Our study of quantitative traits in Soybean has focused on recombinant inbred populations. These have been geneotipically characterized with molecular markers and phenotyped for agronomic traits. These data have been used to identify quantitative trait loci (QTLs) and interactions between QTLs (epistasis).

Quantitative traits are phenotypes, which vary continuously depending on the genotype and on environmental conditions. They are polygenic, determined by a number of genes which in the aggregate determine the phenotype although individually each gene may contribute only a small fraction of the observed trait value. For any particular level of heritability, three important factors will play a major role in the ability to identify and analyze a QTL: 1) the amount of trait variation controlled by a particular locus; 2) the nature of the segregating population being analyzed; and 3) the method used to analyze that population. Because plants are easily inbred and large populations of segregants may be bred to homozygosity as Recombinant Inbred Lines, they represent a very simple system.

Soybean (Glycine max) is an inbreeding plant in which the flower opens after pollination. It was domesticated more than 3000 years ago and has been under strong selection for many characteristics of agronomic interest. In the greenhouse it is possible to obtain three generations in one year; in the field, two. The genome comprises ca. 2500 cM of map distance distributed over 20 chromosomes. With the advent of molecular markers, genetic studies of soybean have progressed rapidly.

Recombinant Inbred populations
Recombinant Inbred (RI) lines are powerful genetic tools. In plants they are particularly useful because large numbers of RI segregants can be prepared and stored as seed. They have homozygous genotypes in which the naturally evolved balance of genetic material has been disrupted, producing new genotypes and often radically different phenotypes. Crossing two genetically distinct parents and then inbreeding to homozygosity produce RI lines. The resulting set of segregants will contain an admixture of the two parental genotypes as a result of chromosome segregation and recombination. If the parental genotypes are quite different, segregant genotypes exhibit transgressive variation in which progeny phenotypes are much more extreme than those of the parents from which they arose. Because of the genetic constancy of RI lines, correlations can be made between different experiments carried out in different environments and/or at different times. Because soybean is an inbreeding plant, RI lines are easily prepared and maintained.

We are using three large recombinant inbred populations developed by Levi Mansur (Mansur and Orf (1995), Crop Sci 35:422-425). They were derived by single seed descent from crosses of 'Minsoy' by 'Noir1'; 'Minsoy' by 'Archer'; or Noir1' by 'Archer'. Each includes ca 250 RI segregants and has been inbred for more than 10 generations. Seed for these populations can be obtained from:
 

Levi Mansur, Ph.D.	Jim Orf
P.O. Box 520   Los Andes, V Region   Chile	Department of Agronomy & Plant Genetics   411 Borlaug Hall, 991 Buford Circle   University of Minnesota   St. Paul, MN, 55108
Phone: 011-56-34-429127	Phone: 612-625-6275
FAX: 011-56-34-425879	FAX: 612-625-1268
levi@entelchile.net	orfxx001@maroon.tc.umn.edu

 

Genetic Characterization of RI populations
In soybean, RFLP and simple sequence repeat (SSR) markers have been developed. At present more than 600 SSR markers are available (mostly tri nucleotide repeats (ATT)n, (CAA)n, or (CTT)n) and an additional 300-400 will be targeted specifically to gaps in the molecular map. P. Cregan at USDA-BARC has developed all of the SSR markers. We have used RFLP and SSR markers to characterize each RI population and to construct genetic maps. These maps show markers in each of the 20 chromosomes as well as their position in the genome.
 

Minsoy X Noir 1 map

Minsoy X Archer map

Archer X Noir 1 map

 

Identification of Quantitative Trait Loci (QTLs)
We have chosen agronomic traits as quantitative phenotypes. The methods for their measurement are well defined and the phenotypes can be readily compared with values in the literature for other soybeans.

Quantitative traits have been carefully measured in all three RI populations by collaborators in several different environments (including Minnesota (J. Orf), Chile, (L. Mansur) and Nebraska (J. Specht)). For each trait and population, the range of values for the segregants was far greater than the values for the parents -i.e. they observed transgressive variation for all phenotypes. The traits measured included such different phenotypes as height, days to flowering (R1) or to maturity (R8), seed weight, seed oil and protein content as well as fatty acid composition, leaf length and width, yield (total weight of seed/area planted), and drought resistance.
 
 

Scatter graph for the Minsoy X Noir1 RI population using the two traits: yield (y-axis) and protein (x-axis).  The Minsoy and Noir1 parents are identified.  There is considerable transgressive segregation for both traits.  There is a fairly high negative correlation between the traits.

Scatter graph for the Minsoy X Noir1 RI population using the two traits: height (y-axis) and leaf width (x-axis).  The Minsoy and Noir1 parents are identified.  There is considerable transgressive segregation for both traits.  There is a moderate correlation between the traits.

 
Identification of QTLs was made on the basis of linkage to qualitative markers. Significance was estimated using analysis of variance, maximum likelihood methods and maximum likelihood methods coupled with Monte Carlo simulations.
 
 

Epistat displays of the phenotypic effects of two markers (T155 and Satt006) on seed weight.  The different color curves represent the different alleles for each marker.  Each plant in the population is grouped according to its allele at the locus and graphed according to its phenotypic value (x-axis) and rank in the group (y-axis).  Linkage to a QTL gives rise to a separation between the two colored curves (the gray curve is the total population).  In this example T155 is linked to a QTL Satt006 is not.

QTLs for agronomic traits have been identified in the MinsoyxNoir population (Mansur, Orf, Chase, Jarvik, Cregan and Lark (1996), Crop Sci 36:1327-1336) and are being identified in the Minsoy x Archer and Noir x Archer populations.
 
 

Genome scans for primary QTL obtained using the composite interval mapping feature of plabqtl.

HEIGHT	LEAF WIDTH
PROTEIN	YIELD

 

(Ht=height; Lodge= lodging; Ht/lodge=a measure of the ability of a tall plant to stand upright; r8=maturity; r1=flowering date; r8-r1= reproductive period; r8/ht=a measure of late maturing short plants; ll=leaf length; lw=leaf width; ll/lw=leaf shape; lllw=leaf area; SW=seed weight; protein=seed protein; yield; YD/SW=seed number; YD/HT= measure of yield in short plants YD/R8=measure of yield in early maturing plants)
 

TRAIT	MARKER	LINKAGE GROUP	MAP POSITION	R2 (percent)	P-VALUE
Ht	A060a	1	11.6	4.63	0.0006
r8/ht	A060a	1	11.6	5.14	0.0008
YD/HT	A060a	1	11.6	6.96	<0.0001
Protein	Satt431	1	83.5	4.63	0.0011
Ht	B124b	2	92.9	5.87	0.0008
R8	B124b	2	92.9	5.93	0.0011
Lodge	B124b	2	92.9	6.85	0.0005
lwmean	Satt045	2	96.5	5.96	0.0001
SW	Satt369	2	108.7	4.67	0.0012
ll/lw	Satt369	2	108.7	6.17	0.0002
R8	A510a	3	166.5	6.9	0.0006
YD/SW	Satt421	3	227.8	5.37	0.0005
SW	Sat_040	3	230.3	9.63	0.0001
llmean	L050q	4	416.5	6.25	0.0008
SW	Satt038	5	422.9	6.53	0.0003
YD/SW	Satt038	5	422.9	6.88	0.0001
Protein	L050b	5	466	5	0.0005
R8-R1	Satt303	5	471.4	5.06	0.0013
lwmean	A378	5	535.1	6.74	0.0004
ll/lw	A378	5	535.1	19.7	<0.0001
llmean	BL004	6	563	5.4	0.0002
ll/lw	L103a	6	583.2	5.26	0.0005
Protein	GMABAB	6	613.9	5.06	0.001
ll/lw	BL015	6	615.3	5.77	0.0002
YD/SW	Satt174	7	732.8	5.77	0.0005
SW	T155	7	736.8	6.96	<0.0001
Protein	Satt225	7	738.1	9.08	<0.0001
Protein	Sat_036	8	819.9	5.05	0.001
SW	Sat_036	8	819.9	6.24	0.0003
Ht/lodge	A295	8	834.1	5	0.0011
R8-R1	A121	9	862.7	4.73	0.0013
llmean	L199a	9	887.4	4.99	0.0007
Protein	L199a	9	887.4	5.07	0.0009
R8-R1	L199a	9	887.4	6.23	<0.0001
SW	L199a	9	887.4	8.37	<0.0001
R8	Satt281	9	906	7.71	0.0001
R8-R1	Satt281	9	906	11.3	<0.0001
ll/lw	Satt205	9	976.7	6.72	0.0003
R8	Satt205	9	976.7	26.4	<0.0001
R1	Satt205	9	976.7	36.8	<0.0001
R8-R1	A109a	9	977.2	7.37	<0.0001
Lodge	A109a	9	977.2	16.8	<0.0001
YD/SW	Satt079	9	979.5	7.59	<0.0001
yield	Satt079	9	979.5	9.11	<0.0001
Ht	Satt079	9	979.5	17.9	<0.0001
YD/HT	BL029	9	980.4	9.39	<0.0001
r8/ht	BL029	9	980.4	15.6	<0.0001
lwmean	Satt353	10	1027	5.04	0.001
ll/lw	Satt353	10	1027	6.22	0.0003
lllw	Satt192	10	1066	5.03	0.001
lwmean	Satt192	10	1066	7.91	<0.0001
ll/lw	BL053c	10	1069	7.66	0.0001
SW	Satt150	11	1158	7.46	<0.0001
yield	Satt150	11	1158	7.97	<0.0001
YD/SW	Satt150	11	1158	11.1	<0.0001
R8	Satt150	11	1158	11.3	<0.0001
R1	Satt150	11	1158	12.5	<0.0001
r8/ht	R079	11	1179	6.24	0.0004
Ht/lodge	R079	11	1179	6.55	0.0002
YD/R8	R079	11	1179	7.97	0.0001
lllw	R079	11	1179	10.8	<0.0001
Ht	R079	11	1179	11.1	<0.0001
ll/lw	R079	11	1179	11.4	<0.0001
YD/SW	R079	11	1179	13.3	<0.0001
R8-R1	R079	11	1179	13.6	<0.0001
lwmean	R079	11	1179	15.8	<0.0001
yield	R079	11	1179	16.1	<0.0001
R8	R079	11	1179	29.6	<0.0001
R1	R079	11	1179	31	<0.0001
R8	A584	11	1195	8.94	<0.0001
lllw	Satt323	11	1198	4.53	0.0006
lwmean	Satt323	11	1198	9.3	<0.0001
ll/lw	Sat_003	11	1199	11.6	<0.0001
R1	Sat_003	11	1199	15.3	<0.0001
Ht/lodge	Satt389	12	1338	4.85	0.0007
Protein	HSP176	13	1467	5.91	0.0003
R8-R1	K644a	13	1468	8.95	<0.0001
Ht/lodge	R045	13	1469	7.76	<0.0001
llmean	BL010	14	1566	7.78	<0.0001
lllw	Satt446	14	1567	7.62	0.0001
R8-R1	Satt182	14	1570	4.94	0.0005
YD/HT	Satt182	14	1570	6.03	0.0003
Lodge	Satt182	14	1570	7.34	<0.0001
r8/ht	Satt182	14	1570	7.48	<0.0001
Ht	Satt182	14	1570	8.02	<0.0001
YD/HT	Satt481	14	1618	12.2	<0.0001
R8-R1	Sat_113	14	1635	8.68	<0.0001
Ht	Sat_113	14	1635	10.3	<0.0001
R8	L050e	14	1638	7.13	0.0006
YD/SW	Sat_099	14	1643	6.33	0.0005
SW	Sat_099	14	1643	7.9	<0.0001
R1	Dt1	14	1657	8.1	<0.0001
lllw	Dt1	14	1657	8.13	<0.0001
R8-R1	Dt1	14	1657	9.69	<0.0001
R8	Dt1	14	1657	11.8	<0.0001
llmean	Dt1	14	1657	14.6	<0.0001
Lodge	Dt1	14	1657	24.6	<0.0001
r8/ht	Dt1	14	1657	29.6	<0.0001
Ht	Dt1	14	1657	35.5	<0.0001
YD/HT	Dt1	14	1657	39.5	<0.0001
Protein	L103b	14	1669	6.39	0.0001
llmean	Satt373	14	1675	9.82	<0.0001
R8	Satt384	18	1799	4.61	0.0011
R1	Satt384	18	1799	4.89	0.0005
R1	Satt046	24	2229	7.36	<0.0001
SW	Sat_077	28	2410	6.23	0.0002

QTLs are listed by their position in the genome.  The table presents the trait, the closest marker to which a QTL for the trait is linked together with the genome position (cM) and linkage group of that marker, the amount of phenotypic variation accounted for by the locus (R²) and the significance of the QTL identification (p-value).

In all, 106 loci were identified with p-values ranging from 0.0013-0.001 for a few, to 56 QTLs which had a significance of <0.0001.  Values for R2 ranged from 4.5% to 40%.  QTLs were clustered on linkage groups 9, 11 and 14 indicating a clustering of agronomic genes and/or pleiotropic effects of single genes.
 

Interactions between QTLs (Epistasis)
Interactions between genes controlling qualitative phenotypes have been known for a long time.  Usually they have been documented as allele specific interactions in which the activity of an allele at one locus is conditional upon a specific allele at another locus.

Interactions occurring between QTLs produce quantitative phenotypes, which cannot be explained by simply summing the phenotypes of the individual loci.  Studies of epistatic effects involving quantitative traits have been for the most part confined to interactions between identified QTLs both of whose individual effects are independently established and usually large (like qualitative traits).  Recently a more general search for interactions between QTLs has been attempted.  In these studies, loci or chromosomal regions were sought which of themselves had no discernible effect or phenotype, but either interacted to produce a phenotype, or altered the quantitative phenotype governed by some other previously identified QTL.  For these studies it was necessary to use the large number of segregants which could be provided by plant populations.  Epistatic effects in the 'Minsoy-Noir' population were suggested by the transgressive segregation of phenotypes (Mansur, L.M., K.G. Lark, H. Kross and A. Oliveira (1993) Theor. Appl. Genet. 86:907-913) and the asymmetric distribution of phenotypic values associated with particular loci (Lark, K.G., J. Orf and L.M. Mansur (1994) Theor. Appl. Genet. 88:486-489.).  We subsequently analyzed the effects of pairs of loci on phenotypic values and evaluated the significance of such effects using  Epistat,  a computer program which identifies and evaluates pairs of loci whose combined effects can not be explained by independent and additive action (Chase, Adler and Lark (1997) Theor. Appl. Genet. 94: 724-730).  For any pair of loci, this program displays the cumulative distributions of phenotypic values of the four subpopulations corresponding to the different possible genotypes and uses maximum likelihood methods together with Monte Carlo simulations to evaluate the significance of non-additive effects (interactions).  The method is extremely robust with respect to differences in the distribution of trait values and has identified interactions in the Minsoy x Noir RI population (Lark, Chase, Adler, Mansur and Orf (1995) Proc. Nat'l. Acad Sci US 92:4656-4660).