Published online 1 July 2008
Published in Crop Sci 48:1337-1340 (2008)
© 2008 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
Mass Selection for Small Seed Size in Natto Soybean Populations and the Resulting Effect on Seed Yield
Elroy R. Cober* and
Harvey D. Voldeng
Agriculture & Agri-Food Canada, Eastern Cereal and Oilseed Research Centre, Bldg. 110, Central Exp. Farm, Ottawa, ON, Canada, K1A 0C6. ECORC Contribution no. 07-098
* Corresponding author (coberer{at}agr.gc.ca).
 |
ABSTRACT
|
|---|
Natto soybean [Glycine max (L.) Merr.] cultivars are typically small seeded and seed yield is generally lower than that of oilseed cultivars. We investigated the effects of mass selection for small seed size on seed yield. Two sets of populations were developed, from crosses of natto and larger-seeded parents, using mass selection; nine tested in three site-years in 2003 and 2004, and four tested from 2002 to 2004. Populations were advanced with mass selection for seed size applied to F3 and F4 seeds using either no. 14 or 15 round-hole sieves. F4:5 recombinant inbred lines (RILs) from AC Colibri/OT91-3, unselected for seed size, were evaluated from 2002 to 2004. Selection for small seed reduced seed weight but did not affect yield or days to maturity when comparing two sieve sizes. The seed yield of natto selections remained at about 80% of the oilseed check cultivars in this study. In nine populations, among progenies resulting from mass-selection for seed size, random selections were significantly higher yielding and later maturing but did not differ in seed weight compared to visually selected lines. In four populations segregating for leaflet shape, mass selection for small seed size resulted in more narrow-leaflet lines. Selection for narrow leaflets may aid in reducing seed size in some populations. The seed size distribution of the RILs suggests that selection for small seed size may cause a bottleneck in mass selected populations and result in small effective population sizes.
Abbreviations: MAD, modified augmented design • RILs, recombinant inbred lines
|
INTRODUCTION
|
|---|
SMALL SEEDS ARE DESIRABLE for natto, a traditional Japanese food made from cooked fermented whole soybean [Glycine max (L.) Merr.] seeds. Generally, beans 5 mm in diameter are preferred (Wilson, 1995); however, exporters from eastern Canada typically use 5.6-mm (no. 14) round-hole screens to condition natto soybean. Oilseed-type soybean cultivars in Ontario and Quebec generally have seeds two times larger in size compared to natto cultivars.
Small-seeded natto cultivars yield less than oilseed cultivars in eastern Canada. From 17 trials grown in Ontario and Quebec in 2000 to 2002, natto cultivars yielded 68 to 87% of a check oilseed cultivar (data not shown). It would be beneficial to use high-yielding, larger-seeded oilseed cultivars as sources of high yield in crosses with natto cultivars. Johnson et al. (2001) used single crosses between small- and normal-seeded parents as well as three-way crosses to determine whether these populations would yield sufficiently small-seeded progeny. The three way [small x (small x normal)] crosses produced acceptable small-seeded lines while the single crosses resulted in limited numbers of small-seeded lines. They did not evaluate the resulting acceptable lines in yield trials.
Recently, some natto exporters have expressed an interest in somewhat larger-seeded lines, especially if the seed yield could be increased as a result. We undertook studies to look at relaxing seed size requirements during mass selection of populations resulting from single and three-way crosses. The objective of this work was to investigate the effects of mass selection for seed size on seed yield and the relationship between seed size and seed yield, particularly to compare selections selected for seed size using small and somewhat larger-sized sieves.
|
MATERIALS AND METHODS
|
|---|
Fourteen populations were developed between crosses of small-seeded natto and larger-seeded high-yielding parents. Two differently structured sets of populations were developed using mass selection for 13 of these populations. The 14th population was developed without selection as a population of recombinant inbred lines (RILs). Thirteen populations were advanced using the bulk method with mass selection. The populations were divided into two subpopulations with mass selection for small seed size applied to F3 and F4 seeds by saving seeds which passed through either no. 14 (5.6 mm) or no. 15 (6.0 mm) round-hole sieves. F4 single plants were randomly selected and grown in F5 progeny rows. The F4:6 and following generations were grown in four-row plots, 1.6 by 5 m in size. Entire plots were combine harvested and seed yield was determined. The mass of 100 seeds randomly selected from each plot was determined and used as a measure of seed size. In the field, days from planting to maturity were observed. The GLM procedure of SAS (Version 8, SAS Institute, Cary, NC) was used for an analysis of variance. Environments were considered random, while all other effects were considered fixed.
Study 1
Nine crosses were made in a factorial mating scheme with two sets of three parents where each parent in Set 1 is crossed to each parent in Set 2. The first set of three parents was composed of natto breeding lines, one of which was commercialized: ‘Heron’ (Voldeng et al., 2000), X3702-15-2-7, and X3703-29-3-5. The second set of three parents was composed of breeding lines that resulted from crosses between ‘AC Colibri’ (Voldeng et al., 1997) and three larger-seeded oilseed-type lines: (X3660-B-B-15-4)AC Colibri/‘AC Albatros’ (Voldeng et al., 1996b), (X3662-B-B-10-3)AC Colibri/PS42, and (X3664-B-B-12-6)AC Colibri/OT91-3. The second set of three parents had an intermediate seed size. The lines in Study 1 resulted from crosses described as [natto x (natto x high yield)] although these crosses were not made using F1 plants as in a true three-way cross and as a result did not allow for as much opportunity for recombination. Each of the nine populations was divided into no. 14 or no. 15 subpopulations for seed size resulting from the use of different-sized sieves for mass selection. Selections among F4:5 progeny rows within each subpopulation were taken randomly and by visual selection for agronomic characters in the field. Approximately 20 lines were selected for each sub-subpopulation. Lines were evaluated in modified augmented design (MAD) trials (Lin and Poushinsky, 1985) at Ottawa and Nepean, ON, in 2003 and at Ottawa in 2004.
Study 2
Four populations were developed from crosses between a high-yielding oilseed cultivar OAC Erin, developed at the University of Guelph, and four natto parents which included two cultivars AC Colibri and ‘AC Pinson’ (Voldeng et al., 1996a) and two breeding lines. Two subpopulations were developed using mass selection with the two sieve sizes mentioned above. Approximately 30 F4:5 lines were randomly selected for each subpopulation. Lines were evaluated at Ottawa in MAD trials in 2002, and in two replicate generalized lattice design trials in 2003 and 2004. These four populations were segregating for leaflet shape at the Ln locus, and since the narrow-leaflet trait is associated with smaller seed size, the lines were scored for leaf shape, Ln or ln. If the selection for seed size did not distort the segregation of leaf shapes, there should be equal numbers of narrow- and ovate-leaflet lines. The chi-square test was used to test for fit of the 1:1 ratio of Ln:ln leaflet lines within and across populations.
Study 3
A RIL population of 208 F4:5 lines was developed from the cross AC Colibri x OT91-3, a larger-seeded high-yielding line without any selection for seed size. These lines were evaluated at Ottawa in MAD trials in 2002, and in two replicate generalized lattice design trials in 2003 and 2004.
|
RESULTS AND DISCUSSION
|
|---|
In Study 1, the use of a smaller sieve size, no. 14 vs. no. 15, for mass selection during two generations resulted in smaller mean seed size for resulting lines (Table 1
). While the reduction in seed size was significant, it was only about a 4% reduction compared to the no. 15 selections. Selection for smaller seed did not result in differences for seed yield or time to maturity. Visual versus random selections in these subpopulations did not result in different seed sizes, however random selections were 2 d later maturing and were higher yielding compared to visual selections. There was a significant cross by selection sieve size interaction for seed size. Therefore, the subpopulation means were plotted individually for seed yield versus seed size pooled over visual and random selections since there was no significant cross by selection type interaction (Fig. 1
). The no. 14 subpopulation always had smaller seeds than the no. 15 subpopulation but the magnitude of the difference varied. The seed yields of the subpopulations with different seed size varied. On average, natto selections yielded about 75% of OT91-3, the larger-seeded check (Table 1).
View this table:
[in this window]
[in a new window]
|
Table 1. Mean agronomic characteristics of check cultivars and subpopulations developed using different selection criteria for seed size from nine soybean populations, Study 1, grown at Ottawa and Nepean, ON, in 2003 and Ottawa in 2004.
|
|
View larger version (10K):
[in this window]
[in a new window]
|
Figure 1. Mean seed yield and size of nine soybean populations (Study 1) developed from crosses between natto and larger-seeded parents. Each population has a unique symbol. The open symbols show subpopulations resulting from mass selection with no. 14 sieves and the filled symbols resulted from selection with no. 15 sieves.
|
|
In Study 2, the use of the smaller sieve also resulted in a slightly but significantly smaller mean seed size (Table 2
) compared to the larger sieve. Similar to Study 1, the use of different sieve sizes to select for smaller seed did not result in significant difference in seed yield or time to maturity. On average, the natto selections yielded about 84% of OAC Erin, the larger-seeded parent. Because the four populations in Study 2 were segregating for leaflet shape, we scored the F5 progeny rows for leaflet shape. The narrow leaf trait has a pleiotropic effect on numbers of seeds per pod (Bernard and Weiss, 1973). In a study involving a cross between ovate- and narrow-leaflet parents, narrow-leaflet F2 plants had smaller seeds compared to ovate-leaflet plants (Dinkins et al., 2002). Following from these results, we hypothesized that selection for smaller seeds would result in more lines with narrow leaflets. The use of the smaller no. 14 sieve resulted in two of four subpopulations not fitting a 1:1 LnLn:lnln leaflet shape ratio, while the use of the no. 15 sieve resulted in one subpopulation which did not fit a 1:1 ratio (Table 3
). Mass selection for small seed size skewed the populations toward narrow-leaflet lines in all eight subpopulations. Using the mean numbers of lines across all four populations resulted in a lack of fit to a 1:1 ratio for both sieve sizes. In an RIL population from a cross between narrow- and wide-leaflet parents, there was a positive association between seed size and leaflet width (Cicek et al., 2006). Selecting for narrow-leaflet plants or progeny rows, in populations segregating for leaflet shape, may result in selection of more lines with smaller seeds.
View this table:
[in this window]
[in a new window]
|
Table 2. Mean agronomic characteristics of check cultivars and subpopulations developed using different selection criteria for seed size from four soybean populations, Study 2, grown at Ottawa, ON, from 2002 to 2004. Populations were developed from crosses between OAC Erin, a larger-seeded cultivar and small-seeded natto lines.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Number of lines with wide (ovate, Ln) or narrow (ln) leaflets, and chi-square value and fit to a Ln:ln 1:1 ratio, following mass selection for seed size with different sieve sizes in four soybean populations, Study 2. These populations were developed from crosses between OAC Erin, a larger-seeded, ovate-leaflet cultivar and four small-seeded, narrow-leaflet natto lines.
|
|
In Studies 1 and 2, we did not keep the larger seeds which did not pass through the sieves, nor did we calculate the proportion of the population that was selected using the mass selection criterion. In Study 3, selection for seed size was not applied and the resulting relationship between yield and seed size was plotted in Fig. 2
for the parents and resulting RILs. The distribution for seed size was skewed toward the small-seeded parent in our study which was similar to another RIL population which resulted from a small- by large-seeded cross (Cicek et al., 2006) To simulate seed size selection in the Study 3 RIL population, we used the mean seed size from lines mass selected in Studies 1 and 2 and applied it to the population in Study 3. We estimated that only 12 lines would have been selected using the no. 14 sieve and, and that 22 lines would have been selected using the no. 15 sieve from a total of 208 RILs in this population. There was little association between seed yield and seed size in this population (r = 0.26).
View larger version (18K):
[in this window]
[in a new window]
|
Figure 2. Seed yield and size of recombinant inbred lines (open circles) developed from a cross between ‘AC Colibri’ and OT91-3 (closed triangles) from trials grown at Ottawa from 2002 to 2004 (Study 3). The vertical reference lines simulate selection for small seed size using no. 14 and no. 15 sieves from mean seed sizes seen in selections observed in Studies 1 and 2.
|
|
In a study investigating recovery of small-seeded lines, both (small x normal) and [small x (small x normal)] crosses were compared (Johnson et al., 2001). Using the small-seeded parent for comparison, only a small proportion of lines were equal to or smaller than the small-seeded parent in size; 4% for the single crosses and 20% for the three-way crosses (Johnson et al., 2001). In our Study 3, only about 1% of lines were equal to or smaller than the small-seeded parent (Fig. 2). Using the estimated seed size from Studies 1 and 2, the selection intensity in Study 3 would have been about 6% using the no. 14 sieve and about 11% using the no. 15 sieve. It is possible that a selection bottleneck occurred at the first mass selection for small seed in our populations in Studies 1 and 2; with the selection intensity being high, the result was a small effective population size from which to work in the second round of mass selection and in the evaluation of progeny rows.
The results of this study lead to research questions regarding the compatibility of small seed size and high seed yield as breeding objectives. Are small-seeded soybeans inherently lower yielding? Are small-seeded soybeans sink limited? If small seed size does not have a pleiotropic effect resulting in reduced seed yield, a method must be developed which allows recombination of favorable alleles for both traits. It may be useful to use recurrent selection for seed size or use relaxed or graduated selection for seed size when introgression of yield is being attempted with larger-seeded parents.
In summary, somewhat relaxed selection for small seed size (use of no. 15 rather than no. 14 sieves) did not result in soybean lines that were higher yielding. The seed yield of these natto selections remained at about 80% of the oilseed check cultivars used in this study. Selection for narrow-leaflet lines in populations that are segregating for leaflet shape may be of some use in reducing seed size in some populations. In populations developed from large- and small-seeded parents, the use of mass selection for small seed size will reduce the population size drastically, necessitating larger initial numbers of F2 plants.
|
NOTES
|
|---|
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.
Received for publication July 16, 2007.
|
REFERENCES
|
|---|
- Bernard, R.L., and M.G. Weiss. 1973. Qualitative genetics. p. 117–154. In B.E. Caldwell (ed.) Soybeans: Improvement, production, and uses. ASA, Madison, WI.
- Cicek, M.S., P. Chen, M.A. Saghai Maroof, and G.R. Buss. 2006. Interrelationships among agronomic and seed quality traits in an interspecific soybean recombinant inbred population. Crop Sci. 46:1253–1259.[Abstract/Free Full Text]
- Dinkins, R.D., K.R. Keim, L. Farno, and L.H. Edwards. 2002. Expression of the narrow leaflet gene for yield and agronomic traits in soybean. J. Hered. 93:346–351.[Abstract/Free Full Text]
- Johnson, S.L., W.R. Fehr, G.A. Welke, and S.R. Cianzio. 2001. Genetic variability for seed size of two- and three-parent soybean populations. Crop Sci. 41:1029–1033.[Abstract/Free Full Text]
- Lin, C.S., and G. Poushinsky. 1985. A modified augmented design (type 2) for rectangular plots. Can. J. Plant Sci. 65:743–749.
- Voldeng, H.D., J.A. Fregeau-Reid, E.R. Cober, and R.J.D. Guillemette. 2000. Heron soybean. Can. J. Plant Sci. 80:821–822.
- Voldeng, H.D., J.A. Fregeau-Reid, R.J.D. Guillemette, D.A. Leonard, and E.R. Cober. 1996a. AC Pinson soybean. Can. J. Plant Sci. 76:803–804.
- Voldeng, H.D., J.A. Fregeau-Reid, R.J.D. Guillemette, D.A. Leonard, and E.R. Cober. 1997. AC Colibri soybean. Can. J. Plant Sci. 77:113–114.
- Voldeng, H.D., R.J.D. Guillemette, D.A. Leonard, and E.R. Cober. 1996b. AC Albatros soybean. Can. J. Plant Sci. 76:151–152.
- Wilson, L.A. 1995. Soy foods. p. 428–459. In D.R. Erickson (ed.) Practical handbook of soybean processing and utilization. AOCS Press, Champaign, IL.
TOP
ABSTRACT
INTRODUCTION
