 

v

d.‘
u

#73

.. ‘ .
$523)"?

.1
:34.

ﬁt; 31‘

up.
'J 13. L

 

.135

T.

av"
rpm”
«

3‘11“." .33" w,
m w

v. ..
,_ an.
.

,3-

, .
'«W‘ '

r
"'4

‘3m
'2".

H

u

.53.". "

1'2"" ‘
,

4,2}? ,.

..
1v:

“ v'
,.
;'.';3{.:.j~;r

..

.3:

ﬁt.

an

~ "' L‘
my!

a
.,.

u.
”a

 

~e~ 1111111111111111111111111111111 11111111111111

l 3 12930

This is to certify that the
thesis entitled
IDENTIFICATION OF RAPD MARKERS ASSOCIATED WITH

CANNING QUALITY IN NAVY BEANS
presented by

Kimberly J. Walters

has been accepted towards fulﬁllment
of the requirements for

Master of Science degree in P1ant Breeding
and Genetics

  

 

Major professor /

Date BI/ZO/gg

0-7639 MS U i: an Affirmative Action/Equal Opportunity Institution

LIBRARY

Mk’higan State
University I

a

PLACE ll RETURN BOX to man this checkout iron your rocord.
TO AVOID FINES Mum on or baton duo duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU to An Afﬂnnutm MioNEmol Opportunity Institution
Walla-9.1

IDENTIFICATION OF RAPD MARKERS ASSOCIATED WITH CANNING .
QUALITY IN NAVY BEANS

By

Kimberly J. Walters

A THESIS

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE

Plant Breeding and Genetics Program
Department of Crop and Soil Science

1995

ABSTRACT

IDENTIFICATION OF RAPD MARKERS ASSOCIATED WITH CANNING
QUALITY IN NAVY BEANS

By

Kimberly J. Walters

RAPD markers associated with QTLs for canning quality in three populations of
navy beans were identiﬁed. Each population consisted of recombinant inbred lines (RILs)
derived from crosses between a superior canning and an inferior canning parent. Materials
were grown and evaluated at two locations for two years. Traits measured included:
visual appeal, texture, washed drained weight, hydration coeﬁcient and percent solids
lost. Marker-QTL associations were established using the General Linear Model
procedure with signiﬁcance set at P<0.05. The putative linkage groups were assigned
using MAPMAKER (LOD = 3.00).

Location and/or population speciﬁcity was common among the markers identiﬁed.
R2 values for individual markers ranged ﬁ'om 0.10 - 0.38 and R2 values for groups of
markers used in multiple regression analyses ranged from 0.15 - 0.74. Heritability
estimates were calculated on an entry mean basis for visual score, texture and washed
drained weight and were moderate to high in value ( 0.48 - 0.96). Negative correlations
were found between visual appeal score (VIS) and washed drained weight (WDWT),
percent solids lost (SL) and WDWT as well as hydration coefﬁcient (HC) and WDWT.
In contrast, VIS and texture (TXT) were positively correlated. Marker data reﬂected

these correlations.

ACKNOWLEDGMENTS

I would like to express my gratitude to my committee members, Dr. George L.
Hosﬁeld, Dr. James D. Kelly and Dr. Mark A. Uebersax for their invaluable assistance in
the design and execution of my research project and for their insight during the
preparation of this thesis. A special thanks goes to Dr. Kelly for helping me to learn the
art and science of plant breeding. I would also like to thank Jerry Taylor, Norm Blakely,
Theresa Wood, Bernie Marchetti and Jenny Clark for their help in the ﬁeld and in the
processing lab and all those who helped with the visual evaluations. Finally, to mom, dad,
Shany, Nate and Chrissy, without your love, support, encouragement, and prayers I

wouldn’t be where I am today-«thank you!

iii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................ v
LIST OF FIGURES ............................................................................................... viii
INTRODUCTION ................................................................................................. 1
LITERATURE REVIEW ...................................................................................... 3
Processing Quality ...................................................................................... 3
Genetics of Canning Quality ....................................................................... 7
Markers for Mapping/Studying QTLs ......................................................... 9
MATERIALS AND METHODS ........................................................................... 18
RESULTS ............................................................................................................. 28
DISCUSSION ....................................................................................................... 46
SUMMARY .......................................................................................................... 53
APPENDIX ........................................................................................................... 55
LIST OF REFERENCES ....................................................................................... 70

iv

Table 1.

Table 2.

Table 3.

Table 4.

Table 5a.

Table 5b.

Table 6a.

Table 6b.

Table 7a.

LIST OF TABLES

Minimum numbers of randomly placed molecular marker loci for
the likely detection of substantial associations with important QTLs
at diﬁ‘erent times after crossing two inbred lines in typical diploid,
tetraploid and hexaploid crop plants (adapted ﬁ'om Lande, 1992).

Signiﬁcant sources of. variation for visual appeal (VIS), texture
(TXT), hydration coeﬂicient (HC), washed drained weight

‘ (WDWT) and percent solids lost (SL) in population 2 (PZ),

population 4 (P4) and population 5 (PS) .........................................

Markers signiﬁcantly associated with canning quality traits in two
of the three navy bean populations. .................................................

Proportion of phenotypic variation accounted for by selected
markers using multiple regression ...................................................

Eﬂ‘ectiveness of selected markers for visual appeal (VIS), texture
(TXT), hydration coeﬁcient (HC), washed drained weight
(WDWT) and percent solids lost (SL) for identifying recombinant
inbred lines with high visual scores in population two ......................

Eﬂ‘ectiveness of a composite of markers for visual appeal (VIS)
ﬁ'om three populations for identifying recombinant inbred lines with
high visual scores in population two ................................................

Effectiveness of selected markers for visual appeal (VIS), texture
(TXT), hydration coefﬁcient (HC), washed drained weight
(WDWT) and percent solids lost (SL) for identifying recombinant
inbred lines with high visual scores in population four. ....................

Effectiveness of a composite of markers for visual appeal (VIS)
from three populations for identifying recombinant inbred lines with
high visual scores in population four. ..............................................

Eﬁ‘ectiveness of selected markers for visual appeal (VIS), texture
(TXT), hydration coefficient (HC), washed drained weight
(WDWT) and percent solids lost (SL) for identifying recombinant
inbred lines with high visual scores in population ﬁve. .....................

16

34

36

37

39

39

41

41

42

Table 7b.

Table 8.

Table 9.

Table 10.

Table A 1.
Table A2.
Table A3.
Table A4.
Table A5.
Table A6.
Table A7.
Table A8.

Table A9.

Table A10.

Table All.

Table A 12.

Table A13.

Table A14.

Eﬁ'ectiveness of a composite of markers for visual appeal (VIS)
from three populations for identifying recombinant inbred lines with

high visual scores in population ﬁve. ............................................... 42
Eﬂ‘ectiveness of selection based on a composite of markers for

visual appeal (VIS) and texture (TXT) from three populations ........ 42
Heritability estimates for visual appeal (VIS), washed drained

weight (WDWT) and percent solids lost (SL). .............................. 44
Correlation coeﬂicients between canning quality traits in three

populations of navy beans. .............................................................. 44
Form of the analysis of variance for VIS in population 2. ................ 55
Form of the analysis of variance for TXT in population 2 ................ 55
Form of the analysis of variance for WDWT in population 2 ........... 55
Form of the analysis of variance for VIS in population 4 ................. 56
Form of the analysis of variance for TXT in population 4 ................ 56
Form of the analysis of variance for WDWT in population 4 ........... 56
Form of the analysis of variance for VIS in population 5 ................. 57
Form of the analysis of variance for TXT in population 5 ................ 57
Form of the analysis of variance for WDWT in population 5 ........... 57

Markers signiﬁcantly associated (P<0.05) with visual score in three
populations of navy beans. .............................................................. 58

Markers signiﬁcantly associated (P<0.05) with texture in three
populations of navy beans. .............................................................. 59

Markers signiﬁcantly associated (P<0.0S) with washed drained
weight in three populations of navy beans ....................................... 60

Markers signiﬁcantly associated (P<0.05) with hydration coeﬁcient
in three populations of navy beans ................................................... 61

Markers signiﬁcantly associated (P<0.05) with percent solids lost in
three populations of navy beans. ..................................................... 62

vi

Table A 15

Table A16.

Table A17.

Table A. 18.

Table A19.

Table A20.

Table A21

Markers selected or multiple regression analysis .............................. 63

Chart showing marker genotypes and their relationship to canning

quality traits in population 2 ............................................................ 64
Chart showing marker genotypes and their relationship to canning
quality traits in population 4 ............................................................ 65
Chart showing marker genotypes and their relationship to canning
quality traits in population 5 ............................................................ 66
- Proposed linkage groups for population two. .................................. 67
Proposed linkage groups for population four ................................... 68
Proposed linkage groups for population ﬁve ................................... 69

vii

Figure 1a.

Figure lb.

Figure 2a.

Figure 2b

Figure 2c.

Figure 2d

Figure 2e

Figure 3

LIST OF FIGURES

Parent lines exhibiting desirable canning quality (‘Seafarer’ and
N81099) .......................................................................................

Parent lines exhibiting undesirable canning quality (‘Cumulus’ and '

N84004) .......................................................................................
RILs ﬁ'om population two exhibiting desirable canning quality ......
RILs from populationtwo exhibiting undesirable canning quality ..

RILs ﬁ'om population four exhibiting desirable canning quality
(top) and undesirable canning quality (bottom). ............................

RILs ﬁom population ﬁve exhibiting desirable canning quality ......
RILs from population ﬁve exhibiting undesirable canning quality ..

Percentage of markers signiﬁcantly associated with canning quality
traits at both locations using data combined over years ..................

19

20

29

3O

31

32

33

35

INTRODUCTION

In the United States, 75% of all dry beans (Phaseolus vulgaris) and 98% of all
navy beans grown are processed. As a result of consumer preferences regarding size,
color, texture and overall appearance of the canned product, processors have developed
culinary standards which beans must meet. Cultivars which fail to meet such standards
may be unacceptable.

Breeding for canning quality is difﬁcult for several reasons. First, canning quality
is a quantitative trait, hence, it can not be evaluated until later generations, which
introduces a measure of inefﬁciency to a brwding program. Lines which have been
selected and advanced based on agronomic traits will have to be discarded if they fail to
meet canning standards. Secondly, to ﬁtlly evaluate canning quality, several component
traits need to be measured. However, the array of objective and subjective tests needed to
adequately assess canning quality are not practical for a large number of samples
exhibiting small diﬂ‘erences, and are costly and time consuming. One alternative to
traditional selection procedures for canning quality may be marker assisted selection
(MAS). With this technique, improvement of a desired trait is accomplished by selecting
for markers tightly linked to the trait of interest. Molecular markers associated with
quantitative trait loci (QTLs) that inﬂuence canning quality would increase selection
eﬁciency by allowing for evaluation in earlier generations, thereby reducing the number
of lines advanced, and providing a faster, less expensive and less subjective measure of

canning quality.

2
Detailed genetic linkage maps have enhanced eﬂ‘orts to identify markers associated

with QTLs in crops such as tomato (Tanksley et al., 1982; Osborn et al., 1987; Martin et
al., 1989) and maize (Stuber et al., 1980, 1982, 1987; Edwards et al., 1987). Fewer,
detailed genetic studies have been conducted in dry beans due to the lack of available
markers. The development of restriction ﬁagment length polymorphism (RFLP)
technology has provided enough markers to construct a linkage map for common bean
(Nodari et al., 1992). Although this linkage map will be useful for some genetic studies
and certain breeding strategies, it may have limitations. Recent reports suggest that there
is limited RFLP-based genetic diversity between commercially adapted genotypes of bean
which necessitates wider crosses for mapping purposes (Chase et al., 1991; Guo et al.,
1991; Nodari et al., 1992). In addition to the difﬁculty of identifying RFLPs among
closely related genotypes, RFLP technology is time consuming and costly.

Recently, Haley et al., (1994a) showed that the random ampliﬁed polymorphic
DNA assay (RAPD) revealed useﬁsl levels of polymorphism within races of P. vulgaris
and suggested that RAPD-based linkage maps for representative populations of common
bean may be useful for breeding purposes. Based on the demonstrated utility of RAPD
markers in common bean, the current research was conducted in order to identify RAPD

markers associated with canning quality traits in three populations of navy beans.

LITERATURE REVIEW

Processing Quality

Acceptable canning quality in dry beans is desired by consumers, processors and
plant breeders, however, what constitutes ‘acceptable” varies over time and for diﬁ‘erent
consumer groups. Consumers are most aware of the texture, wholesomeness, and overall
appearance of processed beans (Wassimi, et al., 1990). Adams and Bedford (1973) stated
that texture can be separated into three components: adhesiveness (force required to
remove material from the mouth), gumminess (energy required to disintegrate the
substance) and ﬁrmness (force required to penetrate the substance). Firmness can be
assessed using a Kramer, Shear Press (Food Technology Corp., Rockville, Md.) which
measures the force required to bring cooked beans to point of rupture and catastrophic
breakdown of individual seeds. Although this measurement does not take into
consideration the other ln'nesthetic properties described for texture, it can be used to
evaluate consumer acceptability as regarding mouthfeel. Ruengsakularach et al., (1994)
reported a strong correlation (r = 0.97) between canned bean shear texture and pasting
torque values of whole bean ﬂour. Because the pasting test requires only 2 g dry bean
solids, it may be useful for selection purposes in early generation breeding lines. Degree
of splitting (tendency of beans to break apart, burst or disintegrate), clumping (adherence
of individual grains because of starch or pectin exudation), processed color, uniformity
and condition of the cooking broth contribute to the overall appeal of canned beans. An
acceptable cooked product would have few split beans, no clumping of beans, a light
amber color, uniform shape and size of individual beans, and no starch extruded into the

3

4
cooking broth. Evaluations for these traits -are generally subjective although color can be

objectively evaluated using a Hunter Lab Color and Color Diﬂ‘erence Meter (Hunter
Associates, Fairfax VA). To conduct visual evaluations, canned beans are displayed on
trays and rated by a panel of qualiﬁed personnel using a hedonic scale. Sample means are
determined and those with standard or above standard ratings are selected. An objective
measurement, such as the washed drained weight (WDWT), may also be useﬁrl as an
indicator of visual appeal. The WDWT is measured by rinsing, draining and weighing
canned samples. A low washed drained weight might be observed if excessive splitting
and loss of solids occurred during processing (Hosﬁeld, 1991).

While processors are constrained by consumer expectations, they also seek beans
which lend themselves to ease of preparation and greater processing emciency. Traits
such as rapid and uniform seed expansion and higher WDWTs are desired (I-Iosﬁeld,
1991). In addition to indicating an intact, desirable product, a high WDWT reduces the
amount of beans nwded to ﬁll a can to the desired volume, thereby increasing proﬁts to
processors. A washed drained weight ratio (WDWTR = WDWT/soaked bean weight) can
also be calculated and can be used to rapidly compare processed bean samples. The
hydration coefﬁcient (HC) is a measure of the weight increase during soaking due to water
uptake (soaked bean weight/dry bean weight). An optimum value of 1.8 indicates a well-
soaked bean, whereas lower values may be less desirable for processing as longer soaking
times are often necessary (I-Iosﬁeld, 1991).

Although canning quality traits are inﬂuenced by seed characteristics, they are also
affected by external factors such as storage conditions, soaking procedures and processing

conditions. Uebersax and Bedford (1980) found that canning quality deteriorated (higher

5
textures and browning) with increased storage time, temperature and relative humidity.

Darkening of seeds during storage has been attributed to the polymerization of phenolic
compounds which is enhanced by high temperatures and high humidity levels (Burr et al.,
1968; Garruti and Boume, 1985). Firm textures may be attributed to intact middle
larnellae and the failure of cotyledonary cells to separate during cooking (Srisuma, et al.,
1989). Absorbed water and heat applied during the cooking process cause degradation of
the middle lamella allowing cells to separate and soften (Uebersax and Ruengsakulrach,
1989). Beans stored at high temperature and humidity may have higher levels of certain
phenolic acids which aide in the cross-linking of pectin in the middle lamella (Srisuma, et
al., 1989). In seeds with intact middle lamellae, cells fail to separate and soften during
processing resulting in high texture values. The presence of an intact middle lamella may
also inhibit water absorption and result in low washed drained weights (Srisuma et al.,
1989). Storage conditions of 75% relative humidity or lower and 20 C or lower have
been shown to produce stable canning quality in navy beans (Uebersax and Bedford,
1980)

Processing conditions can also affect canning quality and must approximate
commercial protocols to properly assess potential varieties. Soaking is the ﬁrst step in
processing beans and plays a critical role in seed coat softening leading to uniform seed
expansion as well as ensuring product tenderness after cooking (Hosﬁeld, 1991).
Diﬂ‘erences in the rate of water uptake have been observed among cultivars (Deshpande,
et al., 1984). Although the precise reason for such differences is not clear, the structure
and composition of the seed coat, hilum and micropyle have been associated with different

rates of water absorption (Agbo et al., 1987). Soaking studies by Snyder (193 6) indicated

6
that, in some bean seeds, water uptake during soaking is primarily through the micropyle

and germinal area. However, Synder’s results and those of Powrie et al., (1960)
suggested that in navy beans, water uptake occurs readily through the seed coat. These
authors sealed the micropyle and germinal areas with beeswax or silicone grease.
Therefore, differences in seed coat structure and composition among cultivars may affect
the rate of water uptake. In a comparison of water uptake in small white, black and pink
beans, it was observed that a loosely arranged cell structure on the raphe-side of the hilum,
a deeply-grooved hilar ﬁssure, a narrow tracheid bar and a thinner seed coat also
contribute to faster water uptake (Deshpande and Cheryan, 1986).

To ensure suﬁcient water absorption during soaking, several factors must be
considered: duration of soak, temperature of soak water, and composition of soak water.
The length of the soaking period is critical to ensure that enough water is absorbed
however, longer soaking times can result in decreased texture values (U ebersax and
Bedford, 1980). Soak-water temperature and calcium content are important factors
because they aﬂ‘ect the rate of water uptake. At higher temperatures and lower calcium
concentrations seeds absorb water faster (Uebersax, and Bedford, 1980). However,
calcium should not be eliminated ﬁ'om the soak water because it forms insoluble salts with
pectic acid, a major constituent of the seed coat, and helps to maintain seed ﬁrmness and
integrity during processing (Uebersax and Bedford, 1980). In their study, Uebersax and
Bedford (1980), evaluated different soaking procedures and different calcium
concentrations in the soak water for their eﬂ‘ect on canning quality. From these studies it
was determined that a two step soaking procedure (30 minute cold soak at 23 C followed

by 30 minute hot soak at 88 C) with at least 50 ppm calcium gave the best results. At

7
levels lower than 50 ppm calcium, greater bean damage and gelatinization were observed

whereas, excessive calcium levels restricted water uptake and caused slightly lower
washed drained weight values.

Although canning quality is inﬂuenced by harvest and post-harvest handling and
processing procedures, there is a genetic component (Hosﬁeld et al., 1984; Wassirni et al.,
1990). The identiﬁcation of diﬂ‘erences between cultivars that arise because of genetic
diﬂ'erences should make breeding for canning quality more efﬁcient. Selection for canning
quality can be achieved by minimizing processing variables. All materials should be
handled and processed in the same manner by using a standard protocol that simulates
commercial processing practices. For processing navy beans, a two stage soaking
procedure is the most desirable because it maximizes differences between genotypes for

water uptake and softening (I-Iosﬁeld and Uebersax, 1980).

Genetics of Canning Quality

Canning quality behaves as a classic quantitative trait because phenotypes are
continuously distributed either because of a number of controlling genes each with a small
eﬂ‘ect or a few genes strongly modulated by environmental conditions. Because of its
quantitative nature, canning quality should be evaluated for several years at more than one
location which requires a large and uniform seed sample. The number of genes inﬂuencing
canning quality traits is unknown with the exception of seed coat color and condition
which have been extensively studied and are well characterized (Leakey, 1988). Genes
controlling pigment formation are numerous and interact closely to create a diversity of

colors and mottles. In navy beans, the most important color factor is the total pigment

8
enabler/suppresser gene (P gene). All genotypes with pp have white ﬂov'vers, white seed

coats and produce no pink, red, or mauve coloration in any vegetative tissue regardless of
the alleles present at other color genes (Leakey, 1988). The .1 gene, (also called the “joker”
gene, is important with respect to canning quality because of its eﬂ‘ect on seed coat color
and condition. Genotypes with the dominant 1.] allele produce proanthocyanidins
(leucobases) which accumulate in the parenchyma layer resulting in a shiny seed coat
(Leakey, 1988). The dominant J allele also results in genotypes that have darkened hilum
rings, darken with aging and become increasingly indigestible (Leakey, 1988). These traits
are undesirable for commercial processing or domestic use, therefore, the recessive ji
genotype is preferred and has been selected for under domestication. Although seed shape
is an important canning quality trait only the truncata and fastigiate seed forms have been
genetically characterized. The homozygous miv miv genotype causes seeds with ﬂattened
ends (truncata) and the fast fast genotype causes seeds to be wider at one end and sloping
at the other end (fastigiate) (Leakey, 1988). Both of these shapes are undesirable in navy
bean cultivars where ovoid seeds weighing 16-20 g/ 100 seeds are ideal (Adams and
Bedford, 1975). In addition to seed shape, seed size is important. Sax (1923) identiﬁed a
gene aﬂ‘ecting seed size in common bean which was linked to the P color locus. Johnson
and Gepts (1995) have since reported that the P locus is linked to the Phs locus, which
codes for phaseolin and has been correlated with seed weight. Vallejos and Chase (1991)
found an isozyme marker linked to a locus (Ssz-l) that affects swd size and overcomes
maternal control over seed size. This locus accounted for 30-50% of variation in seed size
which suggests that there must be other loci c0ntrolling this trait. Brothers et al., (1993)

reported a broad sense heritability for seed size in pinto beans ranging from 0.39 to 0.58

9
and determined that plant architecture and seed size were not associated as was previously

believed.

Canning quality is strongly inﬂuenced by the environment although the nature of
this environmental eﬂ'ect is disputed in the literature. Ghaderi, et al. (1984) found
signiﬁcant cultivar x location (G x E) interactions for clumps and splits, seed weight,
surface color, hydration ratio and processed bean moisture, suggesting that cultivars did
not perform consistently relative to each other in diﬂ‘erent environments. However,
Wassirni et al., (1990) found cultivar x location interactions to be non-signiﬁcant but
suggested that the lack of signiﬁcance may have been due to the use of segregating
populations (F 2 and F 3) rather than narrowly adapted, pure-line cultivars used in other
studies. Signiﬁcant genotype by season (G x Y) interactions for food quality have also
been observed and suggest an inconsistency of cultivar responses to seasonal variation
(Hosﬁeld, et al., 1984). In contrast, practical breeding experience indicates that although
variation for canning quality occurs over locations and years, ranking among breeding
lines and cultivars usually does not change signiﬁcantly (Kelly et al., 1994). Even so, it is
still necessary to conduct evaluations over a number of locations and years to identify

those cultivars with relative stability for the component traits aﬁ‘ecting canning quality.

Markers for mapping/studying Q11:

Evaluating a large number of samples with small diﬂ‘erences is time consuming,
expensive and inefﬁcient particularly if a number of canning quality traits are measured
(Ghaderi, et al., 1984). In addition to the time and expense involved, assessing canning

quality must be delayed until later generations when a large, uniform swd sample can be

10
obtained. This is ineﬁcient and costly because breeders must advance lines for several

generations that will be discontinued due to poor canning quality. Because traditional
selection strategies for canning quality are so costly and time consuming, alternative
selection strategies would be beneﬁcial. Recently, a study demonstrated a signiﬁcant
association (P<0.0001) between low seed density and superior canning quality in kidney
beans (Heil and McCarthy, 1992). Because density is easy to measure and requires a
small seed sample, it could be a useful selection criteria when breeding for canning quality
and could be used to evaluate early generation lines. Similar associations would have to
be established in other market classes of dry bean before this technique would have
general plant breeding utility.

Another selection criterion that needs to be explored is marker assisted selection
(MAS). This technique has received attention as a method for increasing selection
eﬁciency within breeding programs. Mth this technique, improvement in a desired trait
can be accomplished by indirectly selecting for a marker tightly linked to the trait of
interest. Previously, the lack of useful isozyme and morphological markers limited the
application of MAS, but with the development of DNA markers, such as RFLPs and
RAPDs, the technique is becoming a useﬁll tool for breeders. DNA markers linked to
qualitative traits have been found in a number of crops including common bean (Miklas, et
al., 1993; Haley, et a1, 1993; Haley, et al., 1994b), tomato (Martin, et al., 1991), lettuce
(Michelmore, et al., 1991), pea (Dirlewanger et al., 1994), rice (Mohan et al., 1994), oats
(Penner et al., 1993) and barley (Barua et al., 1993) and several are already being used in
breeding programs. While progress has been made toward identifying and utilizing

markers for qualitative traits, less advancement has been made for quantitative traits.

11
The ability to detect markers associated with quantitative traits was ﬁrst

demonstrated by Sax (1923) who reported the association of a seed coat color marker
with seed size in common bean. Other morphological markers (Rasmusson, 1935;
Everson and Schaller, 1955) as well as isozyme markers (Edwards at al., 1987; Tanksley
et al., 1982; Vallejos and Tanksley, 1983) have been found to be associated with QTLs in
crop plants. DNA based markers such as RFLPs and RAPDs oﬂ‘er even more
opportunities for use because of their abundance and neutral character. QTLs have
already been located in various crops using RFLPs. Martin et al., (1989) found three
RFLP markers associated with water use efﬁciency in tomato; Osborn et al., (1987)
detected one RFLP marker linked to soluble solids content in tomato; Nienhuis et al.,
(1987) identiﬁed four RFLP markers associated with the expression of 2-tridecanone-
mediated insect resistance in tomato; Abbo et al., (1992) identiﬁed several RFLP markers
associated with seed weight in lentil. A genetic linkage map with easily scored,
polymorphic loci uniformly distributed throughout the genome has been suggested as a
prerequisite for detailed genetic studies and marker-facilitated breeding approaches
(Stuber, 1992). Such maps are necessary to determine the location and effect of all QTL
for a trait, develop ﬂanking markers for more eﬂ'lcient selection, examine epistatic
relationships, and more accurately assess the phenotypic effect of a QTL. However, a
detailed linkage map is not a necessity for simply ‘tagging” QTL for selection purposes.

Although an RFLP-based linkage map is available for common bean it is not suitable for
evaluating molecular marker associations with canning quality for two reasons: 1.) a lack
of RFLP diversity among commercially adapted material of the same market class and 2.)

the possibility that certain markers are restricted to speciﬁc gene pools in common bean.

12
Studies (Chase et al., 1991; Guo et al., 1991; Nodari et al., 1992) have demonstrated that

because of a lack of RFLP diversity in common bean, wider crosses must be made in order
to facilitate map construction. The linkage map of common bean (Nodari, et al., 1992)
was constructed using crosses between genotypes from the Andean gene pool and the
Middle-American gene pool. Because navy beans all trace their lineage to ancestors
derived from crosses between genotypes of the Middle-American gene pool, mapped
RFLP markers may not have been polymorphic among the parents selected and therefore,
not useful for this study. Mildas et al., (1993) documented the problem of gene-pool-
speciﬁc markers in common bean with a marker linked to the Up-2, rust resistance gene.
The marker was identiﬁed using a BCst population from the Middle-American gene pool
into which the Up-2 gene, ﬁ'om the Andean gene pool, had been introduced. Although
the marker was useful in Middle-American lines for identifying the gene, all Andean
materials, with or without the gene, contained the marker.

Although most marker-facilitated studies of quantitative traits use RFLP markers,
RAPDs are also being used with increasing ﬁ'equency. Because RFLP-based linkage maps
are often developed using inter-speciﬁc crosses, they have. limited, direct application in
breeding programs that utilize intraspeciﬁc variability. Foolad et al., (1993) reported that
RAPD technology would produce sufﬁcient markers to construct an intraspeciﬁc genetic
map in tomato. RAPD markers are also being used to extend the genetic map for pea
(Weeden, 1994) and to construct a RAPD based map in common bean (Jung et al., 1994).
In addition to map construction, RAPD markers are being used to “tag”. QTLs for
quantitative traits. Tagging a QTL is accomplished by establishing associations between a

marker genotype and a QTL that has the desired effect on the trait of interest. Chalmers

13
et al., (1993) used doubled haploid populations to identify RAPD markers linked to QTLs

controlling the milling energy requirement of barley. Abbo et al., (1992) found RAPD and
RFLP markers linked to seed weight in lentil. In common bean, Haley et al., (1994a)
determined that useful levels of polymorphism (3 9%) could be detected among genotypes
of the same market class using the polymerase chain reaction (PCR) based RAPD assay.
In addition to greater levels of polymorphism, RAPDs are easier and less cumbersome to
work with than RFLPs and do not require the use of radioactive materials. Because
RAPDs are useful for Tagging” loci linked to qualitative traits in common bean (Haley et
al., 1993; Haley et al., 1994b; Miklas et al., 1993), this technology may prove fruitful for
‘tagging” QTLs for canning quality traits in navy beans. In light of previous arguments, it
is important to note that RAPD markers identiﬁed in this study may also be gene-pool-
speciﬁc, but they would be speciﬁc to the Middle-American gene pool from which navy
beans derive their lineage. Gene pool speciﬁcity will become an important issue only if
markers that are identiﬁed are to be used‘for selection in crosses between gene-pools or in
market classes originating ﬁom the Andean gene pool.

Regardless of the marker type used, any marker facilitated study must be designed
to give consideration to type of population used, population number and size, number of
markers evaluated, and the statistical analysis chosen to conﬁrm linkages. The types of
populations used vary depending on the crop and the trait to be measured. F2 populations,
backcross populations, and populations of recombinant inbred lines (RILs) derived by
selﬁng or as doubled haploids (DH) are often used. F2 and backcross populations are
generally limited to use with traits that can be measured on a single plant basis whereas,

traits that involve plot measurements such as yield and canning quality will require other

l4 ,
population types (Dudley, 1993). However, F2 populations have been used in maize to

establish marker-QTL associations for yield-related traits that can be measured on a single
plant basis (Stuber et al., 1987). Population size can also inﬂuence the power to detect
marker-QTL linkages. Beavis et al., (1994) noted that the power to detect QTLs is a
function of the number of progeny analyzed. Fewer progeny will reveal fewer QTLs. In
addition, if populations are small and the number of loci controlling the trait is large, the .
probability of ﬁnding the same markers associated in each population is low (Beavis et al.,
1994). Repeatability between populations also depends on whether the linkage phase of
the marker and the QTL is the same in all populations and, if epistasis is important, the
state of other alleles (Dudley, 1993). The number of markers analyzed will also inﬂuence
the probability of detecting substantial associations with QTLs. Lande (1992) has shown
that in a cross of two inbred lines with continued self-fertilization, the number of marker
loci needed to detect substantial associations with important QTLs is not very large (Table
1). These numbers are based on the equation 2Lt + c for random mating and 4L(l-1/2') +
c for complete selﬁng. In these equations, L is the total recombination map length in
Morgans, t is the number of generations alter the F2 or 8; generation, and c is the gametic
number of chromosomes.

Many diﬂ‘erent approaches have been suggested for marker-QTL analysis.
However, most methods fall into two categories: 1.) consideration of marker loci one-at-a
time or 2.) consideration of all marker loci at once. When considering loci one at a time,
simple F tests are appropriate if only two marker genotypes are present as in backcross or

R1 populations (Dudley, 1993). The one-at-a-time approach has several limitations as

15
noted by Lander and Botstein (1989). If the QTL does not lie at a marker locus, more

progeny may be needed to detect marker-QTL association and the phenotypic eﬂ‘ect of the
QTL may be underestimated. In addition, this approach cannot distinguish between a
tight linkage to a QTL with a small effect and a loose linkage to a QTL with a large eﬂ‘ect.
Interval mapping, using a maximum likelihood approach has been proposed as a method to
overcome such difﬁculties (Lander and Botstein, 1989). For interval mapping a
predetermined log of the odds ratio (LOD) score is used to establish regions, associated
with markers, that are likely to contain a QTL. The LOD score is the ratio of two
maximum likelihood estimates and indicates how much more probable it is that the data
arose from the presence of a linked QTL than the absence of a linked QTL. For example, a
LOD score of 3.00 would mean that linkage between the marker in question and a QTL is
1000 times more likely than the assumption of no linkage. One-at-a-time methods have
been compared with the interval mapping method (Stuber et al., 1992; Bubeck et al., 1993)
and both methods showed virtually the same results in detecting QTLs. Another alternative
is a combination of one—at-a-time analysis and multiple regression analysis. In this method,
markers are analyzed individually using F tests and then models including signiﬁcant
markers are analyzed by the General Linear Model procedure. Markers which are
signiﬁcant in this analysis are included in the combined model (Dudley, 1993). Regardless
of the method of analysis, an appropriate probability level for signiﬁcance must be
determined. Lander and Botstein (1989) suggest a genome dependent error rate of 0.001
while in most studies, the signiﬁcance of association is based on a comparison-wise error
rate of 0.05. Although false positives will likely be identiﬁed with a relaxed error rate,

comparisons of possible QTLs across environments and populations will be more fruitﬁll.

16

Table 1. Minimum numbers of randomly placed molecular marker loci for the likely detection of
substantial amociations with important QTLs at diﬁerent times after crossing two inbred lines in

typical diploid. tetraploid and hexaploid crop plants (adapted from Lande, 1992).

 

 

 

 

Generations since hybridization
. 1 5 10
Map length No. of Breeding Minimum number of markers needed to detect
Mus) chromosomes system important QTLs
10 10 Outcrossing 30 110 210
20 20 Outcrossing 60 220 420
30 30 Outcrossing 90 330 630
10 10 Selling 30 49 50
20 20 Selling 60 98 100
30 30 Setting 90 146 150

 

l7
Bubeck et al., (1993) explain that a marker signiﬁcant at 0.001 in one environment may be

signiﬁcant at only 0.05 in another environment. In this case, a relaxed Type I error rate
(falsely accepting a marker-QTL association) reduces the Type 11 error rate (rejecting a
true marker-QTL association).

After marker-QTL associations have been established, their usefulness in a MAS
strategy must be assessed. The value, ease, and cost of measurement and the genetic
control of the quantitative trait will determine if MAS is eﬂ‘ective and eﬂicient. If a trait is
diﬁcult and expensive to evaluate, controlled by a large number of genes with small
effects, or highly inﬂuenced by the environment MAS may be more emcient than
traditional selection based on phenotype (Dudley, 1993). Linkage distance between the
QTL and marker is also important. For MAS to be eﬂ‘ective, recombination between the
marker and QTL must be minimized by using markers tightly linked to the QTL or
markers ﬂanking the QTL. Heritability is another factor that affects the emciency of MAS.
When heritability is low or when the proportion of the additive genetic variance explained
by the marker loci exceeds the heritability of the trait, marker-assisted selection is more
emcient than phenotypic selection (Lande et al., 1990).

Although the long-term goal of this research was to establish a MAS protocol for
canning quality in navy beans, the immediate objective was to identify RAPD markers
associated with canning quality traits. The procedures used to identify markers associated
with canning quality traits are discussed and markers that are identiﬁed are evaluated for

use in MAS.

MATERIALS AND METHODS

Plant Materials

The visual characteristics evaluated (degree of split and clumped beans, soft vs
ﬁrm appearance, surface color, brine color and degree of starch particles present, and seed
size and shape considerations) are part of the standard evaluation for navy beans practiced
at Michigan State University. Parent genotypes for crosses were chosen to represent
extremes for canning quality in navy beans based primarily on visual appearance
characteristics. ‘Seafarer’ and N81099 are navy beans that exhibited good canning quality
over a ten year period while ‘Cumulus’ and N84004 are genotypes that consistently can
poorly (Fig 1). ‘Seafarer’ and ‘Cumulus’ are determinate, bush-type, navy bean cultivars
whereas N84004 and N81099 are upright, indeterminate, advanced breeding lines
developed at Michigan State University. In 1988, a diallel cross was made among parents
with good canning quality (including ‘Seafarer and N81099) and parents with inferior
quality (including ‘Cumulus’ and N84004). Three populations were derived from this
diallel to be used to identify RAPD markers associated with several canning quality traits.
Populations of recombinant inbred lines (RILs) were developed ﬁ'om the crosses
N84004/’Seafarer’ (population 2), N84004/N81099 (population 4), and
N81099/’Cumulus’ (population 5).

Individual F2 plants were advanced, using single seed descent, through the F4
generation with plants grown in the greenhouse and at the MSU Botany Farm, East
Lansing, MI. Moderate levels of ﬁeld selection during this period eliminated progeny that

were not adapted to Michigan growing conditions. F5 seed was planted at a rate of 25

18

19

O 9 , , ‘ a a '.
t- r-foge'etzs era-‘- 1' a.
T I- “ 2: ‘O/' 2:. ‘ .

I ' ,3 .
“'2‘. ‘.o’-

 

Figure 1a. Parent lines exhibiting desirable canning quality (‘Seafarer’ and N81099).

20

 

Figure lb. Parent lines exhibiting undesirable canning quality (‘ Cumulus’ and N84004).

g 21
seeds per row at the Bean and Beet Research Farm, near Saginaw, MI during the summer

of 1992. Single plants were selected from each line and the F55 seed was increased in a
winter nursery in Puerto Rico (1992/1993). A sub-sample was also gown in the

geenhouse to be used for DNA extraction.

Field Daigns

Fm RILs were gown at two locations (Bean and Beet Research Farm, near
Saginaw, MI and MSU Botany Farm, East Lansing, MI) during the summer of 1993 and
F5, RILs were gown at the same two locations in 1994. Prior to planting the seeds, they
were treated with Captan, Lorsban and Streptomycin sulfate, a combination treatment of
fungicide, insecticide, and bacteriocide, respectively. Soil at the Saginaw location is a
Misteguay (ﬁne, mixed, calcareous, mesic Aeric Haplaquepts) and has an average pH of
7.7. Sumcient levels of exchangeable potassium are present therefore, no potash is
recommended for the production of dry beans at the Bean and Beet Farm (Christenson,
1993). Fertilizer (179 kg/hectare of 21-7-0 + 4% Mn + 1% Zn) was applied during
planting. To control weeds, metolachlor (2 kg/hectare) and EPTC (3 kg/hectare) were
applied before planting and plots were cultivated and hand-hoed as needed. Plots were
also sprayed, prior to ﬂowering, with dimethoate (1 kg/hectare) to control leaf hoppers.
At East Lansing, soil is classiﬁed as Capac (ﬁne-loamy, mixed, mesic Aeric Ochraqualfs).
At the Botany Farm a 19-19-19 fertilizer was applied during planting at a rate of 223
kg/hectare and herbicides were applied as pre-plant incorporated (ppi) (EPTC 2

kg/hectare; triﬂuralin 1 kg/hectare; metolachlor 2 kg/hectare). Plots were cultivated 5

22
weeks after planting, hand-weeded throughout the summer and sprayed twice prior to

ﬂowering with dimethoate (1 kg/hectare) to control leaf hoppers.

Populations 2 and 5 were planted in 5x6 lattice desigrs (30 plots/replication) and
population 4 was planted in a 6 x 7 lattice design (42 plots/replication). Three replications
of each population were planted. Each plot consisted of two 6.1-m rows planted 50.8 cm
apart at a seeding rate of 13 seeds/m. For each RIL, days to ﬂowering, gowth habit, and
plot yield were recorded for ﬁrture selection purposes. When plants were mature, 4.3-m

sections ﬁom the middle of both rows in each plot were harvested and threshed.

Processing Procedures and Quality Evaluations

Harvested seed was rolled to remove debris and hand cleaned to remove split,
damaged, diseased, and discolored seed. Seed moisture percentage was determined on
250 g samples using a Burrows (Model 700) digital moisture computer. Duplicate samples
from each plot at each location were adjusted to 100 g solids based on moisture content
(% moisture) and placed in nylon mesh bags. Mesh bags were placed in a cold soak
(distilled water adjusted to 100 ppm calcium at 21 C) for 1/2 hour and then blanched in
water containing 100 ppm calcium at 88 C. Blanched beans were cooled in water for 5
minutes, drained for 10 minutes, poured into coded cans (size 303 x 406) and weighed to
determine the soaked bean weight (SBWT). Cans were ﬁlled with boiling brine (1.25%
sodium chloride, 1.57% sucrose, 100 ppm calcium), exhausted, sealed and processed in a
retort without agitation for 45 min at 116 C. After thermal processing, the cans were
cooled, inverted and stored at least two weeks prior to evaluation. The following traits

were evaluated for each RIL and parent line from two replications at both locations:

23
1. Soaked Bean Weight (SBWT) = weight of sample after hot soak

2. Hydration Coeﬁcient (HC) = SBWT/fresh weight

3. Washed Drained Weight (WDWT): Processed beans were opened and decanted on a
number 8 mesh sieve, rinsed in cool tap water and drained for 2 min at a 15° angle.
The weight of each sample was recorded (WDWT).

4. Texture (TXT): 100g samples of processed beans were placed in standard multiblade
shear compression cell and their textures determined using a Kramer Shear Press. Peak
heights were converted to Newtons using the following equation:

W x 1 kg x peak force x 9.8 = Newtons/100 g
100 2.04 lbs

*transducer force = 3000 lb range = 1/10

5. Percent Solids Lost (SL): 100g samples of textured beans were oven dried and then
weighed. Solids lost were computed using the following equation:

{100 g solids - (WDWT)(dry weight)/100g cooked beans} x 100

6. Visual Appeal (VIS): Processed beans were opened and the contents placed into
Styrofoam containers. A panel of qualiﬁed personnel used a 7 point hedonic scale to
rate the samples. The VIS trait was averaged across judges for each sample. The
seven point hedonic scale was:

1= Very undesirable

2= Moderately undesirable

3= Slightly undesirable

4= Neither desirable nor undesirable
5= Slightly desirable

6= Moderately desirable

7= Very desirable

DNA Extraction and RAPD Protocol
One young trifoliate leaf was harvested, from six plants of every parent line and six
plants of every F55 RIL. Tissue was lyophilized, gound and stored at -80 C. DNA was

extracted using a protocol outlined by Saghai-Maroof et al., (1984) with slight

24
modiﬁcations. Chloroform:isoamyl alcohol (24:1) was used in place of

chloroformzoctanol, RNAase A was added prior to a second chloroformzisoamyl alcohol
extraction, and DNA was resuspended in Tris-EDTA buﬂ‘er (TEN). In some cases, DNA
that was recalcitrant to PCR (polymerase chain reaction) ampliﬁcation was dissolved in a
high-salt solution and re-precipitated in ethanol to remove polysaccharides (Fang, et al.,
1992). DNA ﬁ'om parent lines was ampliﬁed by PCR using 390 diﬂ'erent, random,
decamer primers (Operon Technologies, Alameda, CA) to identify primers that generated
polymorphic bands. Primers were coded using a capital letter to indicate the kit used, an
Arabic number to indicate which. primer in the kit was used, and in some cases a lower-
case letter was used to indicate which band was scored (t= top band, m = middle band,
b=bottom band). Total reaction volume was 18.8 ul and each reaction contained the
following: 19 ng of genomic DNA template and 19 ng of single decamer primer (Operon
Technologies, Alameda, CA), 2 units of Stoffel F ragnent polymerase (Perkin Elmer
Cetus, Norwalk, CT), 1X Tris buﬁ‘er (10 mM Tris-HCl, pH 8.3; 10 mM KCl), 5 mM
MgC12, 200 uM of each dNTP (Perkin Elmer). Reaction components were overlaid with
25 ul of mineral oil and placed in a Perkin Elmer Cetus DNA Thermal Cycler 480.
Cycling conditions were: 3 cycles of 1 min/94 C, 1 min/35 C, 2 min/72 C; 34 cycles of 10
s/94 C, 20 s/40 C, 2 min/72 C; 1 cycle of 5 min/72 C; 1 s “Auto-segment extension” (for
extension portion of reaction). Ampliﬁed DNA was resolved by electrophoresis (1.4%
agarose gel, 0.5 ug ml'l etlridium bromide, 40 mM his-acetate, and 1 mM EDTA) and
each gel was photogaphed with Polaroid ﬁlm. Primers yielding polymorphisms between
N84004 and Seafarer (31 primers), N84004 and N81099 (36 primers), and N81099 and

Cumulus (34 primers) were screened against the respective populations. RILs were scored

25
for the presence (2) or absence (1) of the marker band. The number of markers scored for

each population was: population 2— 31 markers, population 4-- 36 markers, population

5- 34 markers.

Statisa'cal Analysis

Each of the measured traits was analyzed by the analysis of variance (AN OVA)
procedure with a factorial arrangement of treatments. The MSTAT computer program
was used (model number 17). Data from individual RAPDs were then analyzed against
each of the measured traits (VIS, TXT, HC, WDWT, SL) using the statistical progam
SAS proc glm (SAS Insititute, 1988). The probability level for acceptance of marker-QTL
association was P<0.05. In each population, traits were analyzed for each RIL for
separate locations and years (designated: loclyrl, loc2yr1, loclyr2, loc2yr2), locations
combined within years (designated: yearl , year2), locations combined over years
(designated: locl, loc2), and combined over locations and years (designated: comb). To
determine whether the presence or the absence of the marker band was associated with a
desired trait, data combined over years and locations were analyzed using a one-way
ANOVA (MSTAT). MAPMAKER was used to establish linkage goups (LOD
score=3.00; centimorgan distance threshold=15 cM). In each population, markers
associated with canning quality traits (P<0.05) at both locations or in both years were
used for multiple regession analyses where R2 indicated the phenotypic variation
accounted for by the markers (SAS Institute, 1988). Signiﬁcant markers were used in
their respective populations for selection and a composite of markers ﬁom all three

populations was used to select lines from each population. Markers for each trait were

26
ﬁrst tested for their ability to select the better canning parent in each population and then

for their ability to select the RILs with the most desirable canning quality ﬁ'om each
population. In order for a line to be selected it was required to have geater than one-half
of the markers for the trait being considered. Data were recorded for the number of lines
selected, average visual score of selected lines, percent of selected lines scoring above
3.50 for visual appeal and range of visual scores for selected lines. Visual appeal scores
averaged across locations and years were used in all cases. Selection using markers ﬁom
each population in their respective populations was compared with selection using a
composite of markers ﬁ'om all three populations. In addition, selection based on a
combination of visual and texture markers was compared with selection based on visual
markers alone.

Heritability estimates were calculated on an entry mean basis for VIS, TXT, and
WDWT using the formula:

hzgozc/O’z'r

02¢; = genotypic variance

0’: = total variance among means of RILs compared in r replications, I locations and y
years(r=2, l=2,y=2)

Variance components were estimated from the analysis of variance by appropriate

algebraic manipulation of terms comprising the expected values (Allard, 1960). A random

model was used (Table Al - A9). Total variance was calculated using the formula:

021': 020+O’cy/y‘l'O'IGL/14'OJGLYlly+0ﬂlrly

27
Correlations were computed for the following pairs of traits: VIS:WDWT,

VIS:SL VIS:HC, VIS:TXT, WDWT:SL, and TXT2HC. Correlation coeﬁicients were

compared with marker trends across traits.

RESULTS

Within each population there was obvious variation between the lines for canning
quality particularly for visual appeal (Fig 2a,b,c,d,e). For all of the measured traits,
signiﬁcant genotypic differences existed in each of the populations (Table 2). Location
effects were also sigliﬁcant for all traits, in all populations with the exception of VIS in
population ﬁve. Years were also a signiﬁcant source of variation for all traits except VIS
and SL. Interactions between G x L, G x Y, and G x L x Y were not consistently
sigliﬁcant for all traits or for a given trait in all populations. Signiﬁcant genotypic
variation permitted the identiﬁcation of several markers associated with each trait
(P<0.05) in each of the three populations (Table A. 10 - A. 14). R2 values for the markers,
calculated with data combined over years and locations, ranged from 0.13 - 0.29 in
population two, 0.10 - 0.38 in population four and 0.13 - 0.28 in population ﬁve.
Location speciﬁcity was evident as a large proportion of the markers were associated with
traits at only one of the two locations (Fig. 3). Many of the markers were also population
speciﬁc although a few were sigliﬁcantly associated with a canning quality trait in two of
the three populations. None of the markers were associated with a trait in all three
populations (Table 3). Markers selected for multiple regession analysis are listed in Table
A15. R2 values ranged from 0.15 - 0.74 depending on the trait and the population
considered (Table 4). Marker genotypes and their relationships to canning quality traits
are shown in Tables A16.A.18.

The effectiveness of indirect selection based on marker genotype varied for each

population and for the subset of markers used. In population two, markers associated

28

29

 

Figure 2a. RILs from population two exhibiting desirable canning quality

30

 

Figure 2b. RILs ﬁ'om population two exhibiting undesirable canning quality.

 

 

Figure 2c. RILs ﬁom population four exhibiting desirable canning quality (top) and
undesirable canning quality (bottom).

32

 

Figure 2d. RILs ﬁ'om population ﬁve exhibiting desirable canning quality.

33

 

Figure 2e. RILs from population ﬁve exhibiting undesirable canning quality.

34

Table 2. Sigliﬁcant sources of variation for visual appeal (VIS), texture (TXT),
hydration coeﬁicient (HC), washed drained weight (WDWT), and percent solids
lost (SL) in population 2 (PZ), population 4 (P4) and population 5 (PS).

 

 

 

 

 

 

Traits
VIS TXT HC WDWT SL
Populations
Source P2 P4 P5 P2 P4 P5 P2 P4 P5 P2 P4 P5 P2 P4 P5
Year (Y) ns us it 0‘ it it it t. ﬁt it t it #0 n5 n8
Mon (L) .0 it [IS I. it it it 0 it it t. I. .0 t O.
Y X L us it t ﬁt .0 it ti as it it t. it t t. t
Reps/(YL) IIS it ﬁt it #0 ns t it ﬁt I. it it it it it
Genmm (G) I. t. t. it it O. i *3 II t. t. .0 O. t t
G X Y t t it it it ﬁt us it us it t it [IS ns as
G X L as it it t. it 0‘ t 118 us it us it ns ns t
G X L X Y us it it it it #0 #ﬁ us it it us it I ns ns
"' P<0.05
" P<0.01

ns = non signiﬁcant

35

Initial.) ‘
urn (b)
Inc (e)
IWDWNd)
ISL (a)

 

 

 

 

Figure 3. Percentage of markers signiﬁcantly associated with canning quality traits at both
locations using data combined over years.

Table 3. Markers sigliﬁcantly associated with canning quality traits in two of the three

36

 

 

navy bean populations.
Canningguality trait Marker Population 2 Population 4 Population 5
Vrsual Score Q11 Signiﬁcant Signiﬁcant
M11 Signiﬁcant Signiﬁcant
Hydration coefﬁcient C5 Signiﬁcant Signiﬁcant
M10 Signiﬁcant Signiﬁcant
N17 Signiﬁcant Signiﬁcant
X3 Signiﬁcant Signiﬁcant
016 Signiﬁcant Signiﬁcant
Y13 Signiﬁcant Signiﬁcant
Texture F5 Signiﬁcant Signiﬁcant
G4 Signiﬁcant Signiﬁcant
Washed drained weight P5 Signiﬁcant Signiﬁcant
Y4 Signiﬁcant Signiﬁcant
Percent solids lost M10 Signiﬁcant Signiﬁcant
P5 Signiﬁcant Signiﬁcant
N18 Signiﬁcant Signiﬁcant
Yl3 Signiﬁcant Sigliﬁcant
AC2 Signiﬁcant Sigm_ﬁ' cant

 

Signiﬁcant at P < 0.05

37

Table 4. Proportion of phenotypic variation accounted for by selected markers using
multiple regession.

 

 

 

Population 2 Population 4 Population 5
No. of No. of No. of
Trait markers R2 markers R2 markers R2

Visual score 1 0.20 6 0.50 4 0.52
Texture 3 0.39 3 0.29 5 0.40
Hydration

coeﬂ'rcient 2 0.15 5 0.74 6 0.65
Washed drained

weight 6 0.42 0 ~— 3 0.38
Percent solids lost 3 0.35 8 0.52 2 0.31

 

using data combined over years and locations

38
with high HC and WDWT values and low SL values caused the inferior canning parent

(N 84004) to be selected over the superior canning parent (‘Seafarer’). Markers for high
VIS and TXT values caused ‘Seafarer’ to be selected over N84004. When the markers
developed from population two were used for selection in population two, markers
associated with VIS were best for identifying the superior RILs (based on visual score).
Lines selected using VIS markers had an average VIS score of 3.9 and individual line
scores ranged ﬁom 3.4 to 4.5 (range = 1.1). Eighty-one percent of the selected lines
scored above 3.50 for visual appeal. When markers for the other traits were used,
selected lines had lower average VIS scores, larger ranges for VIS scores, and fewer
scored above 3.50 (Table 5a). The composite of VIS markers from all three populations
was not as effective at selecting the best canning lines ﬁ'om individual populations as the
VIS markers used in the populations ﬁ'om which they were identiﬁed (Table 5b). The
average VIS score for all selected lines was 3.7 and individual line scores ranged from 2.7
- 4.5 (range = 1.8). Seventy-seven percent of the selected lines scored above 3.50.

In population four, markers associated with all traits except a high HC identiﬁed
the superior canning parent (N 81099). When markers developed in population four were
used for selection in population four, markers for VIS and TXT were best able to identify
the lines with high VIS scores (Table 6a). However, lines selected using TXT markers had
a wider range of VIS scores and had scores below 3.00. The average VIS score for lines
selected using VIS markers was 3.6 and individual line scores ranged ﬁ'om 3.1 to
4.2 (range = 1.1). Of the selected lines, 47% scored above 3.50. The average VIS score
for lines selected using TXT markers was 3.6 and individual line scores ranged from 2.9 -

4.2 (range = 1.3). Of the selected lines, 63% scored above 3.50. When the composite of

39

Table 5a. Eﬁ‘ectiveness of selected markers for visual appeal (VIS), texture (TXT),
hydration coeﬁcient (HC), washed drained weight (WDWT), and percent solids lost (SL)
for identifying recombinant inbred lines with high visual scores in population two.

 

Range of
visual scores No. and percent of selected
Marker No. of lines Average visual score of for selected lines scoring above 3.50 for

 

ype selected selected lines lines visual score
V18 11 3.9 1.1 9/11 =81%
TXT 9 3.5 2.1 6/9 =66%
HC 7 3.7 1.0 4/7 857%
WDWT 11 3.6 2.1 7/11 =64%
SL 15 3.6 2.1 8/15 =53%

 

Table 5b. Eﬂ‘ectiveness of a composite of markers for visual appeal (VIS) ﬁ'om three
populations for identifying recombinant inbred lines with high visual scores in population
two. '

 

Range of
visual scores No. and percent of selected
Marker No. of lines Average visual score of for selected lines scoring above 3.50 for
type selected selected lines lines visual score
VIS 13 3.7 1.8 10/13 = 77%

40
VIS markers was used for selection, the results were less desirable (Table 6b). The

average VIS score for selected lines was 3.4 and individual line scores ranged from 2.4 -
4.2 (range = 1.8). Only 35% scored above 3.50 for visual appeal. In population ﬁve, the
superior canning parent (N 8 1099) was selected over the poor canning parent (‘Cumulus’)
when markers for all traits, except a high TXT value, were used. Using only markers
developed from population ﬁve for selection, the best results were achieved with markers
for high VIS scores (Table 7a). The average VIS score of lines selected with VIS markers
was 3.4 and individual line scores ranged ﬁ'om 2.7 - 3.9 (range = 1.2). Of the selected
lines, 43% scored above 3.50 for visual appeal. Selection based on markers for other
traits gave lower average scores, wider. ranges for individual line scores and fewer lines
scoring above 3 .50. The composite of VIS markers was less eﬂ‘ective for selecting lines
with high VIS scores (Table 7b). Average VIS score for selected lines was 3.1 and the
range of individual line scores was 1.6 - 3.9 (range = 2.3). Only 23% scored above 3.50
for visual appeal.

Selection was more stringent but also more accurate, when only those lines that
were selected by the composite of VIS markers and by the composite of TXT markers
were considered. In all three populations, the number of lines selected was lower, the
average score of selected lines was higher, the range of VIS scores was smaller, and a
geater percentage of the selected lines scored above 3.50 (Table 8).

Heritability estimates for VIS scores, TXT values and WDWT values were
moderate to high (Table 9). Results ﬁ'om the linkage analysis allowed markers to be
assigned to ﬁve linkage goups in population 2, eight linkage goups in population 4 and

six linkage goups in population 5 (Table A. 19 - A21). The largest estimated distance

41

Table 6a. Effectiveness of selected markers for visual appeal (VIS), texture (TXT),
hydration coeﬂ'rcient (HC), washed drained weight (WDWT) and percent solids lost (SL)
for identifying recombinant inbred lines with high visual scores in population four.

 

Range of
visual scores No. and percent of selected
Marker No. of lines Average visual score of for selected lines scoring above 3.50 for

 

type selected selected lines lrnes visual score
V18 19 3.6 1.1 9/19 = 47%
TXT 8 3.6 1.3 5/8 = 63%
BC 19 3.3 1.7 6/19 - 32%
SL 10 3.4 1.3 4/10 = 40%

 

Table 6b. Effectiveness of a composite of markers for visual appeal (VIS) ﬁ'om three
populations for identifying recombinant inbred lines with high visual scores in population

four.

 

Range of
visual scores No. and percent of selected
Marker No. of lines Average visual score of for selected lines scoring above 3.50 for
type selected selected lines lines visual score ‘
V18 31 3.4 1.8 11/31 = 35%

42

Table 7a. Eﬂ‘ectiveness of selected markers for visual appeal (VIS), texture (TXT),
hydration coeﬁicient (HC), washed drained weight (WDWT), and percent solids lost (SL)
for identifying recombinant inbred lines with high visual scores in population ﬁve.

 

Range of
visual scores No. and percent of selected
Marker No. of lines Average visual score of for selected lines scoring above 3.50 for

 

type selected selected lines lines visualscore
V15 7 3.4 1.2 3/7 =43%
TXT 15 2.9 2.3 3/15 = 20%
KC 8 3.1 1.7 2/8 =25%
WDWT 12 2.6 1.5 1/12 -= 8%
SL 8 2.7 1.6 0/8 = 0%

 

Table 7b. Eﬁ‘ectiveness of a composite of markers for visual appeal (VIS) ﬁ'om three
populations for identifying recombinant inbred lines with high visual scores in population
ﬁve.

 

Range of
visual scores No. and percent of selected
Marker No. of lines Average visual score of for selected lines scoring above 3.50 for
type selected selected lines lines visual score
V18 13 3.1 2.3 3/13 = 23%

Table 8. Eﬂ‘ectiveness of selection based on a composite of markers for visual appeal
(VIS) and texture (TXT) from three populations.

 

Rangeof No.andpercentof
visualscores selected linesscoring

 

Marker No. of lines Avg. visual score of of selected above 3.50 for visual
type Population selected selected lines lines score
VIS:TXT 2 5 3.9 0.5 5/5 = 100%
VIS:TXT 4 l 4.2 1/1 -= 100%

 

VIS:TXT 5 10 3.6 2.3 3/10 = 30%

43
between linked markers was 8.7 cM in population 2, 14.4 cM in population 4 and 8.7 cM

in population 5.

Correlations between the following traits, VIS and WDWT, VIS and SL, VIS and
TXT, VIS and HC, WDWT and SL, TXT and HC, were not high enough to establish
strong associations although certain trends were observed (Table 10). There was a
negative relationship between VIS and WDWT. Correlations ranged ﬁ'om -0.26 to -0.66
with one non-sigliﬁcant value (1994, population 4). VIS and TXT seemed to be
positively associated with correlations ranging ﬁ'om 0.19 to 0.66. WDWT and SL lost
also tended to be negatively associated (correlation = -0.22 to -0.90) with the exception of
population four data from 1993. Markers that were signiﬁcantly associated with more
than one trait reﬂected these trends in most cases. Only one marker (Q11 in population
ﬁve) was signiﬁcantly associated with VIS and WDWT but it demonstrated a negative
relationship between the traits. Selection for high VIS required the presence of the Q11
band whereas selection for high WDWT required its absence. One marker (N 18b ﬁom
population four) was signiﬁcantly associated with VIS and TXT and it reﬂected the
positive correlation between the two traits. Selection for high VIS and a high TXT value
both required the absence of the N18b marker band.

Correlations between traits were not as evident when goups of markers were
considered. In population two, 11 lines were selected with population two-derived VIS
markers. Only 2 of those 11 lines were selected when TXT markers from population two
were used and 5 of the 11 were selected when WDWT markers were used. When the
composite of markers was used, 13 lines were selected based on VIS markers. Five of

these thirteen lines were selected based on TXT markers and 5 lines were also selected

Table 9. Heritability estimates for visual appeal (VIS), texture TXT) and washed

 

 

drained weight (wnwn.
Trait Population 2 Population 4 Population 5 Mean across populations
VIS 0.48 0.55 0.66 0.56
TXT 0.92 0.96 0.59 0.82
WDWT 0.72 0.62 0.67 0.67

 

Table 10. Correlation coefﬁcients between canning quality traits in three populations of

 

 

 

 

navy beans.
Population 2 Population 4 Population 5‘
Cumu- 1993 1994 Cumu- 1993 1994 Cumu- 1993 1994

Correlations lative lative lative
viszwdwt 0.58“ -0.66*"' -0.5 l " -0.26“ -0.49" NS -0.33"‘I -0.39‘"" -0.45"
viszsl 0.14‘I 0.21“ NS -0.12‘ -0.31“ NS NS 0.19" NS
vis:txt 0.49“ 0.66" 0.46" 0.22“ 0.33“ 0.19“ 0.42" 0.43“ 0.29“
viszhc -0.14‘ -0.32” 0.20‘I NS NS NS 0.19“ NS NS
txtzhc NS -0.42“ 0.45" 0.28"”I NS 0.34" 0.14“ NS NS
wdwtzsl -0.53" -0.50" -0.55" -0.52""" 0.27“ 41.90” -0.23" -0.2.‘3‘"‘I -0.22"

‘ P<0.05

“P<0.01

NS = non signiﬁcant

45 .
based on markers for WDWT. In population four, 19 lines were selected with population

four—derived VIS markers. TXT markers ﬁ'om population four permitted selection of 7 of
these 19 lines. When the composite of markers ﬁ'om all three populations was used, 31
lines were selected based on VIS markers. Only one line was selected using TXT markers
but it was a line also selected using VIS markers. Markers for WDWT were used to
select 18 of the 31 lines that were selected when VIS markers were used. VIS markers
speciﬁc to population ﬁve, permitted selection of 7 lines from population ﬁve. Of these 7
lines, 6 were also selected when TXT markers from population ﬁve were used. In
contrast, when WDWT markers were used only 1 of the 7 lines selected using VIS
markers was chosen. When the marker composites were used for selection in population
ﬁve, 13 lines were selected based on VIS markers, 10 of which were also selected when
TXT markers used. Three lines were selected based on WDWT markers, none of which

were selected when VIS markers were used.

DISCUSSION

Although most quantitative traits are inﬂuenced by the environment, many of the
studies associating QTLs with marker loci have been conducted using only one
environment. While this practice simpliﬁes the analysis and reduces the demands on time,
resources and money, it does not allow marker-QTL associations to be tested in different
environments. This may lead to poor decisions when choosing markers for MAS. Marker
studies that have been conducted at multiple locations have identiﬁed environment-speciﬁc
marker-QTL associations (Bubeck et al., 1993; Patterson et al., 1991). The location
speciﬁcity of markers for canning quality traits, previously described, was not surprising
because it is common for the canning quality of a genotype to vary from location to
location and because the analysis of variance indicated that location x genotype
interactions were statistically signiﬁcant for many of the traits in each population. One
possible explanation for the diﬁ‘erence in canning quality across locations could be
environmentally-sensitive QTLs. This would also explain the location speciﬁcity of the
marker-QTL associations.

Markers for QTLs that function consistently at a number of locations would be
preferable for MAS. However, location-speciﬁc markers may also be useful. Testing at
several locations is commonly practiced in order to identify genotypes that perform well
over a range of environments. By pyramiding location-speciﬁc markers into one
genotype, it may be possible to achieve the same results at a faster rate (Patterson, et al.,

1991).

46

47
Another issue that affects the usefulness of marker-QTL associations is population

speciﬁcity. Many of the marker studies that have been conducted utilize only one
population, which may also lead to poor choices or markers for MAS. Population-speciﬁc
marker-QTL associations were prevalent in this study but were expected because of small
population size. However, small populations were unavoidable due to the method of
population development that was utilized and the elimination of unadapted lines. Small
population sizes (27, 37, and 26 RILs in populations two, four and ﬁve respectively) and a
large number of genes inﬂuencing canning quality may be one reason for the population
speciﬁcity of the markers. When fewer progeny are analyzed, fewer QTLs can be
identiﬁed (Beavis et al., 1994). Therefore, if the ability to detect QTLs is reduced because
of small population size and the number of QTLs is large, the probability of detecting the
same QTLs in diﬂ‘erent populations is small. Beavis et al., (1994) demonstrated a similar
anomaly when the QTLs they found associated with yield in maize differed ﬁ'om those
found by Stuber et a1. (1992) using progeny derived ﬁom the same pair of parent lines.
However, the two experiments were conducted at different locations so it is possible that
environmentally sensitive QTLs contributed to the diﬁ‘erences. Other possible causes of
population-speciﬁc marker-QTL relationships are large linkage distances between the
markers and the QTLs and epistatic interactions. If the distance between the markers and
QTLs is sufﬁciently large, recombination may occur resulting in the markers being
separated from the QTLs in some individuals or entire populations. If epistatic
interactions are important for the expression of the QTLs, recombination between the
QTLs and other loci inﬂuencing the QTLs may result in different population speciﬁcities

of the marker-QTL associations. In this case, although the marker may still be linked to

48
the QTL, an individual or population may not have the other necessary genes present, in

. the appropriate state, for the QTL ,to be adequately expressed. Thus, in a statistical
analysis the marker would not be associated with the trait of interest. All of these
possibilities would need to be considered when deciding how to use population-speciﬁc
markers. Obviously, the absence of population-speciﬁc marker-QTL associations would
be more useful for MAS. However, population-speciﬁc markers may still be useful if
population speciﬁcity is due to small population size and a number of controlling genes
rather than large linkage distances and epistasis. If that is the case, using population-
speciﬁc markers to pyramid QTLs may result in lines that have more favorable alleles for
the desired trait. Whether or not the difﬁculty of such a task outweighs the beneﬁts needs
scientiﬁc scrutiny.

Another factor that inﬂuences the eﬁciency of MAS is the heritability of the
desired trait. For highly heritable traits (i.e few genes and little environmental inﬂuence),
MAS oﬁ‘ers no geater selection eﬂiciency than phenotypic selection except when MAS
allows for earlier evaluation of a trait than does phenotypic selection. For traits with low
heritabilities MAS would be more eﬂicient than phenotypic selection if the markers explain
a large proportion of the additive genetic variation (Lande, 1990). Canning quality can be
viewed as a “super” trait because no single variable can adequately describe the
characteristics of a sample. One needs to look at canning quality through dissection into
a number of deﬁned characters that can be individually measured. These individually
measured characters or components of canning quality are heritable but act as
quantitatively inherited traits when they are measured. Because the components of

canning quality behave as quantitative traits, substantial environmental inﬂuence on

49
canning quality was expected and has been documented (Hosﬁeld, et al., 1984; Ghaderi et

al., 1984). It was interesting that components of canning quality were so highly heritable,
indicating limited environmental variance. It was also interesting when the results of the
marker analysis were considered. Markers that were signiﬁcantly associated with each of
the measured traits were often location speciﬁc, suggesting environmentally-sensitive
QTLs. Therefore, one would expect the environment to have a strong inﬂuence on the
expression of these traits, resulting in lower heritability estimates than those that were
calculated. One possible reason for the high heritability values is that evaluations were
conducted at only two locations for two years. Heritability values may have been lower if
evaluations had been made for additional locations and years. It is also possible that
canning quality traits are controlled by a few major genes with large environmental
sensitivities rather than a large number of genes with small effects. If this is the case, one
would expect moderate to high heritability values for TXT and WDWT but would expect
the heritability of VIS to be lower than that of TXT or WDWT. This is because VIS is a
subjective measurement which is inﬂuenced by the perception of the following traits: the
amount of clumping and splitting of processed beans; cooking broth characteristics; and
cooked seed characteristics such as color, size and shape for the market class. VIS would
therefore be governed by the genes inﬂuencing the component traits, each with their own
particular environmental sensitivities. The heritability for VIS would be lower than for the
individual components if each were measured. As observed in all three populations, the
heritability of VIS is lower than the estimates for TXT and WDWT.
An accurate assessment of the effectiveness of the identiﬁed markers for MAS is

beyond the scope of this study. However, selecting among the three populations gave an

50
approximation of how well these markers were able to identify lines with high visual

scores and may have implications for traditional breeding strategies. Markers for VIS
were best for selection purposes in all populations. However, when the composite of
markers for VIS was used, more lines were selected and fewer of the selected lines had
acceptable VIS scores. Although this illustrates the diﬁiculty of using population-speciﬁc
markers to make selections in other populations it does not eliminate the possibility of
using these markers for MAS. Because canning quality is not evaluated until later
generations, the result is a mixture of superior and inferior lines in regards to this trait
that have to be evaluated by prescribed pilot plant methodologies at later generations.
Through the use of these markers, one should be able to separate lines that would be
unacceptable ﬁ'om those that would have acceptable canning quality. However, based on
the preliminary MAS studies some lines with unacceptable canning quality would be
selected. Thus the major advantage of MAS, as opposed to phenotypic selection, for
canning quality is that MAS can be done as early as the F2 generation. Early generation
selection would reduce the number of lines that would be advanced to the point of
consideration for release as cultivars. It is important to note that even if MAS is utilized,
it will still be necessary to process selected lines using a simulated commercial processing
protocol before they are released as cultivars. Using a combination of markers for VIS
and TXT may also increase the number of selected lines with acceptable canning quality
and reduce the number of selected lines with unacceptable quality. However, it may be
necessary to relax the selection criteria to be sure enough lines are advanced. The relative
eﬁciency of marker assisted selection, using the markers identiﬁed in this research,

compared to that of traditional phenotypic selection is a worthwhile research objective.

51
In a breeding program, lines that have been proven to exhibit superior canning

quality are often used repeatedly in crosses designed to combine canning quality with
other desirable traits. Population speciﬁc markers could be used to facilitate this breeding
practice. For example, markers associated with high VIS scores in population two
(‘Seafarer’lN 84004), were all present in ‘Seafarer’. Therefore, these markers could be
used to select for canning quality among progeny from crosses between ‘ Seafarer’ and
various breeding lines. Used in this fashion, population speciﬁc markers may have long-
term utility in a breeding progam.

Most dry bean breeders select for canning quality by visual assessment alone. The
results of the current research seem to validate such a practice. Whether selection is done
with markers or by phenotypic evaluation, one is selecting for QTLs with favorable
effects on the trait of interest. If selection based on markers (QTLs) for VIS is more
eﬁ‘ective than selection based on markers (QTLs) for the other traits, then it would seem
that phenotypic selection for QTLs with favorable effects on VIS would be more effective
than phenotypic selection based on QTLs for other traits. It is still necessary to evaluate
TXT, WDWT and HC but these evaluations can be made when a line is being considered
for release as a cultivar. Breeding lines which deviated substantially from accepted values
for the objectively measured traits would not be good candidates for release as cultivars.
They may, however, be useful as parents for improving other traits in the population such
as yield, disease resistance, plant architecture etc. Since there was a positive correlation
between VIS and TXT, lines selected based on VIS should have acceptable TXT values .

Although beans with high WDWTs are desired by processors, selecting for this

trait either with markers or by gavimetric methods would not be effective in a breeding

52
program. Markers associated with high WDWT values were not as useﬁll as markers for

high VIS for identifying lines with high VIS scores. In addition, the negative correlation
between VIS and WDWT suggests that if selection were based solely on high WDWT,
this procedure would have a negative effect on VIS. HCs do not usually deviate
substantially from the desired value therefore, lines selected on VIS alone are likely to
have acceptable values for this trait.

Based on this preliminary research, markers for canning quality traits may prove to
be useﬁll tools in the process of developing navy bean cultivars with superior canning
quality. Research efforts should now be directed at assessing the utility of the identiﬁed
markers for MAS in other populations of navy beans as well as populations of other

market classes of dry bean.

SUMMARY

The diﬁiculty of evaluating and selecting for canning quality has prompted dry
bean breeders to seek alternative methods of selection. MAS is one possibility that is being
explored. We have identiﬁed markers associated with several canning quality traits that
may be used in a MAS scheme. Because many of the markers are location and/or
population speciﬁc, which markers will be used and in what manner they will be used with
geatest advantage is unclear. Markers for VIS were best able to identify good canning
lines but still selected a number of poor canning lines. Selecting based on VIS markers
and TXT markers increased the average score of the selected lines. However, the number
of lines selected was often quite low which may necessitate reducing the number of
markers required for a line to be selected.

Heritability estimates for VIS, TXT, and WDWT were moderate to high but may
be inﬂated since broad sense heritability estimates were used. Although, lower heritability
values were expected, the calculated estimates may indicate that there are a few major
genes controlling TXT and WDWT. Because VIS is a multi-faceted trait that is
inﬂuenced by other canning quality traits, it would be controlled by more genes with
diﬂ‘erent environmental sensitivities. Thus a lower heritability would be expected for VIS
than for the individual traits. High heritabilities for traits may reduce the eﬂiciency of
MAS compared to traditional phenotypic selection.

In addition to heritability, the correlations between various canning quality traits
will inﬂuence how effective MAS will be. Since negative correlations between VIS and

WDWT were observed and initial selection attempts using markers indicate that markers

53

54
for high WDWT were not effective for selecting lines with desirable canning quality,

selecting for high WDWT would be counter-productive. The best strategy appears to be
selection based on visual evaluation regardless of how the selections are made (MAS or
phenotypic. selection). Recognizing that other traits will need to be measured when a line
is considered for release as a cultivar. Positive correlations between TXT and VIS and
little variation for HC indicate that lines selected for visual appeal are likely to have

acceptable values for other important traits.

APPENDIX

55

Table A. 1. Form of the analysis of variance for VIS in population 2.

 

 

Source Df ss MS E[MS]
Year(Y) (y - 1) 0.040 0.040
Location (L) (1 - 1) 81.877 81.877
LxY (1 - l)(y- 1) 0.768 0.768
Reps/(LXY) (r - l)ly 3.819 0.955
Genotype(G) (g - 1) 77.252 2.664 o1. + r62m+ rlo‘owr ryrr’aL + rlyozo
GxY (g- 1)(y- 1) 34.219 1.180 o3.+ro2m+rlo’ay
G x L (g -1 )(1 - 1) 25.225 0.870 o’,+ ragga-t- ryo’m,
G x L x Y (g - 1x1 - 1)(y - 1) 19.348 0.667 c’.+ ro’cM
Error (r - 1)(_gly - 1) 75.683 0.652 6’.

 

Table A2 Form of the analysis of variance for TXT in population 2.

 

 

Source Df 55 MS ElMS]
Year (Y) (y - 1) 1895220328 1895220328
Location (L) (1 - 1) 3223072473 3223072473
L x Y (1 - l)(y - 1) 765505.093 765505.093
Reps/(LxY) (r - 1)1y 132630.916 33157.729
Genotype(G) (g - 1) 1689453556 58257.019 o’.+ rozmar rlo’o,+ ryoza + rlyo’G
o x Y (3 - 1)(y - 1) 373603.975 12882.896 61,», rozm+ rlo’.W
G x L (g -1 )(l - 1) 484027.004 16690.586 o’.+ ro’ay-t- rya’ox,
G x L x Y (g - l)(l - l)(y - 1) 142905.077 4927.761 o1.+ ro’m
Error (r - l)(gly - 1) 724967.419 6249.719 6’.

 

Table A3. Form of the analysis of variance for WDWT in population 2.

 

 

Source Df 53 MS E[MS]

Yearm (y - 1) 620.817 620.817

Location (L) (l - 1) 28754.704 28754.704

L x Y (1 - l)(y - 1) 1192.604 1192.604

Reps/(LxY) (r - l)ly 2327.492 581.873

Genotype(G) (g - 1) 5406.725 186.439 o’.+ ro3m+ ﬂ030y+ ryo’m, + rlyo’o
G x Y (g - l)(y - 1) 1015.558 35.019 o’.+ ro’m+ rlo’ay

G x L (g -1 )(l - 1) 1260.296 43.458 o’.+ roams» ryo’m,

GxLxY (g-1)(l-1)(y- 1) 784.646 27.057 o2.+rc’m
Error (r - 1);eg - 1) 1611.258 13.890 0'2.

56

Table A4. Form of the analysis of variance for VIS in population 4.

 

 

Source Df ss MS - E[MS]
Year (Y) (y - 1) 1.134 1.134
Location (L) (1 - 1) 22.565 f 22.565
L x Y (1 - l)(y - 1) 5.582 5.582
Reps/(LxY) (r - 1)ly 5.476 1.369
Genotype(G) (g - 1) 79.671 - 1.992 o3. + razou+ rlozoy-t- ryo’aL + rlyo’a
G x Y (g - l)(y - 1) 20.179 0.504 o1,+ ro2m+ r163“
G x L (g -1 )(1 - 1) 35.538 0.888 02,-1- rozau-t- rye’m,
G x L x Y (g - 1)(1 - 1)(y - 1) 32.811 0.820 63.49 rc’m
Error (r - ugly- 1) 49.029 0.306 0“.

 

Table A5. F orm of the analysis of variance for TXT in population 4.

 

 

Source Df 55 MS E[MS]
Year(Y) (y -1) 962306.190 962306.190
Location (L) (l - 1) 1780250323 1780250323
L xY (1 -1)(y - 1) 315526.756 315526.756
Reps/(LxY) , (r - 1)ly 143770.317 35942579
Genotypc(G) (g - 1) 1434287275 35857.182 o3,+ razau-l- rlc’m,+ rye?“L + rlycr’G
G x Y (g - l)(y - 1) 251594.315 6289.858 c’.+ rozaLY+ r163."
G x L (g -l )(1 - 1) 328711.513 8217.788 63.+ rams» ryc’.ll
o x L x Y (g - 1)(1 - 1)(y - 1) 275151.644 6878.791 o3,+ rc’m
Error (r - 1L(_grly - 1) 405886.402 2536.790 63.

 

Table A6. Form of the analysis of variance for WDWT in population 4.

 

 

Source Df 55 MS E[MS]
Year (Y) (y - 1) 183.003 183.003
Location (1.) (l - 1) 26712195 26712195
L x Y (1 - l)(y - 1) 468.488 468.488
Reps/(LxY) (r - 1)ly 17716.921 4429.230
Genotype(G) (g - 1) 7091.735 177.293 02,-!- ro‘mat I‘IO’zay'l' ryo’a,L + rlyc’a
G x Y (g - l)(y - 1) 2583.247 64.581 o3,+ ro3m+ r163.”r
G x L (g -l )(l - 1) 2198.430 54.961 o’.+ ro’m+ tyczm
G x L x Y (g - 1)(l - 1)(y - 1) 2074.137 51.853 'o’,+ ro’our
Error (r - 1)(ngy - 1) 6271.829 39.199 0’2.

 

57

Table A7 . Form of the analysis of variance for VIS in population 5.

 

 

Source or $5 MS E[MS]
Year(Y) (y - 1) 9.832 9.832
Location (L) . (1 - 1) 0.041 0.041
L x Y (1 - 1)(y - 1) 1.020 1.020
Reps/(LXY) (r - 1)1y 3.619 0.905
Genotype(G) (g - 1) 84.974 3.035 o’,+ rolm+ rlo’oy-l- ryo’m, + rlyo’a
G x Y (g - 1)(y - 1) 25.257 0.902 c’.+ [Uzmy't' rlczav
G x L (g -1 )(l - 1) 19.296 0.689 o’.+ ro’gu't- ryo’m,
G x L x Y (g - 1)(l - 1)(y - 1) 15.439 0.551 o’.+ r020“
Error (r - 1)(_gly - 1) 25.545 0.228 6’.

 

Table A8. F cm of the analysis of variance for TXT in population 5.

 

 

Source Df 83 MS E[MS]
Year (Y) (y - 1) 3068117.148 3068117148
Location (L) (l - 1) 2263606882 2263606882
L xY (1 - 1)(y - 1) 752532.771 752532.771
Reps/(LxY) (r - 1)ly 2432.593 608.148
Genotype(G) (g - 1) 3045959477 108784.267 0”. + r63m+ rlo’oyaa ryoﬂg + rlyo’o
G x Y (3 - l)(y - 1) 991859.853 35423.566 o3,+ rczmar rlc’av
G x L (g -1 )(1 - 1) 689272.804 24616.886 o3.+ rolm+ ryo’or,
G x L x Y (g - 1)(1 - 1)(y - 1) 420679.962 15024.284 o’,+~rc’ai,,r
Error (r - 1)(gly - 1) 654820.797 5846.614 6’.

 

Table A9. Form of the analysis of variance for WDWT in population 5.

 

 

Source Df ss MS E[MS]
Year (Y) (y - 1) 823.142 823.142
Location (L) (1 - 1) 18451.556 18451.556
L x Y (1 - 1)(y - 1) 8975.211 8975.211
Reps/(LxY) (r - 1)1y 952.569 238.142
Genotype(G) (g - 1) 7963.754 284.420 o’.+ razou-i- rlo’oyer ryozm, + rlyo’a
G x Y (3 - 1)(y - 1) 2229.233 79.615 o’.+ ro’m+ rlc’ay
G x L (g -1 )(l - 1) 2503.569 89.413 o3.+ rozggyi- ryc’m,
G x L x Y (g - 1)(1 - l)(y - 1) 2086.039 74.501 o’,+ [02913
Error (r - l)(_gly - 1) 3358.181 29.984

 

58

 

 

 

 

 

 

Table A. 10. Markers sigliﬁcantly associated (P<0.05) with visual score in three
populations of navy beans.
Combination of data used in analyses
Yearl Locerl Loc2le Year2 Locer2 Loc2Yr2 Cmb' Locl Loc2
Pop. Signiﬁcant markers
2 A18 A18 A18 A18 A18 A18
L18 L18
Q3 Q3 Q3
Q11 Q11
Y10
4 A14 A14 A14 A14
818 318 B18 B18 B18
C5 C5 C5 C5
H12t H12t
H12b
H18b
12 I2 12 I2 12 I2 12 I2
118t Il8b
19b J9b
M2 M2 M2
M10 M10 M10
M11 M11 M11 M11 M11 M11 M11
N17 N17 N17
N18t N18t N18t
N18m N18m N18m N18m
N18b N18b N18b N18b N18b N18b
016
P16
Q11 Q11 Q11
X7
X17 X17 X17
Y3t Y3t Y3t Y3t Y3t Y3t
Y5 Y5
AC2
5 B15 815
11 11 11 11
L8 L8 L8 L8 L8
M11 M11 M11 M11 M11
M19
016 016
P5 P5
Q11 Q11 Q11 Q11
X3 X3 X3 . X3 X3
Y4b Y4b Y4b
Y13 Y13 Y13

 

'data combined over years and locations

59

Table A 1 l. Markers sigliﬁcantly associated (P<0.05) with texture in three populations of
navy beans.

 

Combination of data used in analyses
Yearl Locerl Loc2le Year2 Loc1Yr2 Loc2Yr2 Cmb' Locl Loc2

 

 

 

 

 

Pg) Sigpiﬁcant markers
2 - F51 F51 F51 F51 F51 F51 F51 F51 F51
F5b
G4 G4 G4 G4 G4 G4 G4 G4 G4
12 12
M19
N17 N17
P5 P5 P5 P5
Y4b Y4b Y4b Y4b Y4b Y4b Y4b
A86 A86
AC8 AC8
4 A14
818 318
12 12
J91
M5 M5 M5 M5
N18b N18b N18b N18b N18b N18b N18b
016
X7 X7 X7 X7 X7
X17 X17
Y3t
Y3b Y3b Y3b Y3b Y3b Y3b Y3b Y3b
Y5 Y5 Y5 Y5
5 F51 F51 F51 F51 F51
G4 G4 G4 G4 G4
L18 L18 L18 L18
P5 P5 P5 P5
Q11 Q11 Q11 Q11 Q11 Q11 Q11 Q11
Y14 Y14 Y14 Y14 Y14

AC2
1’data combined over years and locations

 

60

Table A12. Markers sigliﬁcantly associated (P<0.05) with washed drained weight in
three populations of navy beans.

 

 

 

 

 

 

Combination of data used in muses
Yearl Locerl Loc2le Year2 Locer2 Loc2Yr2 Cmb' Locl Loc2
Pop Signﬁcant markers
2 A18 A18
F51 F51
F5b F5b F5b F5b
12 12
M10 M10 M10 M10 M10 M10 M10 M10 M10
O31 O31 O31 O31
03b 03b 03b 03b
P5 P5 P5 P5 P5
P7
P16
Y4t
Y4b Y4b Y4b Y4b Y4b Y4b Y4b Y4b
Z4 Z4 Z4 Z4
AAl AAI M1 M1 M1 AAl AAI M1 M1
4 A14
818 818
C5
12
19b 19b
M5 M5
N17
N181 N18t N181
N18b N18b
016
X17 X 17
Y3b Y3b
' Y5 Y5 Y5 Y5
5 F51 F51 F51
G4 G4 G4
P5 P5 P5 P5 P5 P5 P5
Q11 Q11 Q11 Q11 Q11 Q11 Q11
Y4b Y4b Y4b Y4b Y4b

AC2
'data combined over years and locations

61

Table A13. Markers signiﬁcantly associated (P<0.05) with hydration coeﬁcient in three populations of
navy beans.

 

 

 

 

 

 

Combination of data used in analyses
Yearl Locerl Loc2le Year2 Locer2 L002Y12 Cmb' Locl Loc2
Pop Signiﬁcant markers
2 C5 C5
F51 F51
G4
H181
H18b H18b H18b H18b
M10
N15
N17 N17 N17 N17 N17
031
03b
Q3 Q3 Q3
Y4b Y4b
AB6 AB6 AB6 AB6
AC8 AC8 AC8 AC8
A017
4 A14 A14 A14 A14 A14 A14 A14 A14
818 , B18 818 818 818 B18 818 818
C5 C5 C5 C5 C5 C5
H12t H121
1181 1181 1181
J91 J91 J91 J91
19b 19b
MZ MZ MZ MZ MZ 112 M2 MZ
M5 M5 M5 M5 M5 M5 M5
M10 M10 M10 M10 M10
N17 N17 N17 N17 N17 N17
N181 N18t N181 N181 N181 N18t
N18m N18m N18m N18m N18m N18m
016
X3 X3 X3 X3 X3
X17
Y5 Y5 Y5 Y5 Y5 Y5
Y13 Y13 Y13 Y13 Y13 Y13 Y13
5 B15 B15 BIS BlS BIS B15 815
F9 F9 F9
J9 J9
L8 L8 L8 L8 L8 L8
M11 M11 M11 M11
N9 N9 N9 N9 N9 N9 N9 N9
016 016 016 016 016 016
P17
X3 X3
Y13 Y13 Y13 Y13
Y14 Y14 Y14 Y14 Y14 Y14 Y14
AA19
AC8 AC8 AC8 AC8

 

'data combined over years and locations

62

Table A 14. Markers signiﬁcantly associated (P<0.05) with percent solids lost in three
populations of navy beans.

 

 

 

 

 

 

Combination of data used in analyses
Yearl Locerl Loc2le Year2 Locer2 Loc2Yr2 Cmb' Locl Loc2
Pop Signiﬁcant markers
2 F51 F51
F5b
M10 M10
031 031
03b 03b
95 P5 95 p5
Q3 Q3
Y4t
Y4b Y4b Y4b Y4b Y4b
Z4- z4 Z4 Z4
M1 M1 AA!
4 A14 A14
818 318
cs C5 C5 C5 C5 C5
H121 H12t H121 H121 r1121 H121
H12b
H181
M2 M2 M2 M2 M2 M2 M2 M2
M10 M10 M10 M10
N17 N17 N17 N17 N17
N18t N18t N18t N18t N18t
P16 P16 P16
Q11 Q11 Q11 Q11 Q11 Q11
X3 X3
X7
Y3t
Y5 Y5 Y5 Y5 Y5
Y13
AC2
5 L8
M11 M11
N18 N18 N18 N18
P5 as as
Y13 Y13
Y14
AC2 AC2 AC2 AC2

 

Tdata combined over years and locations

63

Table A 15. Markers selected for multiple regession analysis.

 

 

 

 

Population/Trait Markers selected
Population 2
Visual score A18
Texture F51, G4, Y4b
Hydration coeﬁcient N17, AC8
Washed drained weight M10, 031, 03b, P5, Y4b, AAl
Percent solids lost PSLY4b, Z4
Population 4
Visual score A14, BIB, 12, M11, N18b, Y3t
Texture N18b, X7, Y3b
Hydration coemcient A14, B18, C5, M2, M5, M10, N17, N18t,
N18m, X3, Y5, Y13
Washed drained weight None
Percent solids lost C5, H121, M2, M10, N17, N18t, Q11, Y5
Population 5
Visual score L8, M11, Q11, X3
Texture F51, G4, L18, Q11, Y14
Hydration coeﬁcient B15, L8, M11, N9 016, Y14
Washed drained weight P5, Q11, Y4b

Percent solids lost

N18, AC2

Table 16. Chart showing marker genotypes and their relationship to canning
quality traits in population 2.

 

Avg. with band Avg. with band

 

 

 

 

 

 

 

 

Trait” Marker absentt present'
Visual score A18 3.38 3,20
Texture F51 _6§_4;_ 587
G4 589 6_6_4;
Y4b 2:5 588
Hydration coefficient N17 __98 1.92
Washed drained weight M10 299 3 294.75
031 m 295.73
03b 295.73 23,59
P5 219,12 295.77
Y4b 294.44 299.53
AAl 300,25 295.42
Solids lost P5 10,; 11.1
Y4b 11.3 , g);
Z4 11 1 10_.5

 

*undcrlining indicates the more desirable value for the trait
”visual score, 1-7; texture, Newtons/100g, hydration coefﬁcient, ratio; washed drained
weight, g; solids lost, %.

65

Table 17. Chart showing marker genotypes and their relationship to
canning quality traits in population 4.

 

Avg. with band Avg. with band

 

 

 

 

 

Trait" Marker absent’ present'
Visual score A14 2,52 3.11
818 2,32 3.09
12 2,22 3.06
M11 2,21 3.12
N18b M 3.06
Y3t 235; 2.58
Texture N18b 221 560
X7 566 612
Y3b 570 @
Hydration coeﬂicient A14 1.86 1.82
318 1.86 1,22
C5 1,82 1.86
M2 1.86 ;&
M5 m 1.87
M10 1a 1.87
N17 LS2 1.86
N18t 1,22 1.85
N18m 1,22 1.87
X3 1 87 Lg
Y5 1 85 2,22
Y13 l 87 w
Solids lost C5 7.8 22
H121 7.8 32;
M2 6._7 7.9
M10 8.0 1,;
N17 7.8 22
N18t 7.8 ﬂ
Q11 12 7.9
Y5 ' 6,2 7.8

 

'underlining indicates the more desirable value for the trai—t
ﬂvisual score, 1-7; texture, Newtons/100g, hydration coefﬁcient, ratio; washed drained
weight, g, solids lost, %.

66

Table A 18. Chart showing marker genotypes and their relationships to
canning quality traits in population 5.

 

Avg. with band Avg. with band

 

 

 

 

 

 

 

 

Trait" Marker absent‘ precent'
Visual score LS 224; 2.57
M11 M 2.54
Q11 2.61 2,19
X3 2g 2.63
Texture F5 J_(_)_8_ 616
G4 616 M
L18 640 73_8
Q11 628 Z4_8
Y14 l_4__5_ 642
Hydration coefﬁcient BIS M 1.92
L8 2% 1.92
M11 _lh9_5_ 1.92
N9 1.92 1,26
016 1.93 _1_,2§
Y14 1:11 1 92
Washed drained weight P5 297,75 291.54
Q11 296,23 289 94
Y4 292.06 296,39
Solids lost N18 12.4 11,2
AC2 12.2 1 5

 

funderlining indicates the more desirable value for the trait
"visual score, 1-7; texture, Newtons/100g; hydration coefﬁcient, ratio; washed
drained weight, g; solids lost, %.

67

Table A 19. Proposed linkage goups for population two.

 

Distance between Total length of

 

1M e group Markers , markers’ W
l F5t 3.8 CM 3.8 CM
G4
2 F10 8.7 CM 8.7 CM
12
3 M10 6.1 CM 14.1 CM
Z4 6.1 CM
O3t 0.0 CM
03b 1.8 CM
M1
4 A18 6.1 CM 12.3 cM
Q11 6.1 cM
L18
5 CS 6.1 CM 6.2 CM
N17 0.00 CM
AC8 0.00 cM
AB6

 

'cM =- centimorgans

68

Table A20. Proposed linkage goups for population four.

 

Distance between Total length of linkage

 

_I_.ii1k_age_goup Marker markers’ group’
1 N18m 4.2 cM 8.5 cM
N181 0.0 CM
CS 4.2 cM
Y5
M2 7.7 CM 19.6 CM
H12t 11.9 cM
Q11
AC2 0.0 cM 14.4 CM
H12b 14.4 CM
M11
118 11.9 CM 27.3 CM
A14 7.7 cM
B18 7.7 cM
X17
19b 14.4 cM 14.4 CM
' Y3
016 7.7 cM 17.3 cM
M10 2.7 cM
X3 2.7 cM
Y13 4.2 cM
M5
AAl 1.3 cM 1 .3 cM
031 0.0 cM
03b
AA19 5.9 CM 20.3 CM
Q3 14.4 CM
N2

 

'cM = centimorgans

69

Table A21. Proposed linkage goups for population ﬁve.

 

 

Distance between Total length of linkage
_Li£k_age_goup Marker markers’ goup'
1 L8 6.4 CM 22.7 CM
N9 4.0 CM
BIS 4.0 CM
Y13 1.9 CM
X3 6.4 CM
016
2 G4 0.0 CM 0.0 CM
F5t
3 L18 6.4 CM 6.4 CM
Q1 1
4 M2 6.4 CM 6.4 CM
' AC2
5 Gl9t 4.0 CM 4.0 CM
M19
6 AB6 6.4 CM 6.4 CM
AC8

 

'cM '- centimorgans

LIST OF REFERENCES

Abbo, S., G. Ladizinsky and N. Weeden. 1992. Genetic analysis and linkage study of
‘ seed weight in lentil. Euphytica 58:259-266.

Agbo, G., G. Hosﬁeld, M. Uebersax, K. Klomparens. 1987. Seed microstructure and its
relationship to water uptake in isogenic lines and a cultivar of dry beans
(Phaseon vulgaris L.). Food Microstructure 6:91-102.

Adams, M.W. and CL. Bedford. 1975. Breeding food legumes for improved processing
and consumer acceptance properties. p. 299-304. In M. Milner (ed) Nutritional
Improvement of Food Legumes by Breeding. John Wiley and Sons, NY.

Allard, RW. 1960. Principles of plant breeding. John Wiley & Sons, NY.

Barua, U.M., K.J. Chalmers, C.A. Hackett, W.T. Thomas, W. Powell and R Waugh.
1993. Identiﬁcation of RAPD markers linked to a Rynchosporium secalis
resistance locus in barley using near-isogenic lines and bulked segegant analysis.
Heredity 71:177-184.

Beavis, W., O. Smith, D. Grant and R Fincher. 1994. Identiﬁcation of quantitative trait
loci using a small sample of topcrossed and F4 progeny from maize. Crop Sci.
34:882-896.

Brothers, ME. and J .D. Kelly. .1993. Interrelationship of plant architecture and yield
components in the pinto bean ideotype. Crop Sci. 33: 1234-1238.

Bubeck, D., M. Goodman, W. Beavis and D. Grant. 1993. Quantitative trait loci
controlling resistance to gay leaf spot in maize. Crop Sci. 33:838-847.

Burr, H.K., S. K011 and HI. Morris. 1968. Cooking rates of dry beans as inﬂuenced by
moisture content, temperature and time of storage. J. Food Sci. and Tech.
22:336.

Chalmers, K., U. Barua, C. Hackett, W. Thomas, R. Waugh and W. Powell. 1993.
Identiﬁcation of RAPD markers linked to genetic factors controlling the milling
energy requirement of barley. Theor. Appl. Genet. 87:314-320.

Chase, C.D., V.M. Ortega and CE. Vallejos. 1989. DNA restriction fragment length
polymorphisms correlate with isozyme diversity in Phaseolus vulgaris L. Theor.
Appl. Genet. 81:806-811.

Deshpande, S. and M. Cheryan. 1986. Microstructure and water uptake of Phaseolus
and winged beans. J. Food Sci. 51:1218-1223.
70

71

Deshpande, S., S. Sathe and D. Salunkhe. 1984. Interrelationships between certain
physical and chemical properties of dry beans (Phaseolus vulgaris L). Qual. Plant.
Pl. Foods Hum. Nutr. 34:53.

Dirlewanger, E., P. Isaac, S. Ranade, M. Delajouza, R Cousin and D. Vienne. 1994.
Restriction fragment length polymorphism analysis of loci associated with disease
resistance genes and developmental traits in Pisum sativum L. Theor. Appl.
Genet. 88:17-27.

Dudley, J .W. 1993. Molecular markers in plant improvement: manipulation of genes
affecting quantitative traits. Crop Sci. 33:660-668.

Edwards, M.D., C.W. Stuber and JP. Wendel. 1987. Molecular-mark -facilitated
investigations of quantitative trait loci in maize. 1. Numbers, genomic distribution,
and types of gene action. Genetics 116:113-125.

Everson, E.H. and CW. Schaller. 1955. The genetics of yield diﬁ'erences associated with
awn barbing in the barley hybrid (Lion x Atlas”) x Atlas. Agon. J. 47:276-280.

Fang, G., S. Hammar, and R Grumet. 1992. A quick and inexpensive method for
removing polysaccharides ﬁ'om plant genomic DNA. BioFeedback 13:52-55.

Foolad, M.R., RA. Jones and R.L. Rodriguez. 1993. RAPD markers for constructing
intraspeciﬁc tomato genetic maps. Plant Cell Reports 12:293-297.

Garruti, R and M. Bourne. 1985. Effect of storage conditions of dry bean seeds
(Phaseolus vulgaris) on texture proﬁle parameters after cooking. J. Food Sci.
50: 1067.

Ghaderi, A, G.L. Hosﬁeld, M.W. Adams and MA Uebersax. 1984. Variability in
culinary quality, component interrelationships, and breeding implications in navy
and pinto beans. J. Amer. Soc. Hort. Sci. 109:85-90.

Guo, M., D.A Lightfoot, M.C. Mok and D.W.S. Mok. 1991. Analyses of Phaseolus
vulgaris L. and P. caccineus Lam. hyrids by RFLP: preferential transmission of P.
vulgaris alleles. Theor. Appl. Genet. 81:703-709.

Haley, S. D. P ..N Miklas, J. R Stavely, J. Byrum, and J. D. Kelly. 1993. Identiﬁcation of
RAPD markers linked to a major rust resistance gene block 1n common bean.
Theor. Appl. Genet. 86:505-512.

Haley, S.D., P.N. Miklas, L. Afanador and J. D. Kelly. 1994a. Random Ampliﬁed
Polymorphic DNA (RAPD) marker variability between and within gene pools of
common bean. J. Amer. Soc. Hort. Sci. 119:122-125.

72

Haley, S. D., L. Afanador and JD. Kelly. 1994b. Identiﬁcation and application of a
random ampliﬁed polymorphic DNA marker for the I gene (Potyvirus resistance)
in common bean. Phytopathology 84:157-160.

Heil, JR and M.J.McCalthy. 1992. Processing characteristics of kidney beans. Kidney
Bean Canning Quality Research Report. Dept. of Food Science and Technology,
University of California Davis, CA

Hosﬁeld, George L. and M.A Uebersax, 1980. Variability in physico-chernical properties
and nutritional components of tropical and domestic dry bean germplasm. J.
Amer. Soc. Hort. Sci. 105:246-252.

Hosﬁeld, George L. M.A Uebersax, and TG. Isleib. 1984. Seasonal and genotypic
eﬂ‘ects on yield and physico-chemical seed characteristics related to food quality in
dry, edible beans. J. Amer. Soc. Hort. Sci. 109:182-189.

Hosﬁeld, G.L. 1991. Genetic control of production and food quality factors in dry bean.
Food Technology 45:98-103.

Johnson, W. and P. Gepts. 1995. The association of size differences with seed storage
protein class in common beans (Phaseolus vulgaris): Sax’s hypothesis revisited.
Plant Genome II abst. p 50.

Jung, G., D. Coyne, P. Skroch, J. Nienhuis, E. Amaud-Santana, J. Bokosi, S. Kaeppler
and J. Steadman. 1994. Construction of a genetic linkage map and locations of
common blight, rust resistance and pubescence loci in Phaseon vulgaris L. using
random ampliﬁed polymorphic DNA (RAPD) markers. Ann. Rep. Bean
Improvement Cooperative, vol. 37.

Kelly, RD., G.L. Hosﬁeld, G.V. Vamer, M.A. Uebersax, M.E. Brothers and J. Taylor.
1994. Registration of ‘Huron’ navy bean. Crop Sci. 34:1408.

Lande, R 1992. Marker-assisted selection in relation to traditional methods of plant
breeding. p. 437-451. In, H.T. Stalker and JP. Murphy (eds. )Plant breeding in
the 1990s. CAB. International, UK.

Lande, R and R Thompson. 1990. Efﬁciency of marker-assisted selection in the
improvement of quantitative traits. Genetics 1241743-756.

Lander, ES. and D. Botstein. 1989. Mapping mendelian factors underlying quantitative
traits using RFLP linkage maps. Genetics 121 : 185-199.

Leakey, C.L.A 1988. Genotypic and phenotypic markers in common bean. p. 245-327.
In P. Gepts (ed) Genetic resources of Phaseolus beans. Kluwer Academic
Publishers, Boston.

73

Martin, G., J. Williams, and S. Tanksley. 1991. Rapid identiﬁcation of markers linked to
a Psuedomonas resistance gene in tomato by using random primers and near-
isogenic lines. Proc. Natl. Acad. Sci. 88:2336-2340.

Martin, B., J. Nienhuis, G. King, and A Schaefer. 1989. Restriction ﬁagnent length
polymorphisms associated with water use eﬁciency in tomato. Science
243 : 1725-1728.

Michelmore, RW., I. Paran, and RV. Kesseli. 1991. Identiﬁcation of markers linked to
disease resistance genes by bulked segegant analysis: a rapid method to detect
markers in speciﬁc genomic regions using segegating populations. Proc. Natl.
Acad. Sci. 88:9829-9832.

Miklas, P.N., J.R Stavely and JD. Kelly. 1993. Identiﬁcation and potential use of a
molecular marker for rust resistance in common bean. Theor. Appl. Genet.
85:745-749..

Mohan, M.S. Nair, J. Bentur, U. Rao and J. Bennett. 1994. RFLP and RAPD mapping of
the rice Gm2 gene that confers resistance to biotype 1 of gall midge (Orseolia
oryzae). Theor. Appl. Genet. 87:782-788.

Nienhuis, J.,' T. Helentjaris, M. Slocum, B. Ruggero and A. Schaefer. 1987. Restriction
ﬁagnent length polymorphism analysis of loci associated with insect resistance in
tomato. Crop Sci. 27:797-803.

Nodari, R.O., E.M.K. Koinange, J.D. Kelly and P. Gepts. 1992. Towards an integated
linkage map of common bean. 1. Development of genomic DNA probes and levels
of restriction ﬁ'agnent length polymorphism. Theor. Appl. Genet. 84: 186-192.

Osborn, T.C., D.C. Alexander and J .F. Fobes. 1987 . Identiﬁcation of restriction fragment
length polymorphisms linked to genes controlling soluble solids content in tomato
ﬁ'uit. Theor. Appl. Genet. 73:350-356.

Patterson, AH, S. Damon, J.D. Hewitt, D. Zamir, H.D. Rabinowitch, S.E. Lincoln, E.S.
Lander and SD. Tanksley. 1991. Mendelian factors underlying quantitative traits
in tomato: comparison across species, generations and environments. Genetics
127:181-197.

Penner, G., J. Chong, C. \Vlght, S. Molnar and G. Fedak. 1993. Identiﬁcation of a
RAPD marker linked to the oat stem rust gene Pg3. Theor. Appl. Genet. 85:702-
.705.

Powrie, W.D., M.W. Adams and U. Pﬂug. 1960. Chemical, Anatomical, and
Histochemical Studies on the Navy Bean Seed. Agonomy Journal 52:163-167.

Rasmusson, J.M. 1935. Studies on the inheritance of quantitative characters in Pisurn: 1.
Preliminary note on the genetics of ﬂowering. Hereditas 20:161-180.

74

Ruengsakulrach, S., N. Srisuma, M. Uebersax, G. Hosﬁeld, and L. Occetla. 1994. Early
generation screening of navy bean breeding lines by canning quality assessment and
pasting characteristics of bean ﬂour. J. Food Quality 17 :321-333.

Saghai-Maroof, M.A, K.M Soliman, RA. Jorgensen, and RW. Allard. 1984. Ribosomal
DNA spacer-length polymorphisms in barley: Mendelian inheritance,
chromosomal location, and population dynamics. Proc. Natl. Acad. Sci. 8128014-
8018.

Sax, K. 1923. The association of size difference seed-coat pattern and pigmentation in
Phaseolus vulgaris. Genetics 8:552-560.

Snyder, EB. 1936. Some factors affecting the cooking quality of the pea and geat
northern types of dry beans. Nebraska Ag. Exp. Sta. Res. Bul. 85.

Soller, M. and T. Brody. 1976. On the power of experimental desigrs for the detection
of linkage between marker loci and quantitative loci in crosses between inbred
lines.

Srisuma, N., R Harnmerschmidt, M. Uebersax, S. Ruengsakulrach, M. Bennink and G.
Hosﬁeld. 1989. Storage induced changes of phenolic acids and the development
of hard-to-cook in dry beans (Phaseolus vulgaris, var. Seafarer). J. Food Sci.
54:311-314.

Stuber, CW. 1992. Biochemical and molecular markers in plant breeding. p. 37-61. In J.
Janick (ed) Plant breeding reviews. John Wiley & Sons, Inc., NY.

Stuber, C.W., S.E. Lincoln, D.W. Wolff, T. Helentjaris, and ES. Lander. 1992.
Identiﬁcation of genetic factors contributing to heterosis in a hybrid from two elite
maize inbred lines using molecular markers. Genetics 132:823-839.

Stuber, C.W., RH. Moll, M.M. Goodman, HE. Schaffer, and BS. Weir. 1980.
Allozyme ﬁ'equency changes associated with selection for increased gain yield in
maize (Zea mays L.). Genetics 95:225-236.

Stuber, C.W., M.M. Goodman and RH. M011. 1982. Improvement of yield and car
number resulting from selection at allozyrne loci in a maize population. Crop Sci.
22:737-740.

Stuber, C.W., M. Edwards, and IF. Wendel. 1987. Molecular-markcr-facilitated
investigations in maize. 11. Factors inﬂuencing yield and its component traits. Crop
Sci. 27:639-648.

Tanksley, S.D., H. Medina-Filho and CM. Rick. 1982. Use of naturally-occurring
enzyme variation to detect and map genes controlling quantitative traits in an
interspeciﬁc backcross of tomato. Heredity 49:11-25.

75

Uebersax, M.A and CL. Bedford. 1980. Navy bean processing: eﬂ‘ect of storage and
soaking methods on quality of canned beans. Mich. State Univ. Ag. Expt. Sta, E.
Lansing. Res. Rpt 410.

Uebersax, M. and S. Ruengsakulrach. 1989. Structural and compositional changes during
processing of dry beans (Phaseolus vulgaris). In J. Jen (ed.) Quality Factors of
Fruits and Vegetables. American Chemical Society, Washington, DC.

Vallejos, CE. and CD. Chase. 1991. Linkage between isozyme markers and a locus
affecting seed size in Phaseolus vulgaris L. Theor. Appl. Genet. 81:413-419.

Vallejos, CE. and SD. Tanksley. 1983. Segegation of isozyme markers and cold
tolerance in an interspeciﬁc backcross of tomato. Theor. Appl. Genet. 61241-247 .

Wassimi, N.N., G.L. Hosﬁeld and M.A. Uebersax. 1990. Inheritance of physico-chemical
seed characters related to culinary quality in dry bean. J. Amer. Soc. Hort. Sci.
1 15:492-499.

Weeden, N., G. Tirnmerrnan and J. Lu. 1994. Identifying and mapping genes of
economic sigliﬁcance. Euphytica 73:191-198.

N1v. LIBRQRIES

~111g111111111311111111 1111111111111111111111

024016