Agronomy Journal Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published in Agron J 100:760-764 (2008)
DOI: 10.2134/agronj2006.0282
© 2008 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
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 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 Google Scholar
Google Scholar
Right arrow Articles by Sabaghnia, N.
Right arrow Articles by Sabaghpour, S. H.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sabaghnia, N.
Right arrow Articles by Sabaghpour, S. H.
Agricola
Right arrow Articles by Sabaghnia, N.
Right arrow Articles by Sabaghpour, S. H.
Related Collections
Right arrow Plant and Environment Interactions
Right arrow Statistics

LEGUMES

Graphic Analysis of Genotype by Environment Interaction for Lentil Yield in Iran

Naser Sabaghniaa, Hamid Dehghania,* and Sayyed Hossain Sabaghpourb

a Dep. of Plant Breeding, Tarbiat Modares Univ. P.O. Box 14115-336 Tehran, Iran
b Dryland Agricultural Research Institute, Kermanshah, Iran

* Corresponding author (dehghanr{at}modares.ac.ir).


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
Selection of lentil (Lens culinaris Medik) cultivars with wide adaptability across diverse farming environments is important, before recommending them to achieve a high rate of cultivar adoption. Seed yield of 11 lentil genotypes, tested in a randomized complete-block design with four replicates across 20 environments in Iran, was analyzed using site regression (SREG) stability model. The biplot technique facilitates a visual evaluation of superior genotypes, which is useful for cultivar recommendation and megaenvironment identification. A substantial amount of genotype x environment (GE) interaction for lentil grain yield was detected. Location (L) and genotype x location (GL) variabilities were the predominant components of total yield variation. The first two principal components (PC1 and PC2) of the SREG model accounted for 76% of the total GE interaction. There were four winning genotypes and three megaenvironments according to the SREG model. The best genotype in one location was not always so in other test locations. According to the ideal-genotype biplot, genotype G5 was better than all other genotypes; G5 exhibited both high mean yield and high stability of performance across environments. According to G + GE sources of variations, the genotypes (G4, G7, G9, and G10) were the most suitable varieties for the lentil-producing regions in Iran.

Abbreviations: ICARDA, International Center for Agricultural Research in Dry Areas • E, environment main effect • G, genotypic main effect • GE, genotype x environment interaction • GGE, G plus GE • MET, multiple-environment trials • SREG, site regression • SVD, singular value decomposition


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
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.

Received for publication October 11, 2006.
   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
NEW GENOTYPES generally need to be tested at many locations and for several years before being recommended for production for a given area. To achieve this goal, multiple-environmental trials (MET) are conducted annually for all major crops throughout the world with the purpose of identifying superior genotypes for the target locations. In most cases, GE interaction is observed and needs to be modeled and interpreted. The term GE interaction commonly refers to yield variation that is caused by the interaction between genotypes (G) and environments (E). Numerous methods have been used in the search for an understanding of the causes of GE interaction (Flores et al., 1998). These methods represent three major analytical categories, univariate parametric analysis of the GE matrix, nonparametric methods, and multivariate approaches for analysis of the GE interaction (Flores et al., 1998). While the univariate analyses (parametric and nonparametric) attempt to define GE interaction by one or two parameters, the objective of the third strategy (multivariate statistics) is to explore multidirectionality aspects of the GE interaction and to attempt to extract additional information out of this component. Methods, such as principal coordinate analysis (Westcott, 1986) the additive main effect and multiplicative interaction (AMMI) model (Gauch, 1988; Zobel et al., 1988), and sites regression (Cornelius et al., 1996), have been used in analyzing GE interaction.

Usually, a large number of genotypes are tested across several locations and years in MET. It is often difficult to determine the pattern of genotypic responses across environments. The biplot technique provides a powerful solution to this problem. Biplots are useful in summarizing patterns of response that exist in the original data. The concept of biplot was developed by Gabriel (1971) to graphically display two-way data.

For cultivar evaluation, both G and GE must be considered simultaneously (Yan et al., 2000; Yan and Kang, 2003). A G + GE (GGE) biplot was shown to effectively identify the GE interaction pattern of the data and to show clearly which genotype won in which environments. In addition, the GGE biplot is useful in selecting superior genotypes and test environments for a given megaenvironment, thst is, a group of locations that consistently share the same best genotype or genotypes (Yan and Kang, 2003).

Lentil is an annual legume better adapted to cool climates and is traditionally grown as a rainfed crop in the Middle East. Lentil is one of the most important food crops in developing countries, such as Iran, and its seed is a rich source of quality protein in human diets in the arid and semiarid areas of west Asia. Iran has been a leader in lentil-breeding efforts in recent years with the support of the International Center for Agricultural Research in Dry Areas (ICARDA). The major objective of these programs has been to increase the genetic yield potential of this crop. Improved cultivars substantially contribute to increased lentil production. Lentil yields, however, in most production regions seem to be no more than one-half of the potential yields of the cultivars (Sabaghpour et al., 2004). This difference reflects powerful production constraints that prevent the true genetic potential of the varieties to be realized. Ensuring the stability of high yield under unfavorable environmental conditions is a major problem in developing improved crop cultivars. The large yield variation due to location, which is irrelevant to cultivar evaluation and mega environment investigation (Gauch and Zobel, 1996), justifies the selection of multivariate procedures such as SREG procedures (Yan et al., 2000). One of the most important proposes in analysis of MET data is megaenvironment identification. The "which-won-where" view of the GGE biplot is an effective tool for this target (Yan et al., 2007). The other important purpose in analysis of MET data is identification of suitable genotypes with both high mean performance and high stability within a megaenvironment (Yan et al., 2007). Thus, the objective of the present investigation was to investigate the GE interaction in a lentil multienvironmental trials using GGE biplot technique (Yan et al., 2000) and to identify megaenvironments for lentil production in Iran.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Plant Material
Eleven lentil genotypes, including nine improved lines and two check cultivars (Kermanshah and Gachsaran), were studied. The check cultivars were the local landraces of Kermanshah and Gachsaran areas. The improved lines were obtained from ICARDA. A further description of these lines is given in Table 1 .


View this table:
[in this window]
[in a new window]

  		Table 1. Mean yield and origin of 11 lentil genotypes studied in seven different locations in Iran.

 
Experimental Locations
The MET were conducted at seven rainfed locations (Gachsaran, Gorgan, Ilam, Kermanshah, Lorestan, Shirvan, and Qazvin) during 3 yr (2002–2004) but in the Qazvin location only in 2002 to 2003. These locations are the main lentil-producing areas in Iran. Shirvan and Gorgan, in the northeastern portion of Iran, are characterized by semiarid conditions and have sandy loam soils. Qazvin in the northwest is characterized by semiarid conditions, but some supplemental irrigation water is applied during dry periods. That location has a complex clay loam soil series of clay loam. Kermanshah, Lorestan, and Ilam (western Iran) represent moderate rainfall areas and have silt loam soils. Gachsaran, in southern Iran, represents a relatively arid area and has a silt loam soil. A detailed description of these test locations is given in Table 2 .


View this table:
[in this window]
[in a new window]

  		Table 2. Agroclimatic characteristics of test environments

 
Culture and Experimental Design
Genotypes were grown in a randomized complete-block design with four replications at each location. Sowing was done by hand. Plot size was 4 m2, 4 m long, four rows and 25 cm between rows, and plants were spaced 10 cm apart within rows. The area harvested was 1.75 m2, that is, only 3.5 m plot length of the middle two rows was harvested. Seed yield was measured at physiological maturity.

Statistical Analysis
Environment-centered matrix, containing the GGE data, was subjected to singular value decomposition (SVD); each element in the matrix was estimated using the following equation:

Formula
where E(Yij) is the expectation of genotype i in environment j; µ is the general mean; βj represents the environment main effect; K is the number of principal components (PC) needed to provide an adequate description of G + GE; {lambda}k is a proportionality constant or singular value for the kth PC (PCk); and {gamma}ik and {delta}jk are the ith genotype score and the jth environmental score, respectively, for PCk. SVD was achieved by providing a scaling factor f to obtain alternative genotype (nik = {lambda}kf{lambda}ik) and environment (mjk = {lambda}kf–1{delta}jk) scores.

The SVD allowed G x E table of means to be displayed in a plot having n points for the genotypes plus m points for the environments. We chose the most straightforward scaling, that is, symmetric scaling (f = 0.5) (Yan, 2002). The statistical theory of this method has been described in detail by Yan and Kang (2003). All biplots presented in this paper were generated using the software GGE biplot package that runs in a Windows environment (Yan, 2001).


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
Analysis of Variance
Variance components for L, G, and GL interaction based on the yearly data showed relative magnitudes of the G, L, and GL interaction (Table 3 ). The L was always the most important source of yield variation (relative to G) accounting for 63.4 to 68.8% of the total variance (G + L + GL), except in 2004 when it accounted for 52.1%. When the GGE model was fitted, the first two PCs explained 76% (PC1 = 61.6% and PC2 = 14.6%) of GGE variation for lentil MET (Table 4 ). For this study F-Gollob (Gollob, 1968) was used to test significance of PCs for the SREG model.


View this table:
[in this window]
[in a new window]

  		Table 3. Genotype (G), location (L), and genotype x location (GL) variance terms for lentil multienvironmental trials, 2002 to 2004.

 

View this table:
[in this window]
[in a new window]

  		Table 4. Site regression (SREG) analysis of variance for grain yield of 11 lentil genotypes

 
The magnitude of GE interactions for grain yield of 11 lentil genotypes tested across seven locations in Iran were larger than that of G main effect, but smaller than that of E main effect (Table 3). The genotypes studied exhibited both crossover and noncrossover types of GE interaction. The former substantially led to differential rankings of genotypes across test environments, thereby making genotypic selection difficult for the rain-fed conditions of Iran. The relative contributions of G and GE interaction effects to the total variation for grain yield found in this study are similar to those found in other crop adaptation studies in rain-fed environments (Alagarswamy and Chandra, 1998; Cooper et al., 1999; Berteroa et al., 2004). This suggests that it would be very difficult to achieve an indirect response to selection over all of the lentil target population of environments from selection in a few environments, ignoring the observed GE interactions. On the other hand, GE interaction that makes it difficult to select the best performing and most stable genotypes is an important consideration in plant breeding programs because it reduces the progress from selection in any one environment (Hill, 1975; Yau, 1995). Results of analyses of variance for the yearly data showed the large yield variation due to location which is irrelevant to cultivar evaluation and megaenvironment investigation (Gauch and Zobel, 1996), justified the selection of SREG (Yan et al., 2000) as an appropriate model for analyzing the MET data of this research.

Megaenvironments Identification
The genotypes that were farthest from the GGE biplot origin (G4, G7, G9, and G10) served as corners of a polygon when these markers were connected with straight lines. The lines that started from the biplot origin and were perpendicular to the sides of the polygon delimited the five sectors (Fig. 1 ). These genotypes were the best in the environments that are in their respective sectors. Therefore, genotype G9 was the highest performer in environments: IL, KE, LO, and QA (Megaenvironment I). Genotypes G7 did not give the highest yield in any of the environments. Genotype G10 gave the highest performance in environments GO and GA (Megaenvironment II) and genotype G4 gave the highest performance in environment SH (Megaenvironment III). Therefore, Fig. 1 suggests that there exist three possible lentil megaenvironments in Iran. However, this megaenvironment pattern needs verification through other multienvironment trials for this target region.


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 1. Site regression (SREG) biplot identification of winning genotypes and their megaenvironments. Eleven lentil genotypes grown in seven locations.

 
As discussed by Yan (2002), the above inferences about polygon view patterns are mostly, but not totally, validated from the original data (data has not shown). For example, it is not G4, but G8, which has the highest yield in the SH megaenvironment according to original data. This realization originates from the incomplete fitting of the GGE model to the original data. However, the model outcome is worthwhile for recommendation purposes since, as first demonstrated by Stein (1956) and later applied to GE modeling by Gauch and Zobel (1988).

Mean Yield and Stability of Genotypes and Ideal Genotype
The mean yield and stability effects of the genotypes were examined by defining an average tester coordinate (ATC). The average (virtual) environment is indicated by a circle and shows the positive end of the ATC x axis (Fig. 2 ). The average yield of the genotypes is approximated by the projections of their markers on the ATC x axis. In this study, the length of the average environment vector was sufficient to select genotypes based on yield mean performances. Genotypes with above-average means (i.e., from G5 to G2) was selected, whereas the rest were discarded. Genotype G5 was the most stable genotype as well as high yielding. Conversely, G10 was the least stable genotype (variable performance) but high yielding. In addition to G10, the performance of genotypes G9 and G7 were also most variable (least stable), whereas genotypes G5 and G11 were highly stable. G11 is a landrace with low yield. Our results confirmed that genotype G5 has high stability as same as landrace G11 and high yield performance, therefore is introduced as the more favorable genotype. The requirement for the use of SREG-based GGE biplots in the identification of superior genotypes is to facilitate the identification of such genotypes (Crossa et al., 2002). The study has clearly shown that the SREG model can analyze patterns and relationships of genotypes and environments successfully as well as provide a valuable prediction. Becker and Leon (1988), however, stated that multivariate methods are too sophisticated to provide a simple measure of yield stability.


Figure 2
View larger version (15K):
[in this window]
[in a new window]

  		Fig. 2. Site regression (SREG) biplot of mean and stability of 11 lentil genotypes for yield and specific genotype x environment interactions.

 
Ideal genotypes are those that should have large PC1 scores (high mean yield) and small (absolute) PC2 scores (high stability) (Yan and Rajcan, 2002). In this study, genotype G5 is the best of all genotypes, followed by G2 and G10 (Fig. 3 ). These genotypes had the high mean yield performance among all genotypes (Table 1) and so it seems that Ideal genotype procedure is identifying high yielding genotypes as the most stable ones. The PC1 and PC2 scores obtained from SREG analysis that represent the G yield and stability respectively are comparable to the G effect (yield) and adaptability parameter (regression coefficient of Finlay and Wilkinson 1963). The relative contributions of stability and grain yield to the identification of desirable genotype found in this study by Ideal genotype procedure of GGE biplot are similar to those found in other crop stability studies in maize (Zea mays L.) (Fan et al., 2007), barley (Hordeum vulgare L.) (Dehghani et al., 2006), wheat (Triticum aestivum L.) (Kaya et al., 2006), and rice (Oryza sativa L.) (Samonte et al., 2005).


Figure 3
View larger version (18K):
[in this window]
[in a new window]

  		Fig. 3. Site regression (SREG) biplot of ideal genotype and comparison of the genotypes with the ideal genotype

 

   ACKNOWLEDGMENTS
 
We wish to thank Dr. W. Yan for making available the figures of GGE biplot, sending his valuable article and Prof. Hugh Gauch for his valuable guidance. Contributions of the cooperators of the Iran Lentil Performance Trials are also gratefully acknowledged. We thank anonymous reviewers for their helpful comments, suggestions, and corrections of the manuscripts.

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.


   REFERENCES
TOP
 NOTES
 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 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 Google Scholar
Google Scholar
Right arrow Articles by Sabaghnia, N.
Right arrow Articles by Sabaghpour, S. H.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sabaghnia, N.
Right arrow Articles by Sabaghpour, S. H.
Agricola
Right arrow Articles by Sabaghnia, N.
Right arrow Articles by Sabaghpour, S. H.
Related Collections
Right arrow Plant and Environment Interactions
Right arrow Statistics

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Crop Science 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