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


     

Published online 24 February 2006
Published in Crop Sci 46:946-949 (2006)
© 2006 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
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 HighWire
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Blanche, S. B.
Right arrow Articles by Myers, G. O.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Blanche, S. B.
Right arrow Articles by Myers, G. O.
Agricola
Right arrow Articles by Blanche, S. B.
Right arrow Articles by Myers, G. O.
Related Collections
Right arrow Crop Cytology
Right arrow Crop Genetics
Right arrow Plant and Environment Interactions

CROP BREEDING, GENETICS & CYTOLOGY

Identifying Discriminating Locations for Cultivar Selection in Louisiana

Sterling B. Blanche* and Gerald O. Myers

Dep. of Agronomy and Environmental Management, Louisiana State Univ. Agric. Ctr., 104 M.B. Sturgis Hall, Baton Rouge, LA 70803

* Corresponding author (sblanche{at}agcenter.lsu.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
Plant breeders generally conduct the selection phase of their program at few locations, mainly on the basis of geography and resource limitations. We conducted this study to identify test locations that optimize genotype selection on the basis of discriminating ability and representativeness. GGE Biplot Pattern Explorer was used to rank six test locations in Louisiana for cotton (Gossypium hirsutum L.) lint yield and fiber length using data from the 1993 to 2003 Louisiana Official Variety Trials (early and medium maturity). Biplots were generated and distances between the "ideal" and actual test locations were measured. Locations with shorter distances were closer to the ideal location and were considered more desirable test locations for the traits of interest. Each test location's distance was standardized by the mean distance of all locations for each biplot. For lint yield, on the basis of its close proximity to the ideal test location and the low standard deviation, the most desirable selection location was St. Joseph loam (Commerce silt loam; fine-silty, mixed, nonacid, thermic, Aeric, Fluvaquent). Winnsboro nonirrigated and Bossier City were not good selection locations for lint yield. For fiber length, Winnsboro irrigated was ranked first and St. Joseph loam was ranked third. Winnsboro nonirrigated was ranked sixth. A composite distance, reflecting the distance between the actual and "ideal" location for lint yield weighted at 60% and fiber length weighted at 40%, was used to determine the desirability of test locations on the basis of simultaneous selection for lint yield and fiber length. St. Joseph loam ranked first, Winnsboro irrigated ranked second, and Winnsboro nonirrigated ranked sixth. St. Joseph loam or Winnsboro irrigated should be used for selecting cultivars for lint yield and fiber length. Winnsboro nonirrigated should not be used for selecting cultivars because of its low level of discrimination and unique behavior.

Abbreviations: Alexandria (ALEX) • Bossier City (BC) • Louisiana Official Variety Trials (LAOVTs) • St. Joseph clay (SJC) • St. Joseph loam (SJL) • Winnsboro irrigated (WIR) • Winnsboro nonirrigated (WNI)


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
GENOTYPE (G) x environment (E) interactions (GEI) have been studied regarding cultivar stability (Wricke, 1962; Finlay and Wilkinson, 1963; Eberhart and Russell, 1966; Baker, 1988; Lin and Binns, 1988; Kang, 1993; Yan, 2001) and environment groupings (Gauch and Zobel, 1997; Atlin et al., 2000; Trethowan et al., 2003; Yang et al., 2005). However, relatively few researchers have studied GEI to determine the desirability of test locations. Yan and Kang (2003) proposed using GGE Biplot Pattern Explorer (Yan, 1999; Yan et al., 2000) to examine GEI with respect to discriminating ability and representativeness of test locations as a measure of desirability.

A highly discriminating location is one that maximizes the observed genotypic variation among genotypes for a given trait. The efficiency and accuracy of cultivar selection for a given trait is greatly enhanced in highly discriminating locations compared with nondiscriminating locations. Therefore, the identification of highly discriminating locations for a single or combination of traits should be of paramount concern to breeders. The discriminating ability of a location is comprised of a variety of factors including soil type, pest pressure, field drainage, temperature, precipitation, soil fertility, and management practices. Some of these factors such as soil type are static and indigenous to each location. For example, Winnsboro, LA, is located on the Macon Ridge and is characterized as a slightly acid light-textured soil with a high aluminum content. An array of genotypes exhibiting any degree of variation in aluminum toxicity would be highly discriminated against at this test site compared with a random test site. St. Joseph, LA, while geographically close (48.3 km) to Winnsboro, is characterized as an alluvial, deep, highly fertile silt loam and a high-yield environment. Any number of the characteristics inherent to a soil type (texture, fertility, organic matter content, etc.) could affect the discriminating ability of a location. Alternatively, dynamic factors such as pest pressure, precipitation, temperature, and management practices fluctuate yearly, although some trends are evident over years. A discriminating location should portray a favorable array of both static and dynamic factors with reasonable repeatability. Ideally, a plant breeder would conduct the selection and early testing phase of the breeding program in the location that provides the most information regarding cultivar separation for each trait. However, limited resources often inhibit that detail and most plant breeders use few test locations for selection (Lubbers, 2003).

In addition to exhibiting a high level of discrimination, an ideal test location should also be representative of the target growing region, or megaenvironment (Lubbers, 2003). Yan (2001) and Yan et al. (2001) discussed the use of GGE Biplot Pattern Explorer to categorize locations into megaenvironments. Traditionally, cotton (Gossypium hirsutum L.) breeding companies have used test locations in various megaregions, (e.g., the Mid-South, Southeast, Southwest, Far West), and cultivar selection at those sites is primarily targeted for that region. The shifted multiplicative model (SHMM) has been used to observe the associations among locations and their similar tendencies to differentiate among genotypes and to identify locations with a high degree of representativeness (Trethowan et al., 2003; Lillemo et al., 2004). In studies conducted by Trethowan et al. (2003) and Lillemo et al. (2004), sites with a low level of association with other global sites indicated that those sites were irrelevant for predicting global yield performance, whereas locations with a high level of association with other global sites would be considered key sites and good predictors of global performance. Glaz et al. (1985) used Shukla's (1972) stability-variance parameter to identify similar location pairs and single degree of freedom interactions to determine which of the location pairs identified contained the most similar locations.

Identification of an ideal test location on the basis of discriminating ability and representativeness implies that selections made at that site would have the highest probability of representing truly superior genotypes that perform well in all locations in the growing region. Major benefits to breeders would include the increased efficiency of selecting in discriminating locations and the discontinued use of poorly discriminating locations. Thus, cultivar development can be achieved most efficiently within the limited resources available to breeders. Multiple trait selection is important in plant breeding because ideal cultivars must exhibit acceptable performance for multiple characteristics such as yield, quality, maturity, pest resistance, etc. Yan and Kang (2003) identify a method for cultivar evaluation on the basis of multiple traits; however, one limitation of their method is that multiple trait selection can be applied for cultivar selection but not determinations of discriminating ability and representativeness of test locations. Thus, a measure of a test location's desirability on the basis of multitrait selection is needed. Our objective was to use the method presented by Yan and Kang (2003) and present a method for determining an ideal test location on the basis of weighted simultaneous selection for multiple traits.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Early and medium maturity groups of the 1993 to 2003 Louisiana Official Cotton Variety Trials (LAOVTs) (LCES, 1993–2003) were used to construct 21 datasets for each year x maturity combination for analysis via GGE Biplot Pattern Explorer (Yan, 1999; Yan et al., 2000). In 1993, the LAOVTs were not separated by maturity group, so the 1993 LAOVT was analyzed as a medium maturity trial resulting in a total of 10 early maturity and 11 medium maturity datasets. The fact that the genotypes were not constant throughout the 10-yr period was irrelevant; genotypes were only used to calculate the desirability of the locations for each biplot.

The LAOVTs have traditionally been conducted at six test locations in Louisiana: Alexandria (ALEX), Bossier City (BC), St. Joseph loam (SJL) (Commerce silt loam; fine-silty, mixed, nonacid, thermic, Aeric, Fluvaquent), St. Joseph clay (SJC) (Sharkey clay; very-fine, montmorillonitic, nonacid, thermic Vertic Haplaquept), Winnsboro irrigated (WIR), and Winnsboro nonirrigated (WNI). While these six test locations do not encompass the entire range of potentially desirable selection environments in Louisiana, they have historically been used for variety testing and are intended to represent the major cotton-growing regions in Louisiana. These six test locations were analyzed to determine which test location of those in the study was most desirable for enhancing germplasm selection. Traits analyzed were lint yield and fiber length, alone and as a component of an indexed value. Yield data were obtained by weighing machine-picked seedcotton and multiplying by lint percentage and were transformed into kilograms of lint per hectare before standardization. Fiber length determinations were made at the Louisiana State University Cotton Fiber Testing Laboratory with HVI instrumentation and reported as the upper-half mean length before standardization.

GGE Biplot Pattern Explorer generates a biplot of an "ideal" tester, which is highly discriminating and an average of the locations in the dataset (representative). Therefore, the two-dimensional distance between the actual test location (ALEX, BC, SJL, SJC, WIR, WNI) and the "ideal" tester is an indication of the desirability of that location with respect to discriminating ability and representativeness for that trait. Yan and Kang (2003) provide a detailed explanation of the biplot calculations and "ideal" test site determinations. The distance (in mm) between each location marker and the "ideal" test location marker was determined (Fig. 1 ) and that distance was then standardized by the mean distance of all locations for each biplot. The standardized distances for each test location were averaged across the 21 datasets to obtain the mean distance from the ideal tester and standard deviation for each test location. The standard deviation was the deviation of the standardized distance of each test location from the ideal test location across 21 yr by maturity biplots. Since standardized data was used, the 21 yr x maturity biplots were treated as replications. These data are presented for lint yield, fiber length, and combined into a single selection index value.


Figure 1
View larger version (34K):
[in this window]
[in a new window]
  		Fig. 1. Example of biplot showing distance between actual and "ideal" location.

 
For the combined selection index value, the distance between each test location and the ideal location for lint yield was measured and given a 60% weight for each year x maturity biplot (i.e., 1996 early). For the same year x maturity biplot (i.e., 1996 early), the distance between each test location and the ideal location for fiber quality was measured and given a 40% weight. The mean distance of each test location's combined selection index value is the weighted average of the distance between the actual and ideal test location for lint yield (60%) and fiber length (40%) for each of the 21 biplots. Standard deviations were calculated as previously described. Generally, breeders select primarily for yield and secondarily for numerous quality components assigning weights to each trait on the basis of personal conviction. The weights given to each trait were assigned arbitrarily to show the method; in reality, the procedure is easily customizable to various scenarios. The resulting selection index value represents the distance from the ideal test location weighted 60:40 for lint yield and fiber length, respectively. Test locations with shorter distances relative to the ideal test location are regarded as the most suitable for maximizing selection progress. Mean separation was done with an F-protected Duncan's Multiple Range Test in the SAS System v. 9.0 (SAS, 2002).


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
The order of location desirability was SJL, SJC, ALEX, WIR, BC, and WNI for lint yield (Table 1). Among the test locations included in this study, SJL, SJC, ALEX, and WIR would be equally sufficient test locations for lint yield, whereas BC and WNI were less effective, either by failing to discriminate among cultivars or not representing the other test sites or growing regions. SJL had the lowest standard deviation of the six test locations indicating that it was consistently close to the ideal test location and fluctuated less across years (Table 1). Lubbers (2003) conducted a GEI study including 16 locations spanning the southeastern cotton belt from lower Alabama to just south of the North Carolina–Virginia border and west to Louisiana to identify ideal test locations for Phytogen Seed Company, LLC. He reported location groupings on the basis of maturity and separated the two maturity groups (early and late) into megaenvironments. Lubbers (2003) found that out of the seven test sites covering Mississippi, Arkansas, Missouri, Tennessee, and Louisiana in the late maturity group, St. Joseph, LA, was the ideal test location to select for lint yield.


View this table:
[in this window]
[in a new window]
  		Table 1. Standardized distances between actual and "ideal" locations, standard deviations, and rankings of six locations for lint yield.

 
Selection for fiber length alone would be most effective in WIR, followed by BC, SJL, ALEX, SJC, and WNI (Table 2). Among the test locations included in this study, WIR, BC, SJL, and ALEX would be equally sufficient test locations for fiber length, but SJC and WNI would be less effective. It is not likely that a separate test location would be used to select only for a quality trait; however, the test location distances for each individual quality trait is needed to create the multitrait selection index.


View this table:
[in this window]
[in a new window]
  		Table 2. Standardized distances between actual and "ideal" locations, standard deviations, and rankings of six locations for fiber length.

 
Selection of an ideal genotype is seldom based on any single criterion but rather on a composite of attributes. To investigate the ability to use a selection index to identify an ideal test location at which to make selections, we formulated one in which the major emphasis (60%) was given to lint yield and lesser emphasis (40%) was given to fiber length. For the individual traits, WIR and SJL ranked fourth and first for lint yield and first and third for fiber length, respectively, indicating that both were desirable test locations to select for each trait individually (Table 3). Table 3 also contains the average ranking of each test location for both traits and the combined indexed value representing simultaneous selection for lint yield (60%) and fiber length (40%). The composite indexed values indicate that SJL would be the most desirable test location for cultivar selection for both traits in Louisiana (Table 3). WIR, ALEX, and SJC were also acceptable test locations, whereas BC and WNI were not desirable either because they provided few meaningful selections or did not well represent the other test sites (growing regions) in Louisiana. It is possible that an exhaustive study using many more test locations than are included in this study would yield different results; however, the subset of potential test locations included in this study are the only selection sites that realistically lend themselves to cultivar evaluation because of available resources and expertise at these locations.


View this table:
[in this window]
[in a new window]
  		Table 3. Locations ranked according to desirability for lint yield, fiber length, and simultaneous selection (lint yield + fiber length) and standard deviations.

 
Plant breeders are usually restrained by resource limitations, and conducting selections in the most desirable test location for each individual trait may not be realistic. However, in some cases the economic potential for improving selection efficiency for a secondary trait may warrant trait-specific selection locations. It should be noted that the test locations included in this study were assumed to represent all possible regions in Louisiana, which was assumed to comprise a single megaenvironment. There may be various megaenvironments within the state in certain years; however, the state cotton breeder is responsible for servicing all of the growing regions in the state with limited resources, and it would be impractical to divide Louisiana into multiple selection environments. In comparison, Lubbers (2003) uses multistate data for identifying desirable selection environments using a regional perspective. The multiple-trait selection techniques employed by the authors in this study are adaptable to many different interests and the number of traits used for determinations and the weights given to each trait are subject to the convictions of the researcher. Certainly test locations that are desirable selection environments for a combination of traits can prove beneficial to breeders interested in cultivar development. Therefore, the most ideal test location for breeders to use as a selection environment is one in which they can select with reasonable effectiveness for ancillary traits without compromising the ability to effectively select for yield.


   ACKNOWLEDGMENTS
 
The authors thank Cotton Incorporated and the Cotton Incorporated Fellowship Program for their support of this research. We greatly appreciate the assistance of all Louisiana Agricultural Experiment Station and Louisiana Cooperative Extension Service employees who helped in conducting the 1993-2003 cotton variety trials. A special appreciation is extended to Mr. David Caldwell for assimilating the 1993 to 2003 Louisiana Official Variety Trial data. Also, we'd like to thank Ivan Dickson, Gladys Carmona, Soni Iyer, and Melissa Ward at the LSU Fiber Testing Lab for providing the high-volume instrumentation data for fiber length. The authors would like to thank the reviewers for their insightful comments and contributions to this manuscript.

Received for publication April 5, 2005.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Crop Sci.Home page
W. Yan, M. S. Kang, B. Ma, S. Woods, and P. L. Cornelius
GGE Biplot vs. AMMI Analysis of Genotype-by-Environment Data
Crop Sci., March 1, 2007; 47(2): 643 - 653.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
X.-M. Fan, M. S. Kang, H. Chen, Y. Zhang, J. Tan, and C. Xu
Yield Stability of Maize Hybrids Evaluated in Multi-Environment Trials in Yunnan, China
Agron. J., January 1, 2007; 99(1): 220 - 228.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
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 HighWire
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Blanche, S. B.
Right arrow Articles by Myers, G. O.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Blanche, S. B.
Right arrow Articles by Myers, G. O.
Agricola
Right arrow Articles by Blanche, S. B.
Right arrow Articles by Myers, G. O.
Related Collections
Right arrow Crop Cytology
Right arrow Crop Genetics
Right arrow Plant and Environment Interactions

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