Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tischner, T.
Right arrow Articles by Lark, K. G.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Tischner, T.
Right arrow Articles by Lark, K. G.
Agricola
Right arrow Articles by Tischner, T.
Right arrow Articles by Lark, K. G.
Related Collections
Right arrow Cell Biology & Molecular Genetics
Right arrow Crop Genetics
Right arrow Crop Growth and Development
Right arrow Soybean
Crop Science 43:464-473 (2003)
© 2003 Crop Science Society of America

CROP BREEDING, GENETICS & CYTOLOGY

Genetics of Seed Abortion and Reproductive Traits in Soybean

T. Tischnera, L. Allphina, K. Chaseb, J. H. Orfc and K. G. Lark*,b

a Dep. of Botany and Range Sci., Brigham Young Univ., Provo, UT 84602
b Dep. of Biology, Univ. of Utah, Salt Lake City, UT 84112
c Dep. of Agronomy and Plant Genetics, Univ. of Minnesota, St. Paul, MN 55108

* Corresponding author (lark{at}bioscience.utah.edu)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
In soybean [Glycine max (L.) Merr.], yield or seed number is determined by the number of seeds per pod, the number of pods per plant, and the number of plants per row. Seeds per pod is a function of the number of ovules per pod and the success of seed set (lack of seed abortion). The objective of this study was to define the genetic basis for seed set. In this study, recombinant inbred (RI) segregants of soybean have been used to identify genetic components, quantitative trait loci (QTLs) that regulate different parameters of seed set. The numbers of ovules per pod and abortions were measured in RI segregants of soybean, as were positions and developmental stages of abortions within pods. Quantitative trait loci were identified from segregants of the Minsoy-Noir 1 RI population grown in Utah and Minnesota. These were confirmed in Minsoy-Archer or Noir 1-Archer RI populations grown in Minnesota. The average abortion frequency in the three RI populations ranged from 10 to 29%. Abortions occurred most frequently in the basal position of the pod and embryo development ceased most often at one of two developmental stages: embryos failed to develop (or ovules were not fertilized), or embryos were partially developed and ceased growth just after cotyledon differentiation. Quantitative trait loci were found on several linkage groups, including U11, U13, and U22. The QTLs in U11 were linked to QTLs for flowering date, reproductive period (RP), and maturity. Two separate QTLs on U13 were linked to genes for male and female sterility (Ms1, Ms6, or St5) and to genes for disease resistance, respectively. One of two distinct QTLs on U22 was associated with a previously identified QTL for water use efficiency. Also of interest was a QTL on U3 linked to Lf1 that determined the number of ovules per pod, suggesting a common regulation of leaf and flower primordia.

Abbreviations: cM, centiMorgan • MA, Minsoy-Archer population • MN, Minsoy-Noir 1 population • NA, Noir 1-Archer population • QTLs, quantitative trait loci • RI, recombinant inbred • RP, reproductive period


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
SEED SET is determined by the number of ovules per fruit or pod, the frequency of embryo abortions, and the number of fruits (pods) per plant. A large number of plant species produce many more ovules than seeds (Wiens, 1984; Lee, 1988) exhibiting high percentages (50–100%) of embryo abortions (Wiens, 1984; Nakamura, 1986; Wiens et al., 1987; Charlesworth, 1989; Wiens et al., 1989; Allphin and Harper, 1997). Several studies have indicated a genetic basis for abortions of developing seeds (Casper and Wiens, 1981; Wiens et al., 1987; Charlesworth, 1989; Allphin et al., 2002). Low fecundity in some species has been shown to be due to genetically programmed, early embryo abortions (Casper and Wiens, 1981; Wiens et al., 1987) such as low seed set in barley (Hordeum vulgaris L.), which is governed by a single codominant gene (Devaux et al., 1992). Moreover, entire plant lineages are known to exhibit low fecundity such as the extremely low seed set that characterizes the milkweeds (Euphorbia spp.; Wyatt and Broyles, 1994). However, more detailed information is needed on the role of genetic loci on low fecundity in plants.

Fruit and seed development have been especially well-studied in soybean (Kato, 1964; Shibles et al., 1975). Soybean ovaries contain anywhere from one to four ovules and the average number per pod is genetically determined (Shibles et al., 1975). Variation in seed yield due to abortion of developing ovules has been established (Halsted, 1917; Kato and Sakaguchi, 1954; Kato, 1964; Shibles et al., 1975; Palmer and Heer, 1984; Stelly and Palmer, 1985) and most commonly occurs in the basal position (Halsted, 1917; Shibles et al., 1975; Palmer and Heer, 1984). This is similar to other multiovuled fruits (including legumes) in which the probability of abortion can be dependent on ovule position within the developing fruit (Halsted, 1917; Kato and Sakaguchi, 1954; Stephenson, 1981; Palmer and Heer, 1984; Lee, 1988; Rocha and Stephenson, 1990).

Ovule number and embryo abortion are examples of quantitative traits resulting from the interplay between the plant genotype and the environment in which it is grown (Barton and Turelli, 1989; Tanksley, 1993). In general, quantitative phenotypes are polygenic, controlled by a large number of genes (QTLs), many of which may have a small effect on the phenotype individually, but which have a large effect in the aggregate.

It is only recently that techniques have become available to identify QTLs (Lander and Botstein, 1989; Dudley, 1993; Falconer and Mackay, 1996). In principle, this is done by linking phenotypic variation to the segregation of closely linked qualitative genetic markers, usually molecular markers involving DNA sequence polymorphisms (Lander and Botstein, 1989). For such analyses, large RI segregant populations are especially useful.

Recombinant inbreds are powerful genetic tools. Recombinant inbred segregants are produced by crossing genetically distant parents and subsequently inbreeding to homozygosity (e.g., Morse, 1978; Burr et al., 1988; Burr and Burr, 1991; Mansur and Orf, 1995). When created from naturally inbreeding plants, this process disrupts the architecture of the genome that had evolved in the course of inbreeding and selection and produces new genotypes that can have radically different phenotypes. The resulting genetic admixtures can exhibit transgressive variation in which some progeny phenotypes are much more extreme than those of the parents from which they arose (e.g., Mansur et al., 1993; Mansur et al., 1996; Orf et al., 1999). Large numbers of homozygous RI segregants can be prepared and stored as seed and because they are genetically reproducible can be used in different experiments in different locations.

For this study, we have used soybean to identify QTLs that regulate the number of ovules in a pod as well as parameters of embryo abortion. In the experiments to be described, we use large RI populations of soybean in which each segregant has been characterized by >600 molecular genetic markers distributed across the genome ({approx}3000 cM comprising 20 chromosomes; Cregan et al., 1999) to identify QTLs that regulate seed set. Specifically, we identify QTLs associated with the number of ovules in a pod, the frequency as well as the pod-position of embryo abortions, and the developmental stage of abortions.

In previous studies, this RI population has been used to identify QTLs controlling yield, seed number, and seed size, as well as loci for flowering date (R1), maturity (R8), and RP (R8-R1) (Orf et al., 1999). Thus, genetic and phenotypic information for ovule number and abortion can be related to other agronomic traits.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Materials for analysis were obtained from RI soybean populations developed by Mansur, containing {approx}240 F9–derived RI lines (Mansur and Orf, 1995; Orf et al., 1999). These RI populations were developed through single seed descent from segregants derived from intraspecific reciprocal crosses of three cultivars (PI 290136, Noir 1; PI 27890, Minsoy; PI 546487, Archer). The three parental lines, and the elite cultivars Clay, Kato, Leslie, McCall, and Sturdy, were also included in the experiments as standards (checks). The RI, parental lines, and checks were planted at two locations: Utah in 1997, and Minnesota in 1998.

The Minsoy-Noir 1 (MN) RI population, as well as Minsoy, Noir 1, and the check cultivars, were planted in Utah at Brigham Young University's Agricultural Station at Spanish Fork, UT (40° N, 111° W). The plants were sown in hill plots rather than rows. Seeds were placed in rectangular spacing (75 cm by 90 cm apart). Each of the lines was replicated three times. Each replicate was randomly assigned to hill placement in the field. To reduce the effects of interplant competition, all hills were thinned to three seedlings. Plants were grown under irrigated conditions of optimal moisture. On maturity, the three plants at each hill were harvested by hand and placed in paper bags. Fifty randomly chosen pods from each hill were analyzed for seed set traits.

Cultivars were also planted in Minnesota under typical agronomic yield trial conditions for soybean at Rosemount (45° N, 93° W). Three RI populations were grown: MN, MA (Minsoy-Archer) and NA (Noir 1-Archer). Parental and check cultivars were included. The field consisted of a randomized complete block design with two replications of each RI line. The plots were 2 rows, 1.5 m long, with 75-cm spacing between rows. The plants were rain-fed. Three plants in each plot were randomly selected from each 50-plant row, and harvested by hand on maturity. The cultivars were packaged in paper sacs, and taken to Brigham Young University in Provo, UT, for analysis. All of the pods from each plant were analyzed.

Pods collected from both sites were analyzed for the number of ovules per pod, the number of filled seeds per pod, the seed:ovule ratio, and the developmental stage and the position of abortions within each pod. Pods contained anywhere from 1 to 4 ovules. The basal position was characterized as Position 1, with Positions 2, 3, and 4 moving to the apex of the pod. Because large numbers of pods were required for genetic analysis, a simplified classification of developmental stages was adopted, suitable for rapid analysis by visual inspection. We describe four general stages of embryo abortion (illustrated in Fig. 1) based on the classification of Kato (1964). The earliest stage (Fig. 1a) is characterized by a basal visible ovule or proembryo, where either the ovule is not fertilized, or cell division has terminated at a very early stage of embryo development (3–8 cells; Kato, 1964). Shibles et al. (1975) estimate that embryo abortion in this initial stage of proembryo development occurs within 1 to 7 d of fertilization. The second stage of embryo abortion (Fig. 1b) is a later stage of proembryo development characterized by a more advanced growth of the proembryo (Kato, 1964). Embryos aborted at this stage of development consist of {approx}100 cells (Kato, 1964), and are found {approx}7 to 15 d following fertilization (Shibles et al., 1975). A third stage of embryo abortion (Fig. 1c) occurs just after cotyledon differentiation, and is characterized by partial growth of the embryo (Kato, 1964). On average, the embryo has been developing for 15 to 26 d before aborting, and the cotyledons fail to fill. Finally (Fig. 1d), a seed may contain 80% of its dry weight, protein, and fat, and be characterized by a wrinkled appearance and likely represents an abortion at a late stage. However, viability of these seeds was not determined.



View larger version (64K):
[in this window]
[in a new window]
  		Fig. 1. Examples of the four developmental stages of soybean ovule abortion scored (i.e., stages of development at which growth ceased). The four general stages of embryo abortion are based on the classification of Kato (1964). The first, (none) characterized by a basal visible ovule or pro-embryo; the second (early) is a later stage of development characterized by a more advanced growth of the pro-embryo; the third (partial) occurs just after cotyledon differentiation; the fourth (late) may contain 80% of seed dry weight, protein, and fat, and is characterized by a wrinkled appearance.

 
These reproductive traits were compared with genetic markers using existing genetic linkage maps of the RI genotypes from this inbred population (Cregan et al., 1999). Genetic markers comprised restriction fragment length polymorphisms and microsatellites (Lark et al., 1994; Mansur et al., 1996; Orf et al., 1999). Quantitative trait loci contributing to the reproductive trait variation were identified and analyzed for interactions using the Epistat computer program (Lark et al., 1995; Chase et al., 1997). This computer software uses maximum likelihood methods for both the identification and evaluation of significance of interactions between pairs of QTLs (Chase et al., 1997). The simple interval mapping feature of the computer package PLABQTL (Utz, 1996) also was used for detecting QTLs. This program uses a multiple regression approach to interval mapping with marker order and distances determined by Mapmaker (Lander et al., 1987). We established empirical log likelihood (LOD) thresholds for QTL detection using permutation tests (Churchill and Doerge, 1994). We used analyses of variance to partition the total variance into genetic and environmental components. Heritability estimates (Hanson et al., 1956) were computed as: H2 = sG/(sG + se/r), where H2 = heritability, sG = genotypic variance, se = error, and r = the number of reps for the trait.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
Phenotypes of Inbred Genotypes
Parameters of seed set were measured in two quite different environments: In Utah, as irrigated, spaced hill plots and in Minnesota as normal field conditions, high density, rain-fed, row plots. Table 1 compares data from the two field trials. Whereas the number of ovules per pod may be somewhat higher in Utah, the number of seeds per pod was much higher in Utah (Table 1, i). This could be attributed to a much lower rate of abortion in Utah (higher number of seeds per ovule) where growth was less competitive and water and light more available to individual plants (see Egli, 1999). Heritability values were higher for the number of ovules per pod (H2 = 0.84) than for seed set (H2 = 0.57) or abortion (H2 = 0.48). On average, most of the RI segregants set less seed (seeds per pod) than elite cultivars (checks, used as comparative standards). This probably reflects a genetic contribution from the Minsoy parent.


View this table:
[in this window]
[in a new window]
  		Table 1. Comparison of soybean seed set and abortion parameters in Utah and Minnesota: (i) The means and standard errors for Minsoy, Noir 1, the recombinant inbred population, and the group of five elite cultivars (checks). Percentage abortion = 100% seeds per ovule. (ii) The fraction of ovules that aborted in the position indicated (e.g., in Utah, 24% of ovules in Position 1 aborted). (iii) The percentage of ovules aborting at each of the four stages in Fig. 1 [e.g., for Minsoy grown in Utah: abortions = 17.3% (or 9.21 + 1.47 + 6.45 + 0.18) = 100 - 82.7% (seeds per ovule)].

 
More detailed data for the RI segregant population is shown in Fig. 2. Values from Minnesota are compared with those from Utah. As might be expected, the average values of reproductive parameters for the RI segregant population were between the two parents, but closer to Minsoy than to Noir 1. The range of values, however, often exceeded both of the parents as well as the check lines (transgressive segregation). For all three parameters, the mean values in Minnesota were different from those in Utah, as were the slopes of the regressions. In general, the RI segregants had more abortions (less seeds per ovule) in Minnesota than in Utah (Fig. 2a) and, as a consequence, a lower seed set (seeds per pod). Moreover, in Minnesota, seeds per ovule decreased in segregants with higher numbers of ovules per pod (Fig. 2b), leading to less of an increase in seeds per pod as ovules per pod increased (Fig. 2c). The physical environment in Utah is not optimal for the growth of soybean. The high desert results in low humidity and cold nights. Nevertheless, the abortion frequency was much lower in the Utah environment and abortions did not increase significantly with increasing pod size. It seems most likely that the greater abortion rate in Minnesota is the result of competition for water within densely planted rows fed by rainfall or for light under the denser canopy found in such row plots (Egli, 1999).



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 2. Relationships between seed/ovule, seeds/pods or ovules/pod for soybean segregants within the Minsoy-Noir 1 recombinant inbred population. Values are shown for Utah (•), Minnesota ({circ}) and the parental cultivars.

 
As observed for other legumes (Halsted, 1917; Rocha and Stephenson, 1990), abortions in both environments occurred most frequently in the first position (Table 1, ii), becoming less frequent in distal positions. This was true in both environments and may result from differences in timing of fertilization (Kambal, 1969), leading to differences in vigor between the distal and basal embryos (Shibles et al., 1975; Casper, 1984, 1988).

In both Utah and Minnesota, abortions were most frequent in the development stages showing no embryos (H2 = 0.60) or partially developed embryos (H2 = 0.48); whereas embryos aborted in early or late stages were infrequent (Table 1, iii). The no development ovules could either be ovules that were undeveloped, that failed in fertilization, or that underwent very early abortions at the initial stages of proembryo development. Limitations of light microscopy made it difficult to distinguish among these possibilities. Our partially developed seeds most likely were aborted during the transition stage from proembryo to heart-stage, consistent with earlier findings that this stage most commonly aborts (Kato and Sakaguchi, 1954; Kato, 1964). Abortions leading to no observable embryos (none) were almost equally frequent in the two environments, whereas abortions of partially developed embryos were much more frequent in Minnesota, accounting for most of the difference in overall abortion frequency between the two environments. Finally, the lack of observable embryos was strikingly reduced in the terminal position (Table 2). Note that only the none category lacks abortions in single ovule pods. Unlike other stages of abortion, the frequency of the none category drops off in the terminal position of the pod. This effect is most striking in pods with four ovules. A similar effect was observed in the Minsoy-Archer and Noir 1-Archer RI populations discussed below (data not shown).


View this table:
[in this window]
[in a new window]
  		Table 2. Soybean abortion frequency in the Minsoy-Noir 1 population in Minnesota. Frequencies are categorized according to the stage and position of embryo development, and the number of ovules per pod. Embryo position is numbered from basal (1) to distal (4).

 
Most of the high yielding elite lines used as checks had few if any pods with four ovules, either in Utah or Minnesota (data not shown). On the other hand, several RI segregants had a significant number of pods with four ovules when grown either in Utah or Minnesota (H2 = 0.72). This phenotype was found in RI segregants previously shown by Orf et al. (1999) to produce higher yields in several environments (data not shown).

Quantitative Trait Loci Associated with Seed Set
Utilizing the DNA polymorphic markers that define the genotypes of the RI population, we have identified QTLs that regulate variation in a number of traits associated with seed set. Table 3 summarizes our results. For each phenotype we present the number of QTLs identified at a significance of LOD > 2.5 and the amount of phenotypic (R2) as well as heritable (R2/H2) variation that they can explain. Heritabilities could not be calculated for infrequent abortion types (Table 3). However, for almost all those parameters for which heritabilities were determined, QTLs accounted for {approx}50% of the heritable variation. The one exception was four ovules per pod. In Tables 4 and 5, we describe those QTLs with LOD > 3 in detail.


View this table:
[in this window]
[in a new window]
  		Table 3. Summary of soybean seed set and abortion quantitative trait loci (QTLs) identified using recombinant inbred segregants from the Minsoy-Noir 1 population. For each of the traits shown, the number of QTLs that were identified at a significance of LOD > 2.5 is given. The amount of phenotypic variation (R2) accounted for by these QTLs is presented, together with the heritabilities of the trait (H2) and the amount of heritable variation (R2/H2) explained by the QTLs.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Soybean quantitative trait loci (QTLs) (at LOD > 3.0) affecting seed set, identified in the Minsoy-Noir 1 recombinant inbred segregant population. Phenotypic values were from the combined Utah and Minnesota environments. The linkage group (LG), markers (MKRs), and support interval locate the QTL. The significance (LOD) as well as the phenotypic variation explained by the QTL (R2) are also given, together with the parent [Minsoy (M) or Noir 1 (N)] contributing the allele with the higher value.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Soybean quantitative trait loci (QTLs) affecting abortions, identified in the Minsoy-Noir 1 recombinant inbred segregant population. Phenotypic values were from the combined Utah and Minnesota environments. The linkage group (LG), markers (MKRs), and support interval locate the QTL. The significance (LOD) as well as the phenotypic variation explained by the QTL (R2) are also given, together with the parent [Minsoy (M) or Noir 1 (N)] contributing the allele with the higher value.

 
As seen in Table 4, pod size (ovules per pod), which is highly heritable (Table 3), is regulated by four QTLs of LOD > 3. Two loci on U13 influence pod size (ovules per pod) and consequently the number of seeds per pod. Of these, the major QTL, located on Linkage Group 13, can control as much as 19% of the ovule per pod variation. This QTL is located in a region known to contain several male sterile loci [e.g., Ms1, Ms 6, and St 5 (Palmer and Kilen, 1987; Cregan et al., 1999)], suggesting that alleles at these loci may influence the number of ovules in a pod. The second QTL on U13 (near R045_1) is linked to loci controlling disease resistance [e.g., Rpg1, Rps3, and Rpv1 (Cregan et al., 1999)]. This region of U13 also includes QTLs for RP and for seed weight (Orf et al., 1999). In the MN RI population, segregants with the Noir 1 haplotype of the RP QTL had shorter RPs. Increasing ovules per pod (Table 4) while decreasing RP could have an effect on seed weight. Other QTLs regulating ovules per pod include one on U14 near the determinate stem locus, Dt1, and one on U22. The U14 locus was found to interact (P <= 0.0001) with a locus on LG-U6 (Rpg4 or another locus closely linked to this disease resistance gene), such that the effect of the U14 QTL was conditional on the presence of a Minsoy allele at the U6 locus (data not shown). The locus on U22, that also affects the frequency of four ovule pods, occurs in the same region as another QTL for seed weight that segregates in the NA population (Orf et al., 1999). Effects of pod size on seed weight are not surprising, especially if resources become limiting during growth. Finally, we have listed a weak QTL (LOD < 3.0) in U3 that is supported by a similar, very strong QTL segregating in the NA RI population (to be discussed). This QTL is closely linked to Lf1 (Cregan et al., 1999), a locus controlling leaflet number (Fehr, 1972). This suggests that a common regulatory mechanism may be operating in both leaf and floral primordial tissue.

As expected, the number of seeds per pod was regulated by the same QTL haplotypes on U13 that determined the number of ovules per pod. However, this trait was not affected by QTLs on U14 or U22, probably because the increase in ovules per pod was offset by abortions controlled by QTLs in the same linkage groups (see below). Another seeds per pod QTL, located on U1, occurred in the same region as a QTL for water use efficiency (Specht et al., 2001) as well as a QTL for resistance to brown stem rot.

Seeds per ovule, a measure of abortion frequency, is regulated by three QTLs (Table 4). One, located on U11, is in the same region as QTLs for flowering date, RP, and maturity, as well as other traits (e.g., yield) that are affected by control of these reproductive stages (Orf et al., 1999). The locus on U14 is within the support interval that contains Dt1. One effect of the U14 Noir 1 haplotype is to increase the number of ovules per pod. Fig. 2 suggests that under competitive conditions—for example, the Minnesota environment—increasing the number of ovules per pod decreases the number of seeds per ovule. Thus, the QTL on U14 may have pleiotropic effects stemming from the regulation of the number of ovules per pod. The most important QTL affecting abortion is found on U22 in the same region in which a QTL for water use efficiency was identified using 13C stable isotope fractionation (Specht et al., 2001). The ability to fix carbon efficiently under conditions of even very short term stress could easily influence abortion frequency.

Table 5 presents QTLs for the frequencies of abortions occurring at the two developmental stages at which abortions were most prevalent. The QTL on U22 exerted its effect on partially developed embryos (Table 4), affecting such embryos in all pod positions. It had no effect on embryos in other stages of development. Abortion of partially developed embryos was much more frequent in Minnesota than in Utah (Table 1), and in the rain fed, high-density row plots in Minnesota, competition for water would more easily result in water stress than in the irrigated hill plots in Utah. The QTL on U9 is linked to loci for QTLs for RP and maturity, both of which can affect the completion of seed development. Quantitative trait loci affecting abortions at the time of fertilization (none) are different from those affecting abortion of partially developed embryos. They include major QTLs affecting flowering date on U11 as well as the region containing several disease-resistant loci on U13. Other position-specific QTLs on U2 and U7 could not be associated at this time with loci affecting other phenoytpes.

In summary, abortions at the time of fertilization (no development) are associated with QTLs for flowering date, placing the embryos at hazard with respect to environmental effects of climate and photoperiod. In contrast, abortions at the later stage of partially developed embryos are associated with QTLs that can affect water stress or that control the increase in number of ovules in a pod, placing a greater drain on the resources available to individual embryos.

Comparison of Three Recombinant Inbred Populations
The three RI populations—Minsoy-Noir 1, Minsoy-Archer and Noir 1-Archer (Orf et al., 1999)—were compared in the 1998 Minnesota environment. In this trial, we also measured yield and counted the number of pods per plant. Table 6 compares the seed set parameters of the three populations. It can be seen that the Archer populations had more seeds per pod and fewer abortions. Although the NA population had more seeds per pod, it had significantly fewer pods per plant. As a result, the overall number of seeds per plant was about the same as the MA population. Nevertheless, the NA population had, on average, much higher yields. Moreover, whereas the seeds per plant were correlated with yield in the MN and MA populations, no such correlation was observed in the NA population (data not shown). Since the yield increase and increase in seed number (yield/seed weight) could not be attributed to the number of seeds per plant, the data suggest that the increase came from an increase in the number of plants per row. In the future, it should be possible to identify QTLs for this parameter by comparing seeds per plant and yield in segregating populations. A preliminary attempt to do this identified QTLs for plants per row in the NA population on U1 (LOD 3.3) and U26 (LOD 3.6). In both cases, the elite parent, Archer, provided the allele for higher plant density. The U1 locus was linked to a QTL for water use efficiency (Specht et al., 2001) and also affected the number of seeds per pod (Table 3).


View this table:
[in this window]
[in a new window]
  		Table 6. Means of soybean seed set parameters, pods per plant, and yield in the Minsoy-Noir 1 (MN), Minsoy-Archer (MA), and Noir 1-Archer (NA) recombinant inbred populations grown in the 1998 field trial in Minnesota.

 
Using the seed set and abortion parameters described above, we identified QTLs in the three populations (Tables 7 and 8). Table 7 presents the number of QTLs, LOD >= 2.5, and the percentage of the phenotypic variation that they explain. Quantitative trait loci identified in all three populations explained comparable amounts of variation in ovules per pod and seeds per pod. However, segregants of Minsoy parents explained more of the variation in QTLs affecting abortions. Table 8 identifies specific QTLs in the three populations. In several instances, QTLs identified in the MN population (Tables 4 and 5) also were found in either the MA or NA populations [Table 8 ({dagger})], supporting the significance of those QTLs. Most striking was a very strong QTL on U3 for ovules per pod found in the Noir 1-Archer population in the region of Lf 1, that supported the relationship between ovules per pod and number of leaflets discussed above (Table 4). The QTL for ovules per pod on U13 in the region of male sterile loci found in the MN population also segregated in the MA population affecting the number of seeds per pod as well. In addition, the second, lesser QTL on U13 could be identified in the MA population.


View this table:
[in this window]
[in a new window]
  		Table 7. Summary of soybean seed set and abortion quantitative trait loci (QTLs; LOD >= 2.5) parameters for the three recombinant inbred populations grown in Minnesota [Minsoy-Noir 1 (MN); Minsoy-Archer (MA); and Noir 1-Archer (NA)]. The number of QTLs and the percentage phenotypic variation explained (R2) are presented.

 

View this table:
[in this window]
[in a new window]
  		Table 8. Details of soybean quantitative trait loci (QTLs) identified in the three recombinant inbred populations grown in Minnesota in 1998: Mapping, significance, and effect of parental alleles. Quantatative trait loci are located by linkage group (LG), support interval, as well as by a linked marker within the support interval. Quantatative trait loci are noted ({dagger}) in the Minsoy-Noir 1 (MN) and Noir 1-Archer (NA) populations that support QTLs already identified in two environments using the Minsoy-Noir 1 (MN) population (Tables 4 and 5). In many instances ({ddagger}), QTLs were supported by finding similar QTLs of lesser significance (2.5 < LOD < 3.0) in a different recombinant inbred population [e.g., the QTL for ovules per pod found on LG U10{ddagger} in the MN population is supported by a lesser QTL found in the NA population].

 
On U22, the QTL for ovules per pod (Table 4; U22 support interval 18–48) was confirmed as a QTL for seeds per pod in the NA population (Table 8; U22 support interval 14–44). Similarly, the QTLs for partial abortion (Table 5; U22, support interval 90–126) were found in the MA population. In addition, another QTL occurs in the MA population in an adjoining region (Table 8; U22 support interval 52–66). Abortion data for seeds per ovule suggest that the support interval on U22 between 96 and 128 could contain more than one QTL; that is, one between 96–118 and the other between 118–128. However, this relationship should be clarified as more markers become available for this region.

Finally, we note that although the Ln locus (four seeds per pod, Palmer and Kilen, 1987) on U17 (Cregan et al., 1999) could not be detected in the MN or NA RI populations, this locus segregated in the MA population (Table 8) with a significant LOD score (4.1).

Quantitative Trait Loci affecting Seed Set and Abortion: Conclusions
Quantitative trait loci on U11, U13, and U22 had the greatest effect on parameters of seed set. The QTL(s) on U11 was linked to QTLs previously shown to affect flowering date, RP, and maturity. The effect of this QTL on abortion was confined almost entirely to lack of development of embryos (none). This might most easily be interpreted as environmental effects on fertilization during flowering.

A major QTL on U13 affecting ovules per pod was associated with genes for male sterility (Ms1, Ms6) as well as a desynaptic locus (St5) affecting both male and female sterility. This raises the possibility that in the distal position, failure of ovule development or of ovule fertilization terminates pod elongation. If this were so, we might expect that fewer none abortions would be observed in terminal positions. This, in fact, was the case (Table 2). Unlike the mutants of Ms1, Ms6, or St5 studied by Palmer and others (Johns et al., 1981; Palmer and Kaul, 1983; Skorupska and Palmer, 1989), the effects that we see would have to be attributed to alleles with low penetrance, possibly interacting with environmental effects. It is interesting to note that this QTL on U13 also affects partial abortions in Position 1 (Tables 5 and 8). This probably is due to resource limitation when the number of ovules per pod increases (Fig. 2b). In support of this explanation, we have observed a 30% correlation between partial abortions in Position 1 and the number of ovules per pod.

Another QTL on U13 affecting ovules per pod occurs in a region characterized by genes for disease resistance. Although this may be coincidental, it could reflect an effect of low levels of infection on resources available for pod development or it may be that alleles producing disease resistance may reallocate resources away from reproduction.

U22 contains a region with a QTL that strongly affects abortion of partially developed embryos as well as a QTL that affects water use efficiency. This linkage demonstrates the importance of water use efficiency during seed development. We also observed a linkage between a QTL on U1 for water use efficiency and QTLs for seeds per pod (Table 4) and, possibly, for the number of plants per row.

Finally, the number of ovules per pod may be strongly regulated by a QTL on U3 linked to Lf1, as seen in segregants from the NA RI population (Table 8), suggesting that manipulation of Lf1 and Lf2 could be used to increase yield and seed number.

Our data suggest that when resources are limiting (high plant density as in Minnesota), abortions increase with increasing numbers of ovules per pod. The number of pods per plant may also vary as can the optimal number of plants per row. In the interest of improving yield, it may be useful to optimize the trade-offs between these parameters, locating the QTLs that play an important role in balancing these effects. As a first step it would be useful to determine the heritable variation in these parameters. Our study indicates that this could easily be made a part of ongoing yield experiments by counting: (i) the seeds per plant; (ii) the number of partial abortions per plant (the most frequent abortion type); and (iii) the number of plants per row. Use of differential screens could easily separate abortions of partially developed seed (Fig. 1b) from fully or almost fully developed seed using a few test plants chosen at random from each row in yield tests.


   ACKNOWLEDGMENTS
 
The authors thank Earl Hansen and employees at the Brigham Young University (BYU), Spanish Fork Farm for assistance with care and watering of the soybean crop. Special thanks are given to Susan Crook for her field assistance. We also thank several undergraduate assistants from BYU: Tiffany Isaksen, Jennifer Lambert, Suzanne Reeves, Randall Knight, Nancy Nelson Tischner, Tamara Heaton, Dave Gammon, Eric Shupe, and Dane Hatch. The College of Biology and Agriculture, BYU, provided funding for this project. Research at the University of Utah and at the University of Minnesota was supported by grants from the United Soybean Board.

Received for publication April 10, 2002.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tischner, T.
Right arrow Articles by Lark, K. G.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Tischner, T.
Right arrow Articles by Lark, K. G.
Agricola
Right arrow Articles by Tischner, T.
Right arrow Articles by Lark, K. G.
Related Collections
Right arrow Cell Biology & Molecular Genetics
Right arrow Crop Genetics
Right arrow Crop Growth and Development
Right arrow Soybean

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome