to broaden
Disease resistance. This is a critically important the genetic base of breeding populations in corn (Zea
objective in many bean breeding programs. The lack of mays L.). Four levels of hierarchy were established, to
resistance can dramatically affect the productivity of reflect elite, high, intermediate and base levels of pop-

euph4504.tex; 8/06/1998; 13:24; v.7; p.7


350

Figure 1. Breeding pyramid for yield improvement in dry bean (Phaseolus vulgaris).

ulation performance. A five year analysis of this hierar- ment time would be greatest at the lowest level. The
chical system resulted in its establishment as a viable exploitation and maintenance of this diversity would
breeding program for genetic gain in corn (Cramer be optimized at the lower level while uniformity and
& Kannenberg, 1992). Another breeding program, the yield performance would be sought at the apex of the
Recurrent Introgressive Population Enrichment (RIPE) pyramid. Since breeding approaches will differ at each
program, was designed to incorporate genetic diversi- level, they will be detailed separately, recognizing that
ty into elite lines of a barley breeding program (Kan- a strong continuity among all three levels is necessary.
nenberg & Falk, 1995). This program entailed five According to Welsh et al. (1995), gene exchange
sequential steps that led to the incorporation of new between parents of the same bean race should pro-
germplasm into elite lines, and was facilitated by the ceed without metabolic, hormonal, or physiologi-
use of male sterility. Both HOPE and RIPE appear to be cal hindrance because of evolutionary similarities. In
potential frameworks for incorporating new diversity contrast, greater degrees of biochemical adaptation
into a breeding program. and modification would be necessary to accommo-
In order to emphasize yield breeding in dry bean, date gene exchange between distantly related parents.
given the constraints mentioned earlier, we propose This implies that different breeding strategies must be
a structured program based on a three-tiered breed- employed to exploit these differences and the breeding
ing pyramid (Figure 1) similar to the HOPE and RIPE pyramid allows for such an approach.
breeding systems (Kannenberg & Falk, 1995). The
approach to yield breeding would be different at each Breeding pyramid for dry bean
level of the breeding pyramid, encompassing the vari-
ability within dry bean germplasm. Materials devel- The apex of the breeding pyramid. This level would
oped at the lower levels would be designed to move involve elite by elite crosses. Every attempt should
upwards sequentially or be maintained for additional be made to cross parents with maximum diversity but
improvement. The genetic diversity and the develop- within the limits of market class. Crosses would be

euph4504.tex; 8/06/1998; 13:24; v.7; p.8


351

restricted to within gene pools, within races, and with- line material. Crosses in this level would be restricted
in market classes. Selection for constraints other than within gene pools and between races. Crosses between
yield is eliminated since genotypes at this level are rela- market classes of similar seed size would be encour-
tively uniform for these traits. Breeding strategies in the aged. There also would be flexibility to cross between
elite group would utilize a limited number of crosses growth habits (I/II or II/III), among different architec-
within different market classes rapidly advanced using tural types and between adapted material differing in
SSD to the F5 generation after which preliminary yield maturity. A larger number of crosses would be made
trials (PYT) would be conducted. Selection intensity since no prior prediction of combining ability is known
could vary from 20 to 35% depending on factors such and larger F2 populations would be developed. Single
as environment, resources, and optimization of subse- plant selection would be rigorously practiced for agro-
quent advanced yield trial (AYT) nurseries. The select- nomic and seed traits (i.e. those demonstrating high
ed genotypes from the PYT would be advanced to the heritability) which can be readily fixed in this gener-
AYT, where selection pressure would increase to 10– ation. Selected individuals could be advanced rapidly
20%. Top performers from AYT would be considered to the F5 generation by SSD and entered into the PYT
for release as cultivars after extensive additional test- where selection would be based on performance com-
ing, consistent performance and consumer preference. pared to elite parental or other lines. Lines showing
Early generation testing could be used in the apex of above-average performance for yield would enter the
the breeding pyramid provided resources are available, elite group at the apex to be tested and utilized accord-
since many of the constraints (traits) listed above are ingly. Given that more flexibility in the choice of par-
fixed. Selection based on YSA could also be utilized at ents is possible at this level, breeders should consider
this level since seed size and growth habit differences developing an ideotype for the different seed classes
between parents is limited, allowing emphasis to be grown in their production zone. Flexibility in breed-
placed on biomass, maturity and partitioning data as ing for an ideotype or plant architecture is possible at
the basis of selection for yield. this level. In addition, based on data collected from
In soybean, where seed size is fixed and variabil- wider crosses or information in the literature, breeding
ity in maturity and growth habit traits is restricted, should emphasize combining ability and breeders need
breeders have made annual progress in yield through to know the combining ability of all parental material.
EGT (F4 ) of large numbers of lines derived primar- There may, however, be more opportunity to exploit
ily from elite by elite crosses (Ininda et al., 1996). GCA at the intermediate level, since a broader array of
Multi-location testing, mechanized direct harvest, and crosses is possible at this level. Use of three-way cross-
electronic field measurements have contributed signif- es, pedigree, modified inbred-backcross, and limited
icantly to this success. Similar successes have been cyclic selection should all be considered as breeding
observed in dry bean where private breeders, out of methods in the intermediate level.
necessity, concentrate their efforts on the apex of the
breeding pyramid and utilize germplasm derived from The base of the breeding pyramid. It would represent
public programs as a source of variability without a foundation of diversity upon which the intermedi-
which progress would be limited. Typically, newly ate and eventually elite populations will rely for future
released cultivars represent the highest available yield progress. The failure to increase yield in dry bean can
potential at the time of release. A review of the pedi- be attributed to a lack of desirable alleles in the base
gree information of contemporary cultivars (McClean population, low heritability, high genotype by environ-
et al., 1993) demonstrates that they are the product of ment (G  E) interactions, yield component compen-
elite by elite crosses. The need to continue elite by sation, low or negative GCA within and between gene
elite crosses in dry bean within the apex of the breed- pools, and dependence on visual selection for yield in
ing pyramid will ensure the short term improvement early generations (Singh, 1991). Improvement in any
of yield in the different commercial classes and gene one of these areas should reflect higher yield poten-
pools. tial in dry bean. An analysis of individual elements
suggests that all the elements are interdependent, so
The intermediate level of the breeding pyramid. This any change in one will affect the others. The lack
would be structured to introgress greater levels of diver- of desirable alleles for yield in the base population
sity into breeding materials, and to identify top per- is a fundamental limitation but the real problem lies
forming intermediate lines to test and cross with elite in identifying these desirable alleles. Once identified,

euph4504.tex; 8/06/1998; 13:24; v.7; p.9


352

they can easily be transferred into new germplasm. out solely targeting single genes. Singh (1996) cau-
Plant breeders in many crops recognize the value of tions that the use of random intermating to simultane-
variability, but identifying sources of useful variability ously improve yield and seed size in interracial pop-
for complex traits is formidable. In soybean, recur- ulations will result in mostly small-seeded progeny.
rent selection was used to increase the variability and The need to maintain some large-seeded recombinant
contribution of plant introductions in five G. max pop- plants for additional intermating in recurrent selection
ulations (Ininda et al., 1996). No increase in genetic programs was demonstrated in interracial crosses used
gain for seed yield was observed in the four populations to improve Durango race pinto bean (Kelly & Adams,
ranging from 25 to 100% PI parentage when compared 1987).
to the population derived solely from elite adapted cul- A conversion program, recently established
tivars and experimental lines. Clearly, within soybean, between the USDA-ARS Tropical Agriculture
elite germplasm must carry desirable alleles for seed Research Station (TARS) in Puerto Rico and breeders
yield, and recurrent selection was ineffective in identi- in North America, is a second example of manipulation
fying desirable alleles for yield in the cultivated PIs. A in the base population. The conversion program, mod-
similar situation probably exists in dry bean. Factors eled on a similar successful program with sorghum
such as dwarf lethal genes, negative GCA, and yield (Sorghum bicolor; Patterson, 1998), was established
component compensation all limit the exploitation of with TARS to better utilize and exploit, through col-
inter-gene pool variability in the cultivated dry bean. laboration, the tropical Andean germplasm which,
Increased diversity, through both intraspecific and because of photoperiod sensitivity and type IV growth
interspecific hybridization, would be sought at the base habit, makes both crossing and evaluation difficult in
level. Intraspecific hybridization would utilize a wide temperate regions. The simple introgression of deter-
range of germplasm from within and between gene minacy (fin gene) and photoperiod insensitivity (ppd,
pools. Opportunities at this level would attempt to hr genes) will allow for the testing and evaluation of
exploit yield potential, such as that demonstrated by the introgressed tropical germplasm in temperature zones.
Durango and Jalisco races of dry bean (Kornegay et al., Crosses and F1 progeny are produced in the tropics.
1992). Effective introgression of these races into other Determinate growth habit and photoperiod sensitivity
gene pools or races would be critical. Another method in the F2 are selected for in temperate regions after
to increase diversity in the base population, is through which progeny are returned to the tropics for addition-
interspecific hybridization. To meet this requirement, al crossing with tropical materials. Again the cyclic
the breeder must have specific objectives and adequate nature of this breeding program is essential to achieve
technical expertise. Desirable yield traits found in oth- these objectives.
er Phaseolus species can be introgressed into a base Other structured breeding programs could be envi-
P. vulgaris type. Once incorporated into cultivated P. sioned, but systems that involve cyclic intermating
vulgaris, efforts would be focused on bringing these such as recurrent selection (Beaver & Kelly, 1994; Kel-
traits into the intermediate level, and finally the elite ly & Adams, 1987); inbred-backcross (Bliss, 1993);
level for evaluation. congruity backcross (Urrea & Singh, 1995); or con-
The value of the base population is dependent on ical crossing (Fouilloux & Bannerot, 1988; Bett &
its utilization in the higher levels of the breeding pyra- Michaels, 1994) are the most likely to be effective.
mid. In order to achieve this objective, the selection Some form of recurrent selection will be needed to
and advance of progeny from these wide crosses needs create new combinations of desirable traits from the
to be rigidly structured. Long-term selection and cross- different gene pools, because the different genetic
ing programs need to be established to facilitate this and developmental associations between germplasm
objective. One example of such a program was a recur- from different gene pools have evolved over millennia
rent selection program designed to improve the yield (Welsh et al., 1995). However, breeders need patience
of Caribbean red beans. Yield of determinate Andean to give these programs time to achieve some level of
red beans was improved through a system of F5 recur- recombination and assortment before identifying lines
rent selection used to identify high yielding progeny for advancement to the intermediate level of the breed-
that were recombined in additional cycles of recurrent ing pyramid. The inbred-backcross line method is well
selection (Beaver & Kelly, 1994). The system demands suited for the transfer of genes controlling a quanti-
patience and resources but it offers a realistic avenue to tative trait from an unadapted or exotic source into
effectively combine diverse germplasm sources with- a commercially acceptable type (Bliss, 1993). Con-

euph4504.tex; 8/06/1998; 13:24; v.7; p.10


353

gruity backcrossing over recurrent backcrossing is sug- uation of novel traits in a cultivated background. The
gested to maximize recombination between and retain exploitation of the wild relative of tomato (Lycopersi-
desirable traits from distantly related parents used in con pimpinellifolium) using advanced backcross-QTL
interracial crosses (Urrea & Singh, 1995). The conical analysis has the potential to enhance elite tomato cul-
crossing system (Fouilloux & Bannerot, 1988) has the tivars for a number of quantitative traits (Tanksley et
potential to combine genes for quantitatively inherit- al., 1996). In rice, Xiao et al. (1997) used a high
ed traits from a number of different sources. In this density molecular map and BC2 interspecific (Oryza
system, breeders may choose the degree of diversi- rufipogon) test cross populations, to identify two QTL
ty by varying the number of elite or non-elite par- each associated with a 17% increase in yield over the
ents used in the design. Efficiency of conical crossing elite cultivated hybrid (Tanksley & McCouch, 1997).
could be improved by combining this approach with Similar approaches need to be considered in Phaseo-
gamete selection (Singh et al., 1996) for those traits lus breeding. Use of the wild species is essential since
where linked dominant markers are available. Com- the genetic variability from one cultivated gene pool
bining ability may actually be lower in the very wide has limited use for improving quantitative traits in the
crosses that would be conducted at the lower level of other gene pool.
the breeding pyramid. Initially, wild germplasm from within the same
Finally, the use of the wild P. vulgaris species as gene pool of P. vulgaris should be used in breeding
an alternative to continued crossing among races of the cultivated crop prior to using the wild germplasm
the cultivated species, should be evaluated in the base from the other gene pool. With the discovery of a
population of the pyramid. Independent domestication third ancestoral gene pool in Ecuador (Kami et al.,
events in Phaseolus suggest that the germplasm with 1996), the opportunity to find additional novel vari-
the greatest potential for future improvement should ability is available. The selection of suitable wild par-
be the wild species, particularly within the gene pool ents for crossing would be facilitated using marker
from which the cultivated crop originated. The genetic technologies to identify unique germplasm within the
control of the ‘domestication syndrome’ (marked phe- wild gene pools (Tohme et al., 1996) and related wild
notypic differences between the cultivated and wild species (P. costaricensis; Freytag & Debouck, 1996).
bean) in cultivated bean involves few genes with large The proposed breeding pyramid would offer the struc-
effects (25–30%) which account for a substantial (40– ture needed to exploit wild germplasm as a source
50%) part of the phenotypic variation (Koinange et of desirable alleles for yield in the base population.
al., 1996). In addition, these same genes are localized Backcross populations (BC2 ) and adequate linkage
in three regions of the genome. Although variability maps of wild P. vulgaris constructed within gene pools
was greatly reduced by domestication, the localiza- need to be developed to identify QTL associated with
tion of the genes involved in domestication implies improved seed yield. The identification of QTL for
that additional useful (novel) variability, not associ- yield would bypass the other limitations such as low
ated with these regions, should exist within the wild heritability, and high G  E interactions that reduce
species. The simple genetic control of these phenotypic progress in the improvement of complex traits such as
differences suggests that the introgression of diversi- yield. Although yield testing in rice involved hybrid
ty from the wild to the cultivated P. vulgaris should test crosses for hybrid development whereas pure lines
present no special difficulties (Koinange et al., 1996). are the cultivars of choice in dry bean, the importance
However, the unique alleles are masked by the pre- of EGT would be minimized since marker assisted
ponderance of deleterious alleles in the wild species selection (MAS) would replace direct selection. This
which restricts the use of the wild species for the assumes that the useful QTL are not genotype spe-
genetic improvement of quantitative traits. Alternative cific and would not have to be evaluated for every
mating schemes which increase the recovery of proge- genetic cross in which they were utilized. One might
nies with a cultivated phenotype should be investigated perceive an initial program of backcrossing wild P. vul-
(Singh et al., 1995). With only a few linkage groups garis species and advancing the recombinants into the
involved in the domestication syndrome, pre-breeding base population based on MAS for specific QTL for
efforts in the wild species may be feasible and aimed at yield. It is important to assess this type of germplasm
incorporating those genes associated with determinacy, and introgress desirable traits (many unknown) into
photoperiod-insensitivity and seed size. Manipulating the cultivated background. By shifting the paradigm
these traits in the wild species should permit the eval- from phenotypic to genotypic selection, improvement

euph4504.tex; 8/06/1998; 13:24; v.7; p.11


354

of quantitative traits will be accelerated. Tanksley & of improving yield without neglecting the other non-
McCouch (1997) outline a breeding strategy that can yield related traits.
identify these QTL while simultaneously transferring
the desired alleles into adapted genotypes. Once iden-
tified, these novel characteristics could be effectively Acknowledgements
evaluated and introgressed into more elite germplasm
at a higher level in the breeding pyramid. The authors thank J. Myers and K. Grafton for their
As plant breeders, we need to recognize the level at critical review of the manuscript.
which we are currently breeding and to which level we
need to look for further improvement. These decisions
can vary dramatically between breeding programs but References
the model should be constant. If a breeder works exclu-
sively with elite lines, the lack of diversity in one’s Acosta-Gallegos, J.A., J.Z. Castellanos, S. Nunez-Gonzales, R.
Ochoa-Marquez, R. Rosales-Serna & S.P. Singh, 1995a. Reg-
germplasm base may be limiting further progress. The istration of ‘Flor de Mayo M38’ common bean. Crop Sci 35:
development of an intermediate level program or col- 941–942.
laboration with other breeders may be critical. If the Acosta-Gallegos, J.A., R. Ochoa-Marquez, F. Ibarra-Perez, A.
breeder is working on germplasm enhancement using Pajarito-Ravelero & I. Sanchez-Valdez, 1995b. Registration of
‘Pinto Villa’ common bean. Crop Sci 35: 1211.
wide crossing, the breeding program needs to be ade- Adams, M.W., 1967. Basis of yield component compensation in crop
quately structured to identify high-yielding lines that plants with special reference for field beans, Phaseolus vulgaris.
would be useful at a higher level in the pyramid or Crop Sci 7: 505–510.
which could be valuable to other breeders. Many breed- Adams, M.W., 1973. Plant architecture and physiological efficiency.
In: Potential of Field Beans and other Food Legumes in Latin
ers may recognize the level of the breeding pyramid at America, pp. 226–278. CIAT, Cali, Colombia.
which he/she expends the most effort with the realiza- Adams, M.W., 1982. Plant architecture and yield breeding in Phase-
tion that they are not achieving their desired objective. olus vulgaris L. Iowa State J Res 56: 225–254.
Beaver, J.S. & J.D. Kelly, 1994. Comparison of two selection meth-
Our challenge to those breeders is to become involved
ods for the improvement of dry bean populations derived from
in all three levels of the pyramid since many programs crosses between gene pools. Crop Sci 34: 34–37.
may be stagnated at one or two levels. Collaboration Beaver, J.S., C. Paniagua, D. Coyne & G. Freytag, 1985. Yield
and germplasm exchange between breeding programs stability of dry bean genotypes in the Dominican Republic. Crop
Sci 25: 923–926.
will continue to be the most effective strategy to max-
Beaver, J.S., E. Arnaud-Santana & D.P. Coyne, 1996. Yield stability
imize the diversity entering one’s program. of determinate and indeterminate red mottled beans. J Agric Univ
P R 80: 187–189.
Beebe, S.E., I. Ochoa, P. Skroch, J. Nienhuis & J. Tivang, 1995.
Genetic diversity among common bean breeding lines developed
Conclusions for Central America. Crop Sci 35: 1178–1183.
Bett, K.E. & T.E. Michaels, 1994. Accumulation of blight resistance
Given the wide range of variability available in P. vul- using a conical cross. Ann Rep Bean Improv Crop 37: 75–76.
garis, the challenge to breeders interested in improving Bliss, F.A., 1993. Breeding common bean for improved nitrogen
fixation. Plant and Soil 152: 71–79.
yield is to optimize ways to accumulate genes for yield Brothers, M.E. & J.D. Kelly, 1993. Interrelationship of plant archi-
in specific genotypes. The breeding pyramid is an orga- tecture and yield components in the pinto bean ideotype. Crop
nized system which encourages the search for yield Sci 33: 1234–1238.
genes in diverse materials, while also offering breed- Castellanos, J.Z., H. Guzman-Maldonado, A. Jimenez, C. Mejia,
J. Munoz-Ramos, J.A. Acosta-Gallegos, G. Hoyos, E. Lopez-
ers ways to exploit those genes already present in elite Salinas, D. Gonzales-Eguiarte, R. Salinas-Perez, J. Gonzales-
materials. Parental selection and combining ability are Acuna, J.A. Munoz-Villalobos, P. Fernandez-Hernandez & B.
key factors in this process. Since many wide crosses Caceres, 1997. Preferential habits of consumers of common bean
(Phaseolus vulgaris L.) in Mexico (English summary). Archivos
result in inferior progeny as a result of poor combining
Latinoamericanos de Nutricion 47: 163–167.
ability, care should be given to the choice and type Cooper, R.L., 1985. Breeding semidwarf soybeans. Plant Breed Rev
of diversity used and the methods used to effective- 3: 289–311.
ly exploit valuable variability needed for the future Coyne, D.P., J.R. Steadman, D.T. Lindgren & D.S. Nuland, 1991.
‘Starlight’ – Great northern dry bean. HortScience 26: 441–442.
improvement of yield in dry bean. Recognizing the
Coyne, D.P., D.S. Nuland, D.T. Lindgren & J.R. Steadman, 1994.
many constraints which limit progress in bean breed- ‘Chase’ pinto dry bean. HortScience 29: 44–45.
ing programs, breeders must accomplish the objective

euph4504.tex; 8/06/1998; 13:24; v.7; p.12


355

Cramer, M.M. & L.W. Kannenberg, 1992. Five years of HOPE: McClean, P.E., J. Myers & J.J. Hammond, 1993. Coefficient of
the hierarchical open-ended corn breeding system. Crop Sci 32: parentage and cluster analysis of North American dry bean culti-
1163–1171. vars. Crop Sci 33: 190–197.
Donald, C.M., 1968. The breeding of crop ideotypes. Euphytica 17: Nienhuis, J. & S.P. Singh, 1985. Effects of location and plant density
385–403. on yield and architectural traits of common bean. Crop Sci 25:
Elia, F.M., G.L. Hosfield, J.D. Kelly & M.A. Uebersax, 1997. Genet- 579–584.
ic analysis and interrelationships between traits for cooking time, Nienhuis, J. & S.P. Singh, 1986. Combining ability analysis and
water absorption, and protein and tannin content of Andean dry relationships among yield, yield components and architectural
beans. J Amer Soc Hort Sci 122: 512–518. traits in dry bean. Crop Sci 26: 21–27.
Fehr, W.., 1987. In: J.R. Wilcox (Ed.), Soybeans: Improvement, Nienhuis, J. & S.P. Singh, 1988. Genetics of seed yield and its
Production and Uses, pp. 250–255. ASA 16. components in common bean (Phaseolus vulgaris L.) of Middle
Fouilloux, G. & H. Bannerot, 1988. Selection methods in the com- American origin. II. Genetics, variance, heritability and expected
mon bean (Phaseolus vulgaris). In: P. Gepts (Ed.), Genetic response from selection. Plant Breed 101: 155–163.
Resources of Phaseolus Beans, pp. 503–542. Kluwer Academic Panter, D.M. & F.L. Allen, 1995. Using best linear unbiased pre-
Publication, Dordrecht, The Netherlands. dictions to enhance breeding for yield in soybean: I. Choosing
Freytag, G.F. & D.G. Debouck, 1996. Phaseolus costaricensis. A parents. Crop Sci 35: 397–405.
new wild bean species (Phaseolinae leguminosae) from Costa Parades, O.M. & P. Gepts, 1995. Extensive introgression of Middle
Rica and Panama, Central America. Novon 6: 157–163. American germplasm into Chilean common bean cultivars. Gen
Gepts, P. & F.A. Bliss, 1985. F1 hybrid weakness in the common Res and Crop Evol 42: 29–41.
bean. J Hered 76: 447–450. Park, S.J. & J.C. Tu, 1987. Comparison of near-isogenic lines for
Grafton, K.F., J.R. Venette & K.C. Chang, 1997a. Registration of anthracnose Are are genes in common bean, Phaseolus vulgaris
‘Hatton’ pinto bean. Crop Sci 37: 288. L. Euphytica 36: 251–256.
Grafton, K.F., J.R. Venette & K.C. Chang, 1997b. Registration of Patino, H. & S.P. Singh, 1989. Visual selection for seed yield in the
‘Maverick’ pinto bean. Crop Sci 37: 1672. F2 and F3 generations of nine common bean crosses. Ann Rep
Ininda, J., W.R. Fehr, S.R. Cianzio & S.R. Schnebly, 1996. Genetic Bean Improv Coop 32: 79–80.
gain in soybean populations with different percentages of plant Patterson, A.H., 1998. Molecular dissection of complex traits. pp.
introduction parentage. Crop Sci 36: 1470–1472. 187–195. CRC Press NY.
Kami, J., V. Becerra Velasquez, D.G. Debouck & P. Gepts, 1995. Rasmusson, D.C., 1987. An evaluation of ideotype breeding. Crop
Identification of presumed ancestral DNA sequences of phaseolin Sci 27: 1140–1146.
in Phaseolus vulgaris. Proc Natl Acad Sci USA 92: 1101–1104. Singh, S.P., 1991. Breeding for seed yield. In: A. Van Schoonhoven
Kannenberg, L.W., 1981. Activation and deployment of genetic & O. Voysest (Eds.), Common beans. Research for Crop Improve-
resources in a maize breeding program. In: G.G.E. Scudder & ment. CAB Int CIAT, Colombia.
J.L. Reveal (Eds.), Evolution Today, pp. 393–399. Proc 2nd Intl Singh, S.P., 1992. Common bean improvement in the tropics. Plant
Congr of Systematic and Evolutionary Biology, Hunt Institute Breed Rev 10: 199–269.
for Botanical Documentation, Carnegie-Mellon University, Pitts- Singh, S.P., 1994. Gamete selection for simultaneous improvement
burgh, PA. of multiple traits in common bean. Crop Sci 34: 352–355.
Kannenberg, L.W. & D.E. Falk, 1995. Models for activation of plant Singh, S.P., 1995a. Selection for water-stress tolerance in interracial
genetic resources for crop breeding programs. Can J Plant Sci 75: populations of common bean. Crop Sci 35: 118–124.
45–53. Singh, S.P., 1995b. Selection for seed yield in Middle American
Kelly, J.D. & M.W. Adams, 1987. Phenotypic recurrent selection in versus Andean  Middle American interracial common bean
ideotype breeding of pinto beans. Euphytica 36: 69–80. populations. Plant Breed 114: 269–271.
Kelly, J.D., M.W. Adams & G.V. Varner, 1987. Yield stability of Singh, S.P., 1996. Effect of random intermating in early generations
determinate and indeterminate dry bean cultivars. Theor Appl on seed yield in interracial populations of common bean. Can J
Genet 74: 516–521. Plant Sci 76: 683–688.
Kelly, J.D., M.W. Adams, A.W. Saettler, G.L. Hosfield, G.V. Varner, Singh, S.P., 1997. Single seed descent versus gamete selection for
M.A. Uebersax & J. Taylor, 1990. Registration of ‘Sierra’ pinto breeding beans for resistance to insects. Agron Abstracts p. 71.
bean. Crop Sci 30: 745–746. Singh, S.P. & J.A. Gutierrez, 1990. Effect of plant density on selec-
Kelly, J.D., G.L. Hosfield, G.V. Varner, M.A. Uebersax, P.N. Miklas tion for seed yield in two population types of Phaseolus vulgaris
& J. Taylor, 1992a. Registration of ‘Alpine’ great northern bean. L. Euphytica 51: 173–178.
Crop Sci 32: 1509–1510. Singh, S.P. & C.A. Urrea, 1995. Inter- and intraracial hybridization
Kelly, J.D., G.L. Hosfield, G.V. Varner, M.A. Uebersax, N. Wassimi and selection for seed yield in early generations of common bean.
& J. Taylor, 1992b. Registration of ‘Aztec’ pinto bean. Crop Sci Euphytica 81: 131–137.
32: 1509. Singh, S.P., C. Cijiao, J.A, Gutierrez, J. Garcia, M.A. Pastor-Corrales
Kisha, T.J., C.H. Sneller & B.W. Diers, 1997. Relationship between & F.J. Morales, 1989a. Selection for seed yield in inter-gene pool
genetic distance among parents and genetic variance in popula- crosses of common bean. Crop Sci 29: 1126–1131.
tions of soybean. Crop Sci 37: 1317–1325. Singh, S.P., C.A. Urrea, J.A. Gutierrez & J. Garcia, 1989b. Selection
Koinange, E.M.K., S.P. Singh & P. Gepts, 1996. Genetic control for yield at two fertilizer levels in small-seeded common bean.
of the domestication syndrome in common bean. Crop Sci 36: Can J Plant Sci 69: 1011–1017.
1037–1045. Singh, S.P., R. Lepiz, J.A. Gutierrez, C. Urrea, A. Molina & H. Teran,
Kornegay, J., J.W. White & O. Ortiz de la Cruz, 1992. Growth habit 1990. Yield testing of early generation populations of common
and gene pool effects on inheritance of yield in common bean. bean. Crop Sci 30: 874–878.
Euphytica 62: 171–180. Singh, S.P., H. Teran, A. Molina & J.A. Gutierrez, 1992a. Combining
ability for seed yield and its components in common bean of
Andean origin. Crop Sci 32: 81–84.

euph4504.tex; 8/06/1998; 13:24; v.7; p.13


356

Singh, S.P., C.A. Urrea, A. Molina & J.A. Gutierrez, 1992b. Perfor- Voysest, O., M.C. Valencia & M.C. Amezquita, 1994. Genetic diver-
mance of small-seeded common bean from the second selection sity among Latin American, Andean and Mesoamerican common
cycle and multiple-cross intra- and interracial populations. Can J bean cultivars. Crop Sci 34: 1100–1110.
Plant Sci 72: 735–741. Wallace, D.H., J.P. Baudoin, J.S. Beaver, D.P. Coyne, D.E. Halseth,
Singh, S.P., A. Molina, C.A. Urrea & J.A. Gutierrez, 1993. Use of P.N. Masaya, H.M. Munger, J.R. Myers, M. Silbernagel, K.S.
interracial hybridization in breeding the race Durango common Yourstone & R.W. Zobel, 1993. Improving efficiency of breeding
bean. Can J Plant Sci 73: 785–793. for higher crop yield. Theor Appl Genet 86: 27–40.
Singh, S.P., A. Molina & P. Gepts, 1995. Potential of wild common Wallace, D.H., P.N. Masaya, R. Rodriquez & R.W. Zobel, 1995.
bean for seed yield improvement of cultivars in the tropics. Can 
Genotype, temperature, and genotype temperature interaction
J Plant Sci 75: 807–813. effects on yield of bean (Phaseolus vulgaris L.). In: M. Pessarak-
Singh, S.P., C. Cardona, F.J. Morales, M.A. Pastor-Corrales & O. li (Ed.), Handbook of Plant & Crop Physiology, pp. 893–915.
Voysest, 1996. Gamete selection for upright Carioca beans with Marcel Dekker, NY.
resistance to leaf hopper and four diseases. Ann Rep Bean Improv Welsh, W., W. Bushuk, W. Roca & S.P. Singh, 1995. Characterization
Coop 39: 86–87. of agronomic traits and markers of recombinant inbred lines from
Stewart-Williams, K. & J. Myers, 1995. 46th Annual Report of the intra- and interracial populations of Phaseolus vulgaris L. Theor
National Cooperative Dry Bean Nurseries. Univ of Idaho, SP 17. Appl Genet 91: 169–177.
Tanksley, S.D. & S.R. McCouch, 1997. Seed banks and molecular White, J. & A. Gonzales, 1990. Characterization of the negative
maps: Unlocking genetic potential from the wild. Science 277: association between seed yield and seed size among genotypes
1063–1066. of common bean. Field Crops Res 23: 159–175.
Tanksley, S.D., S. Grandillo, T.M. Fulton, D. Zamir, Y. Eshed, White, J.W. & S.P. Singh, 1991. Sources and inheritance of earliness
V. Petiard, J. Lopez & T. Beck-Bunn, 1996. Advanced backcross in tropically adapted indeterminate common bean. Euphytica 55:
QTL analysis in a cross between an elite processing line of tomato 15–19.
and its wild relative L. pimpinellifolium. Theor Appl Genet 92: White, J.W., J. Kornegay, J. Castillo, C.H. Cajiao & G. Tejeda, 1992.
213–224. Effect of growth habit on yield of large-seeded bush cultivars of
Tohme, J., D.O. Gonzalez, S. Beebe & M.C. Duque, 1996. AFLP common bean. Field Crops Res 29: 151–161.
analysis of gene pools of a wild bean core collection. Crop Sci Xiao, J., J. Li, S. Grandillo, S.N. Ahn, L. Yuan, S.D. Tanksley &
36: 1375–1384. S.R. McCouch, 1997. Identification of heterosis-enhancing QTL
Urrea, C.A. & S.P. Singh, 1995. Comparison of recurrent and con- alleles from a wild rice relative, Oryza rufipogon. Plant & Animal
gruity backcrossing for interracial hybridization in common bean. Genome V, P147, abstract.
Euphytica 81: 21–26.

euph4504.tex; 8/06/1998; 13:24; v.7; p.14