| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Crop Science, North Carolina State Univ., Raleigh, NC 27695-7631 USA
b USDA-ARS and Dep. of Crop Science, North Carolina State Univ., Raleigh, NC 27695-7631 USA
tommy_carter{at}ncsu.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and methodsResults and discussionREFERENCES |
|---|
Abbreviations: CP, coefficient of parentage • MDS, multidimensional scaling • NC, northern China, i.e., Huanghe, Huaihe, and Haihe valleys • NEC, northeastern China • SC, southern China • SP, spring • SU, summer • FA, fall • WI, winter
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionREFERENCES |
|---|
The prolific nature of Chinese soybean breeding makes systematic assessment and conscious preservation of diversity in the Chinese breeding pool difficult. An awareness of the diversity patterns in Chinese soybean breeding may make the task of diversity management easier. Regional and provincial isolation, the introduction of new ancestors over time, the intended cropping system for which a cultivar was developed, and less tangible factors such as important breakthroughs in disease resistance or yield advances and breeder preferences, all potentially influence diversity patterns in Chinese cultivars. The objective of this study was to quantify genetic diversity in Chinese cultivars via coefficient of parentage (CP), and to determine the relative importance of these factors in explaining that diversity. Genetic diversity patterns analyzed and characterized in this way may help identify genetic bottlenecks in Chinese breeding and help elucidate the best use of modern Chinese soybean cultivars in future breeding programs.
|
Materials and methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionREFERENCES |
|---|
Cultivars were grown primarily in the region and province of origin. Therefore, each cultivar was assigned to its respective growing region, province, release era, and cropping system (Cui et al., 1999). A matrix of CP values for the 651 cultivars and their ancestors was computed by FORTRAN as described by Cui et al. (2000). The CP between two individuals is defined as the probability that a random allele at a random locus in one individual is identical by descent to a random allele at the same locus in another individual (Malécot, 1948; Kempthorne, 1957). Briefly, the CP values were computed by the formula,
CPXY = 
, where Y is a genotype, A and B are the parents of Y, and X is a second genotype that is not a descendent of Genotype Y. Mean CP was computed within and among the large descriptive groups using the MEANS procedure (SAS Institute, 1985a). This CP matrix and its subsets were subjected to various analyses described below.
Multivariate and Cluster Analysis of Coefficient of Parentage Data
Multidimensional Scaling
The approach employed here is similar to that described by Gizlice et al. (1996). The CP measures of similarity (i.e., 0 = unrelated, 1 = identity) are relative and do not represent coordinates in Euclidean space, a prerequisite for a cluster analysis described below. Thus, the first step in the analysis was to generate a set of Euclidean coordinates for each cultivar by multidimensional scaling (MDS) (SAS Institute, 1992). Inspection of the 651 by 651 CP matrix revealed that 104 of the 651 cultivars had very little or no pedigree relation with each other or the remaining Chinese cultivars. Because these cultivars did not have potential to influence cluster analysis results and reduced computational efficiency by their presence, these cultivars were dropped from the MDS analysis before proceeding. The MDS procedure was then applied directly to the CP matrix derived from the 547 remaining cultivars in a series of analyses (SAS Institute, 1992). By trial and observation, we found that an analysis with 70 dimensions produced output with an excellent fit to the 547 by 547 CP matrix, as measured by stress or badness-of-fit criterion (2.7%) and R2 (0.92). Stress is a measure of the extent to which a geometrical representation falls short of a perfect match with the original CP matrix (Johnson and Wichern, 1992). Low stress value indicates a better fit. A stress value of 0% represents a perfect fit while 20% is taken to indicate a poor fit. The R2 was calculated from the comparison of input CP data with predicted values derived from the MDS coordinates. The options used in MDS analyses (SIMILAR = 1, COEF = IDENTITY, and LEVEL = ABSOLUTE) are the same as those described by Gizlice et al. (1996).
Cluster Analysis
The 547 Chinese cultivars were subjected to the nonhierarchical cluster analysis FASTCLUS procedure using MDS-derived Euclidean coordinates as source data (SAS Institute, 1985b). For each FASTCLUS analysis, it was necessary to specify the number of clusters desired prior to analysis. To find an optimum analysis, we performed 46 separate cluster analyses, specifying 5 to 50 clusters. The MEANS procedure was used to calculate mean CP within and among clusters for each analysis.
Graphical Representation of Diversity Patterns
Mean CP values between all pairs of the 25 provinces were formatted into a matrix and subjected to MDS employing the ABSOLUTE option, two dimensions, and with SIMILAR set to max, where max is the maximum value of the off-diagonal CPs. A similar approach was taken to develop the graphical representation of clusters. Two-dimensional graphs were constructed from MDS-derived coordinates by PROC PLOT (SAS Institute, 1985a). The graphs depicted approximate genetic distances among provinces and clusters in terms of the original CP values (Gizlice et al., 1996). The R2 values indicated that two-dimensional MDS explained 38 and 40% of the variation among 25 provinces and 20 clusters, respectively.
The Relative Importance of Discriminator of Genetic Diversity in Chinese Cultivars
To determine the importance of a diversity pattern, we computed the percentage of variation for which the discriminator accounted in the CP matrix. The 651 by 651 CP matrix was subjected to regression analyses with clusters, growing regions, provinces, cropping systems, and release eras treated as independent class variables. The R2 from regressions was used to determine the relative importance of the variables in explaining CP. An arithmetic shortcut described by Gizlice et al. (1996) was adopted to compute the regression R2 in GLM (SAS Institute, 1985b). Models with single and combinations of variables were fitted and R2 values compared to address the relative importance of the factors. The 104 cultivars not included in FASTCLUS analyses (see details in multivariate and cluster analysis section) and those cultivars which could not be assigned to acceptable clusters, post hoc, (see details in results and discussion section) were pooled together and treated as a special catch all cluster in the regression analysis as a computational device.
|
Results and discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionREFERENCES |
|---|
|
|
|
|
Genetic Diversity among Contrasting Cropping Systems
The classification of cultivars for cropping systems is based on seven historical cropping systems in China. In NEC, only the spring-planted soybean (NECSP) production system is employed with soybean grown as a full-season crop. In NC, summer-planted soybean (NCSU) predominates and is grown in a double-cropping system after a winter crop such as wheat (Triticum aestivum L.). Spring-planted or full-season soybean (NCSP) is grown to a lesser extent. Because of the relatively warmer climate in SC, multiple-cropping rotation systems are used: spring- (SCSP), summer- (SCSU), fall- (SCFA), and occasionally winter-planted (SCWI) soybean. Most cultivars were grown in only one of the seven cropping systems.
It is believed that the NECSP, SCSP, and SCSU cropping cultivars have been genetically independent for many centuries (Gai, 1997). The low mean CPs between cropping systems for cultivars developed in this century confirmed the continued genetic separation of the cropping systems (Table 4) . The mean CPs between different cropping systems were virtually zero. The basis for the continued genetic independence of cropping systems to the present day is that modern cultivars developed for a particular cropping system were derived mainly from landraces also grown in that same cropping system. The CP values were generally very low within- as well as among-cropping systems, indicating a great degree of diversity among cultivars for each cropping system.
|
Genetic Diversity among Eras of Release
Of the 651 Chinese cultivars released between 1923 and 1995, approximately 60% were released after 1980. Half of these newer releases occurred in Northeastern China. Concomitant with cultivar release, genetic diversity among Chinese soybean cultivars has increased in recent decades (Table 5)
. The mean within-era CP decreased monotonically from a high of 0.072 for cultivars released in the 1960's to a low of 0.016 for those released in the 1990's. This trend was brought about through the continual introduction of new ancestors (Chinese landraces and exotic germplasm) to Chinese breeding (Cui et al., 2000). A similar low CP was achieved by the introgression of new germplasm into U.S. oat (Avena sativa L.) cultivars released from 1951 to 1985 (Souza and Sorrells, 1989). In China, the between-era mean CPs were low and ranged from 0.01 to 0.06, indicating that release eras were almost independent.
|
Cluster Analysis of Breeding Patterns
The previously identified factors, i.e., regions and provinces of China, soybean types for contrasting cropping systems, and release eras, are not likely to explain all trends that existed in Chinese soybean breeding. Breeding approaches usually go beyond geographical factors to take opportunistic advantage of breakthroughs in yield and pest resistance improvement as they occur. A popular new productive cultivar might provide sufficient interest to trigger its use by several breeders as a parent, for example. Such information may not be revealed by analyses described, thus far, in this paper. However, an empirical clustering technique based on CP may allow the identification and interpretation of some of these trends. To this end, we employed MDS to generate coordinates for cultivars and then FASTCLUS to group Chinese cultivars into clusters based on these coordinates (Gizlice et al., 1996).
A series of MDS analyses revealed that 70 dimensions produced an excellent Euclidean representation of the CP matrix (Johnson and Wichern, 1992). This fit was most clearly depicted by employing the MDS coordinates to construct a predicted CP matrix. Only 1.2% of the estimates deviated from the actual CP values by as much as 0.10 and only 0.4% by as much as 0.15. A series of nonhierarchical, disjoint cluster analyses using the FASTCLUS procedure and the MDS output indicated that a FASTCLUS analysis employing 43 cluster dimensions assigned the most cultivars (270) to acceptable clusters (20) and was the only clustering output presented in this paper (Tables 6 and 7) . An acceptable cluster was defined by the following criteria: (i) number of cultivars within a cluster was at least 4; (ii) the CP between the cluster under consideration and any other was less than 0.25; (iii) the average CP within a cluster was at least 0.25. This definition was similar to that employed with success by Gizlice et al. (1996) and Sneller (1994). For the acceptable clusters, within-cluster means ranged from 0.25 to 0.47 (equivalent to half-sibs and full-sibs relations, Table 6), while between-cluster mean CPs were very small or zero indicating that clusters were nearly unrelated. Among the 23 unacceptable clusters (based on the criteria above) from the 43-cluster analysis, three were rejected for small size (<4 members) and 20 for low within-cluster mean CP (<0.25).
|
|
|
|
Although the within-cluster mean CP was greater than 0.25 for the 20 acceptable clusters, most cluster members were not true full-sibs or half-sibs. Parent-offspring and grandparent-grandchild relationships accounted for a large portion of the pedigree ties among cluster members. The close relations within clusters also tended to be derived from long pedigrees involving the mating of cousins, etc. The minimum value of 0.25 for mean CP of an acceptable cluster was chosen in part because of the findings of Manjarrez-Sandoval et al. (1997), which indicated that cultivar pairs with CP values greater than 0.25 do not tend to make good parents in terms of yield improvement. By employing the value of 0.25 as a minimum for cluster means, we suggest that clusters identified in this analysis have breeding relevance, and that a breeder should avoid the mating of cultivars within a cluster unless there was good reason to the contrary.
Relative Importance of the Diversity Patterns
The goal of this paper was to identify genetic diversity patterns in Chinese soybean cultivars by coefficient of parentage. These patterns may help breeders establish a strategy for selecting parents, and thereby, achieve better management of genetic resources in their breeding efforts. Regression analysis showed that the three growing regions, the most easily identifiable breeding factor in Chinese soybean breeding, although important, accounted for only 11%
of the total variation in cultivar relationships. This result reinforces the notion that other major patterns of diversity must exist among Chinese soybean cultivars. Subdividing the regions further into seven cropping systems did not raise the R2 appreciably and, thus, did not explain genetic diversity patterns in Chinese cultivars very well. Dividing the three regions into 25 provinces raised the R2 to 0.19 and, thus, identified an important diversity pattern extant in Chinese soybean. This was noted graphically in Fig. 1, where Dimension 1 separated the regions and Dimension 2 separated provinces. The six release eras accounted for only 2% of the total variation in cultivar relationships, indicating their relative unrelatedness. In contrast, 20 clusters accounted for 41% of the variation in diversity patterns and indicated that clusters were more efficient than other factors in describing diversity patterns. The degree of success encountered using clustering analysis here is similar to that reported by Gizlice et al. (1996) for U.S. soybean.
Cluster and province taken together accounted for 49% of the variation in the CP matrix while clusters taken jointly with cropping system, region, or era accounted for only 45, 44, or 42% of the cultivar variation in CP, respectively. Thus, cluster, province, and region of origin appeared to offer the best descriptive analysis of CP in breeding terms. For example, cultivars taken from contrasting clusters and provinces were not likely to be related.
U.S. Cultivars in Chinese Breeding
Although the ancestral base of U.S. soybean cultivars has not increased appreciably in the past 40 yr, genetic diversity in Chinese soybean cultivars has continued to increase in recent decades (Cui et al., 2000; Delannay et al., 1983; Hymowitz and Bernard, 1991; Gizlice et al., 1994; Sneller, 1994). One important factor increasing diversity in Chinese cultivars has been the use of U.S. germplasm in Chinese soybean breeding. Since the 1974 release of the first Chinese soybean cultivar Shang Qiu 4212 with U.S. germplasm, Mamotan, in its pedigree, 145 additional Chinese soybean cultivars have been released with some portion of U.S. ancestry in the pedigrees (Table 7). A total of 24 U.S. strains appeared in the pedigree of the 146 Chinese cultivars developed during 1974 to 1995. Out of these cultivars, 108 (74%) had 25% or higher portion of U.S. ancestry, indicating U.S. germplasm was the parent or grandparent of those cultivars.
The importance of exotic germplasm in China is seen not only in the expansion of genetic diversity, but in the improvement of agronomic traits as well. High yield potential and pest resistance were often observed in the 146 cultivars which had U.S. ancestry. More than 20 of these cultivars are grown commercially in China at present (J. Gai, 1999, personal communication). For example, `Si Dou 11' was selected from the cross of `Si Dou 2 Hao' x `Williams' and released in 1987 in Jiangsu province. It has ranked first in Jiangsu yield trials to the present. `Ji Dou 7 Hao' was developed from the cross Williams x `Cheng Dou 1 Hao' and released in 1992 in Hebei province. Its yield set a record in a nation-wide competition sponsored by the National Soybean Breeding Program. At least seven cultivars were developed using frogeye leaf spot (Cercospora sojina Hara.) resistance derived from the U.S. cultivars Amsoy, Clark 63, and Corsoy. Two Chinese cultivars were derived with soybean cyst nematode (Heterodera glycines Ichinohe) resistance from the U.S. cultivar Franklin (Cui et al., 1999). Cultivar improvement employing intercrosses between U.S. and Chinese materials has been a remarkable breeding success story in China.
The basic pattern for use of U.S. strains was to hybridize northern U.S. materials with NEC or NC breeding stock. In rare cases, northern U.S. and SC materials were intercrossed as well as southern U.S. and NC or NEC materials. Some U.S. cultivars were used successfully in Chinese breeding programs with comparatively high frequency: Clark 63 (10 crosses), Amsoy (8 crosses), `Beeson' (8 crosses), Williams (6 crosses), `Wilkin' (5 crosses), and `Harosoy 63' (5 crosses) (Cui et al., 1999). Usually, U.S. cultivars were the male parents in crossing (83% of all cases). However, some U.S. cultivars, such as Williams, Beeson, and Amsoy were also used as female parents in crossing and therefore contributed cytoplasm to Chinese soybean cultivars. Amsoy cytoplasm is found in `Ji Dou 5 Hao' and `Feng Jiao 7607'; Beeson cytoplasm is found in `Ji Lin 27' and `Fen Dou 31'; and Williams cytoplasm is found in `Ke Feng 35', `Zhao Shu 18', `Yu Da Dou 2 Hao', and `Ji Dou 7 Hao' (Cui et al., 1999).
Despite the use of U.S. cultivars in Chinese breeding, the majority of the 490 cultivars released during 1973 to 1995 did not have U.S. germplasm in their pedigrees. The U.S. germplasm contributed only 7.3% of the total Chinese genetic base (Cui et al., 2000).
Implications to U.S. Breeding
More than 70 yr of breeding have improved U.S. soybean production substantially, raising yield levels and providing resistance to potentially devastating diseases (Specht and Williams, 1984). One unintended consequence of breeding progress, however, is that genetic diversity has decreased in applied U.S. soybean breeding programs. Pedigree analysis has shown that many recent U.S. cultivars are as closely related as half sibs (Carter et al., 1993; Sneller, 1994). Some have suggested that the highly related nature of U.S. soybean cultivars could pose a barrier to further breeding progress (Manjarrez-Sandoval et al., 1997; Zhou et al., 1998). One straightforward method to avert further erosion of diversity in the USA is to add new breeding stocks to applied programs. The successful introgression of U.S. germplasm into Chinese cultivars may indicate that Chinese cultivars could be used to increase diversity in U.S. cultivars as well. In the USA, recent field studies have shown that many modern Chinese soybean cultivars are much higher yielding than random plant introductions from the USDA Soybean Germplasm Collection (Carter et al., 2000; Bernard et al., 1998). Four DNA marker studies demonstrated that Chinese and U.S. cultivars constituted very distinct groups (Carter et al., 2000). These findings imply that Chinese cultivars constitute an untapped reservoir of diversity for U.S. breeding.
In choosing breeding stocks for U.S. cultivar improvement, an obvious rationale for selection of Chinese cultivars is to choose those with no U.S. pedigree unless there is a clear reason to the contrary (Table 7). In addition, the CP cluster analysis presented here may pose an efficient method to categorize Chinese cultivars for potential utilization in U.S. breeding. Clusters may represent winning breeding strategies in China by the very fact that so much breeding effort was devoted to a particular type of pedigree represented by the cluster. Choice of the most appropriate cultivar within a cluster may be determined by agronomic performance or, in the absence of other data, by recency of release. Only 12 of the 20 clusters we identified included some U.S. parentage. In cases where U.S. parents appeared, 50% or more of the cluster members were free of U.S. pedigree (Table 7). Therefore, there is much opportunity for breeders to introduce Chinese cultivars into U.S. breeding.
|
ACKNOWLEDGMENTS |
|---|
Received for publication February 4, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Smith Intellectual Property Protection for Plant Varieties in the 21st Century Crop Sci., July 1, 2008; 48(4): 1277 - 1290. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. M. Nichols, W. Lianzheng, Y. Pei, K. D. Glover, and B. W. Diers Variability among Chinese Glycine soja and Chinese and North American Soybean Genotypes Crop Sci., May 31, 2007; 47(3): 1289 - 1298. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Wang, R. Guan, L. Zhangxiong, R. Chang, and L. Qiu Genetic Diversity of Chinese Cultivated Soybean Revealed by SSR Markers Crop Sci., March 27, 2006; 46(3): 1032 - 1038. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. S. C. Smith, D. N. Duvick, O. S. Smith, M. Cooper, and L. Feng Changes in Pedigree Backgrounds of Pioneer Brand Maize Hybrids Widely Grown from 1930 to 1999 Crop Sci., November 1, 2004; 44(6): 1935 - 1946. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. N. Ude, W. J. Kenworthy, J. M. Costa, P. B. Cregan, and J. Alvernaz Genetic Diversity of Soybean Cultivars from China, Japan, North America, and North American Ancestral Lines Determined by Amplified Fragment Length Polymorphism Crop Sci., September 1, 2003; 43(5): 1858 - 1867. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Zhou, T. E. Carter Jr., Z. Cui, S. Miyazaki, and J. W. Burton Genetic Diversity Patterns in Japanese Soybean Cultivars Based on Coefficient of Parentage Crop Sci., July 1, 2002; 42(4): 1331 - 1342. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Cui, T. E. Carter Jr., J. W. Burton, and R. Wells Phenotypic Diversity of Modern Chinese and North American Soybean Cultivars Crop Sci., November 1, 2001; 41(6): 1954 - 1967. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Li, L. Qiu, J. A. Thompson, M. M. Welsh, and R. L. Nelson Molecular Genetic Analysis of U.S. and Chinese Soybean Ancestral Lines Crop Sci., July 1, 2001; 41(4): 1330 - 1336. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Zhou, T. E. Carter Jr., Z. Cui, S. Miyazaki, and J. W. Burton Genetic Base of Japanese Soybean Cultivars Released during 1950 to 1988 Crop Sci., November 1, 2000; 40(6): 1794 - 1802. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
0.0985 indicate no relationship. (Ningxia and Jiangxi provinces were deleted from the graph because cultivars from these provinces were not related to any other provinces)