Crop Science Grow Your Career with CSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
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 HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Signor, C.E.-L.
Right arrow Articles by Charcosset, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Signor, C.E.-L.
Right arrow Articles by Charcosset, A.
Agricola
Right arrow Articles by Signor, C.E.-L.
Right arrow Articles by Charcosset, A.
Related Collections
Right arrow Maize
Right arrow Plant and Environment Interactions
Crop Science 41:663-669 (2001)
© 2001 Crop Science Society of America

CROP BREEDING, GENETICS & CYTOLOGY

Interpretation of Genotype x Environment Interactions for Early Maize Hybrids over 12 Years

C.Epinat-Le Signora, S. Doussea, J. Lorgeoub, J.-B. Denisc, R. Bonhommed, P. Caroloe and A. Charcosset*a

a INRA-INAPG-UPS, Station de Génétique Végétale, Ferme du Moulon, F91190 Gif sur Yvette
b Association Générale des Producteurs de Maïs, Station expérimentale, F91720 Boigneville
c INRA, Station de Biométrie, Route de Saint-Cyr, F78026 Versailles
d INRA, Unité de recherche en Bioclimatologie, F78850 Thiverval-Grignon
e Rustica-Prograin Génétique, 117 avenue de Vendôme, F41000 BLOIS

* Corresponding author (charcos{at}moulon.inra.fr)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 
Genotype x environment interaction was investigated for grain yield of early maize (Zea mays L.) hybrids. Data were obtained from the French Association Générale des Producteurs de Maïs trial network and included 132 hybrids and 229 environments over 12 yr, following an unbalanced design. Analysis of genotype x environment interaction was done for the 1-yr data sets, for the two successive years data sets, and for the 12-yr data set. The magnitude of genotype x environment interaction variance was equal to, or greater than the genotypic variance. Interaction effect was modeled by factorial regression analysis using additional genotypic and environmental information. Genotypic covariates considered were the sum of growing day degrees (GDD) necessary from sowing to flowering and the GDD necessary from flowering to maturity. Environmental covariates were the mean temperature from sowing to the 12 leaf stage, the mean temperature from the 12 leaf stage to the end of the linear grain-filling stage, the water balance around flowering, and the sum of solar radiation around flowering. These six covariates explained about 40% of the interaction effect in all analyses, with equal contribution of genotypic variates (20%) and environmental variates (20%). Flowering earliness of hybrids, water balance around flowering, and mean temperature from the 12 leaf stage to the end of the grain filling phase were determinants of genotype x environment interaction for grain yield in the considered area. A biological interpretation of the interaction was attempted through examination of the regression parameters.

Abbreviations: AGPM, Association Générale des Producteurs de Maïs, France • AMMI, Additive Main effects and Multiplicative Interaction analysis • GDD, sum of Growing Day Degrees • GDDs_f, GDD from sowing to flowering • GDDf_m, GDD from flowering to maturity • GE, genotype x environment • Mg ha-1, ton per hectare • RSD, Residual Standard Deviation • SRf, sum of radiation around flowering (from 06-20 to 08–20) • SS, Sum of Squares • TMs_12l, mean temperature from sowing to 500 GDD (12 leaves stage) • TM12l_e, mean temperature from 500 GDD to 1425 GDD (end of linear grain-filling phase) • WATf, water balance around flowering (rainfall + irrigation – evapotranspiration from 06–20 to 08–20)


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 
NEWLY REGISTERED CULTIVARS generally need to be tested at many locations and for several years before being recommended for a given zone. To achieve this goal, multi-environment trials form the core of varietal testing programs in many countries. These programs have to face the recurring problem of genotype x environment (GE) interactions. Indeed, differential genotypic responses to variable environmental conditions, especially when associated with changes in genotypic ranking, limit the identification of superior, stable hybrids. The GE interactions are as much a function of the environmental variables as a function of the genotypic, morphological, phenological, and physiological traits of the varieties (Nachit et al., 1992). Identification of causal factors of the GE effect and quantification of unexplained variation are of prime importance for selecting for stability or to recommend environmentally specific varieties. During recent decades, new developments have been achieved in crop physiology, agronomy, and statistics and some integrated approaches appeared for GE interactions evaluation (Brancourt-Hulmel, 1999). Many fixed or mixed models have been used for detecting and characterizing GE interaction (van Eeujwick, 1995a,b; Yan and Hunt, 1998; Vargas et al., 1999).

Until now, there have been few attempts to analyze this interaction for the newly registered varieties of maize over an important series of years. Only van Eeujwick et al. (1995b) reported results concerning maize multi-environment trials over a series of 11 yr but they studied forage percent dry-matter content and not yield. Little is known about the most relevant environmental or genotypic variables to consider for maize grain culture. Kang and Gorman (1989) studied the influence of several environmental factors on the interaction effect and concluded that none of them significantly affected interaction for corn yield. The aim of this study was to quantify and model GE interaction for maize grain yield in northern France over 12 yr, from 1986 to 1997. The efficiency and ability of our model to enhance biological interpretation of the GE interaction will be discussed.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 
Experimental Data
Data used for this study were collected within the early maize hybrids trial network of AGPM (Association Générale des Producteurs de Maïs, France) from 1986 to 1997. Each year, the trials were conducted in about 30 locations chosen in northern France. Within each location in a given year, varieties were planted following a randomized complete block design with three replicates. Complete procedures for the tests were described in the annual reports of AGPM (AGPM, 1986–1997). Entries of the trials (about 25) consisted of checks (four to six reference varieties), about 15 first-year and about five second-year entries. Since 10 first-year entries out of 15 were eliminated on the basis of low performance, about 10 varieties (five second-year entries + reference varieties) were common to two successive years and only the reference varieties were therefore common to three or more years. Some trial locations within the northern part of France varied from one year to the other.

This study focuses on grain yield (in Mg ha-1) at 150 g kg-1 of moisture. For each location, data available in the AGPM database were the average values of the varieties over the three blocks and the corresponding residual error. Trials were selected on the basis of their residual standard deviation (RSD) and coefficient of variation (CV) of grain yield: only trials with RSD less than 0.7 Mg ha-1 and CV less than 10% were retained. Eliminated trials lacked precision or had a low average yield (mean yield less than 5 Mg ha-1) revealing poor conditions (for example, because of strong drought stress, high lodging, pathogen attack, etc.). About 20 trials out of 30 were retained each year on the basis of these criteria. Considering all selected trials, the average residual error variance associated with the mean performance of an hybrid in a given trial was estimated as 0.0968 (Mg ha-1)2. The total data set considered includes 132 hybrids and 238 trials. Because of the objectives of the network, the percentage of missing GE combinations was 85%.

Environmental and Genotypic Covariates
An environment was defined as a combination of a year and a region. A region was defined as an area with homogeneous pedologic and climatic conditions and, when useful, presence versus absence of irrigation (about 15% of the considered trials were irrigated). About 60 regions were defined in the considered zone. When more than one trial was available in a given environment, the average values of agronomic data were considered. The final data set included 229 environments (see their partition across years in Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Variance decomposition for individual years, according to Model [1]. Number of varieties and regions are indicated for each year.

 
Climatic measures such as daily air temperature (°C), radiation (MJ m-2), rainfall (mm), Penman evapotranspiration (mm), and irrigation (dates and quantities, mm) were available for each environment. Average daily temperature was estimated over key phases of maize culture: from sowing to 12 visible leaves stage (leaf area development) and from 12 visible leaves stage to the end of the linear grain-filling phase (grain development). The average 12 visible leaves stage of the considered germplasm was determined as the date when the GDD from sowing reached 500. The 12-leaf stage corresponds roughly to both maximum leaf area index and ear differentiation. The end of the linear grain-filling phase was determined as the date when the GDD from sowing reached 1425. The maize culture is very susceptible to environmental conditions during flowering, ear growth, kernel set, and the beginning of the linear grain-filling phase, particularly to solar radiation (Kiniry and Ritchie, 1985; Otegui and Bonhomme, 1998) and water availability (Hall et al., 1981; Westgate and Boyer, 1986). Water balance and solar radiation were therefore calculated over a period flanking flowering stage of early hybrids, i.e., from 20 June to 20 August. Water balance (WAT) was defined as rainfall + irrigation - evapotranspiration. The four environmental covariates retained were the mean temperature from sowing to the 12-leaf stage (TMs_12l), the mean temperature from the 12-leaf stage to the end of the linear grain filling phase (TM12l_e), the sum of daily radiation around flowering (SRf), and the water balance around flowering (WATf).

Earliness of varieties is a key adaptation factor for northern Europe and was considered as the most relevant genotype characteristic. It depends on cycle length and attempts to distinguish the flowering phase from the post-flowering phase to reach maturity (Derieux and Bonhomme, 1982a, b). Physiological maturity was defined as 320 g kg-1 grain moisture. Two genotypic covariates were used in this study: GDD necessary for a given variety to reach flowering after sowing (GDDs_f) and GDD necessary to reach physiological maturity following flowering (GDDf_m). The GDD are calculated with 6°C as base temperature and 30°C as maximal temperature (Gilmore and Rogers, 1958). The average values of these two covariates over the informative trials, for each genotype, were used in modeling.

Statistical Approaches
Variance Components Estimation
An analysis of variance model was applied to the grain yield data by the General Linear Models (GLM) and the REML method of Varcomp procedures of SAS (SAS Institute, 1989). The estimate of the residual experimental error variance {epsilon}ij (0.0968 (Mg ha-1)2) was a priori included in the analysis for estimating the variance components and testing the interaction effect. The complete interaction model was

(1)
where {alpha}i is the effect of Genotype i, ßj is the effect of Environment j, ({alpha}ß)ij is the GE interaction effect, and {epsilon}ij is the residual experimental error. All effects were considered as random following van Eeuwijk et al. (1995a,b) and Frensham et al. (1998). We consider that the reference population for hybrids was that of hybrids submitted to registration. The reference population for locations was all possible locations within northern France (target population of environment according to Podlich et al., 1999).

For data sets involving more than one year (2-yr and 12-yr studies), environmental effect could be divided into a year effect, a region effect, and a year x region interaction effect. Numerous components of variance should be estimated with, in some cases, very few of degree of freedom due to the unbalanced nature of the data sets. Furthermore, because of nonindependence of mean squares, variance ratios are no longer exactly F-distributed, while the hypotheses being tested depend on the structure of the data (Searle, 1971). The analysis of variance has been used on these data sets as an exploratory tool to get a rough idea of the distribution of the variation over the various sources (results not shown).

Modeling of GE Interaction Effect with Factorial Regression
According to the factorial regression model, the complete interaction model could be partitioned as (Denis, 1980, 1988):

(2)
where {rho}k is the regression coefficient of genotypic covariate Gik, {alpha}i is the residual genotype main effect, {delta}h is the regression coefficient of environmental covariate Ejh, ßj is the residual environment main effect, {theta}kh is the regression coefficient of the cross-product of covariates Gik and Ejh, {alpha}'ih is the Genotype i specific regression coefficient of environmental covariate Ejh, ß'jk is the Environment j specific regression coefficient of genotypic covariate Gik, and {epsilon}ij is the residual interaction effect. All the parameters of Model [2] were considered as fixed.

The factorial regression analysis was performed using the INTERA package (Decoux and Denis, 1991). After standardization, covariates were always introduced in the following same order: GDDs_f, GDDf_m for the genotypic covariates and WATf, TM12l_e, TMs_12l, SRf for the environmental covariates. The order was based on the order of appearance of covariates in a stepwise regression process realized on the 12-yr data. It must be emphasized that the sum of squares explained by a given term of the model is adjusted for the term(s) previously introduced (sequential or type I sum of squares). All terms in Model [2] were tested with the experimental residual error [0.0968 (Mg ha-1)2].


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSION
 REFERENCES
 
Variances Estimation
Considering single year data sets, the components of variance due to varieties, regions, and varieties x regions were highly significant (Table 1). Estimates of variety and variety x region variance components were comparable (each about 10% of the total phenotypic variance defined as region + genetic + GE + residual variances). The environmental variance represents, on the average, 80% of the phenotypic variance. The 2-yr and 12-yr (results not shown) analyses showed on average comparable results with a tendency towards a higher relative magnitude of GE effect. For these data sets, about 80% of the total variance was explained by environment differences, about 11% by GE differences, 6% by variety differences, and 3% by residual experimental error.

May and Kozub (1995) found somewhat different results for a data set of barley (Hordeum vulgare L.) genotypes tested in the Western Cooperative Two-Row barley Registration Test from 1987 to 1993. When selecting the few varieties common to 2 yr, they found that GE interaction was small in comparison to the genotypic effect. Van Eeuwijk et al. (1995b) studied silage dry matter content data from the Dutch maize variety trials with 18 selected varieties evaluated from 1980 to 1990 at four locations. They also found that variety x environment interaction was small in comparison to the variety main effect.

Besides differences in species, traits, and countries (climatic conditions in the Netherlands being more homogeneous than in France), the difference between our results and previous results from the literature may be partly explained by the structure of the data sets that were considered and the selection of the varieties. The study of van Eeuwijk et al. (1995b) was restricted to reference varieties selected for high stability.

Modeling of the GE Interaction Effect with Factorial Regression
The average proportion of the environmental main effect explained by the four environmental covariates was 35% for the 1-yr analysis (Table 2), 30% for the 2-yr analysis, and 22% for the 12-yr analysis (Table 3). In the 1-yr analyses, the mean temperature in the second part of the cycle (TM12l_e) and water balance around flowering (WATf) were the most important (each explaining about 10.5%, on average), followed by mean temperature in the first part of the cycle (TMs_12l; 7% on average) and radiation around flowering (SRf; about 5% on average). This trend is similar to the 2-yr analysis: WATf and TM12l_e explained each about 10% of the environmental effect, TMs_12l and SRf explained about 5% each. In the 12-yr analysis, the most important covariate was WATf (10%), then TM12l_e (6%), TMs_12l (3.5%), and SRf (2.5%). The examination of the regression coefficients associated with each environmental covariate ({delta}h) shows that high values of water balance, mean temperatures, and solar radiation were generally favorable to maize growth (data not shown). However, the effect of water balance was negative for years 1987 and 1992 indicating that environments with the lowest water balance were also the most productive. This can be explained by the high precipitation values observed for these two years (Table 4). When water is not limiting, environments with the lowest water balance values (and therefore the smallest rainfall) are likely favored by a high temperature during the second part of the cycle and high solar radiation around flowering.


View this table:
[in this window]
[in a new window]
  		Table 2. Factorial regression analysis (Model [2]) for individual years, including four environmental covariates: WATf, TM12l_e, TMs_12l, SRf and two genotypic covariates: GDDs_f and GDDf_m. Decomposition of main effects (environment and variety) as well as decomposition of the interaction effect (GE) are presented. Degrees of freedom (DF), percentage of SS within the corresponding main or interaction effect (%SS), mean squares [(MS, 100 Mg ha-1)2] and significance of F test are indicated. In each column, the rows of the table are successively adjusted for the preceding ones.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Factorial regression analysis (Model [2]) for pairs of years and for the 12 yr, including four environmental factors: WATf, TM12l_e, TMs_12l, SRf and two genotypic covariates: GDDs_f and GDDf_m. Decomposition of main effects (environment and variety) as well as decomposition of the interaction effect (GE) are presented. Degrees of freedom (DF), percentage of SS within the corresponding main or interaction effect (%SS), mean squares [MS, 100 (Mg ha-1)2] and significance of F test are indicated. In each column, the lines of the table are successively adjusted for the preceding ones.{dagger}

 

View this table:
[in this window]
[in a new window]
  		Table 4. Annual average, minimum, and maximum values of the sum of Growing Day Degrees (GDD) from 25 April to 31 October and sum of rainfall (mm) from 10 June to 31 August for the early maize zone in France.

 
The average proportion of the variety main effect explained by the two genotypic covariates (GDDs_f and GDDf_m) was about 20% for the three analyses. In all three analyses, the two covariates each explained about 10% of the variety effect. Examination of the {rho}k regression coefficients associated with genotypic covariates clearly showed a trend across the 12 yr. From 1986 to 1991, the two coefficients were positive, indicating that late flowering and late maturing genotypes were also the most productive. For some years from 1992 to 1995, the two regression coefficients became negative, indicating that early flowering and early maturing genotypes were the most productive. Since 1996, a positive coefficient appeared for GDDs_f and a negative one for GDDf_m, indicating that genotypes that flower late but mature early were the most favorable. This could be an illustration of the evolution of the variety type across these 12 yr, with the recent appearance of late flowering genotypes with a faster rate of maturing.

When considering the interaction, the eight genotypic x environmental covariates cross-products (Gik {theta}khEjh in Model [2]) were pooled in Tables 2 and 3 for the sake of presentation simplicity and detailed in Table 5. The sum of cross-product terms were significant for 6 yr, for 10 pairs of years, and for the 12-yr data sets. They accounted on average for 3.9% of the interaction effect sum of squares in the 1-yr analyses, 2.7% in the 2-yr analyses, and 1.4% for the 12-yr analysis. The most meaningful cross-products were GDDs_f x WATf (0.86, 0.72 and 0.61% on average of interaction SS in the 1-, 2-, and 12-yr analyses) and GDDf_m x TMs_12l (1.03, 0.38, and 0.25% on average of interaction SS in the 1-, 2-, and 12-yr analyses). To develop a biological explanation of the interaction (presented below), the 12-yr cross-products of covariates effects were considered. This data set shows the best accuracy for the coefficients {theta}kh because they are estimated with a large range of environments and genotypes (see discussion below).


View this table:
[in this window]
[in a new window]
  		Table 5. Decomposition of cross-products effects of covariates (Model [2]) for years, pairs of years and for the 12 yr. Degrees of freedom all equal 1, percentage of SS within the interaction effect (%SS) and significance of F test is indicated.{dagger}

 
The decomposition of the interaction effect through the six regression terms {alpha}'ihEjh and ß'ikGik accounted on average for 37% of the interaction effect sum of squares in the 1-yr analyses, 15% being explained by genotypic covariates and 22% by environmental covariates (Table 2). These proportions were on average 35, 16, and 19% for the 2-yr analyses, respectively. These proportions were 34, 18, and 16% for the 12-yr analysis, respectively (Table 3). The percentage of the interaction explained by the two groups of covariates was comparable, indicating a better efficiency for the two genotypic covariates, which explained the same percentage of interaction sum of squares as environmental covariates with fewer degrees of freedom.

From the interaction of the genotypic (respectively environment) covariate with the residual environmental (respectively genotypic) variation, it appears from Tables 2 and 3, that earliness of flowering (GDDs_f x env), water balance (WATf x var), and mean temperature in the second part of the cycle (TM12l_e x var) were the most explicative of the GE interaction for early hybrids in the trial network of AGPM. The genotype covariate GDDs_f was as explicative as the two environmental covariates, WATf and TM12l_e, together. These results are in agreement with those of Argillier et al. (1994) who studied 40 hybrids in 21 environments over 2 yr, 1992 and 1993. They showed that GE interaction effect for silage yield in maize could be due to a large extent to earliness effects and yield limiting factors such as lodging susceptibility and water stress. Indeed for years 1992 and 1993, the most explicative covariates in our study were temperature for flowering (GDDs_f) and water balance (WATf). Van Eeuwijk et al. (1995b) reported comparable results using factorial regression modeling of the variety x year interaction on their complete data set of 18 varieties over 10 yr. They selected four environmental covariates: number of days with mean temperature below 10°C in May, mean temperature over the period July-August, total radiation during the growing season (May–August), and the mean dry-matter content (DMC) by environment. The four covariates accounted for 52% of the variety x year interaction. Excluding the mean DMC by environment, the three environmental covariates accounted for 33% of the interaction effect.

In contrast to these results, Kang and Gorman (1989) found that factorial regression was unable to explain GE interaction for grain yield when studying 17 maize hybrids grown in 12 environments (4 locations x 3 yr). Rainfall during the season and preseason rainfall were the most explicative covariates, although their individual contribution did not exceed 1.5% of the interaction sum of squares.

Biological Interpretation of Interaction Effect
Factorial regression provides a good basis for the biological interpretation of the interaction effect. Analysis of the highly significant cross-products in the 12-yr data set allows the identification of general trends in the combined effect of both genotypic and environmental covariates on the interaction. The most explicative cross-product was GDDs_f x WATf (Table 5), thus confirming a major contribution of flowering earliness and water supply to the interaction effect. The positive regression parameter associated with this cross-product ({theta}11 = 0.11) indicates that the interaction effect of early varieties increases (i.e., becomes favorable) as the water supply of the environment decreases. Consequently, the interaction effect of late varieties increases as the water supply of the environment increases. The advantage of early varieties in stressed conditions can be explained by an escape strategy since they avoid water deprivation near flowering and at the end of the crop cycle (Hall et al., 1981). In addition, total leaf area of early varieties is lower than that of late varieties, which limits evapotranspiration and is therefore favorable in the case of restricted water supply (Shaw, 1988).

The second most explicative cross-product was GDDf_m x SRf (Table 4). The negative regression coefficient cross-product [{theta}24 = -0.01) indicates that the interaction effect of late maturing varieties increases as solar radiation around flowering decreases. As a consequence, the interaction effect of early maturing varieties increases as solar radiation increases. This can be interpreted as an advantage of a long grain-filling period, associated with a long light interception period, in the case of limited average daily radiation (Muchow et al., 1990).

Biological interpretation of the two other significant cross-products (GDDf_m x TMs_12l and GDDf_m x TM12l_e) is more complicated. Indeed the interaction effect of early maturing varieties decreases as the mean temperatures in both cycle phases decreases ({theta}23 = -0.15 and {theta}22 = -0.02). This is unexpected because of the well-known adaptation of early varieties to low mean temperatures. However, this can be interpreted as an indirect effect of the association of high temperatures with high solar radiation and limited water balance, which both favor early varieties.

Despite the significance of the previous cross-product terms and the discussed general tendencies, the six regression terms {alpha}'ihEjh and ß'jkGik were the most explicative terms in Model [4]. This illustrates, in particular, that there are variety-specific responses to environmental covariates, which could not be solely accounted for by differences in earliness.


   CONCLUSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
REFERENCES
 
This study first underlines the magnitude of the environment effect which accounts for about 80% of the total variance of maize grain yield in northern France. Genotype x environment interaction effect also appears important when compared with variety effect. The factorial regression model facilitated the biological interpretations of GE interaction. Earliness of flowering (GDDs_f), water balance (WATf), and mean temperature in the second part of the cycle (TM12l_e) were contributing to the GE interaction for maize grain yield. Flowering earliness was particularly favorable in situations with restricted water supply. This effect appears larger than the adaptation of early varieties to low temperatures. In addition to the effect of earliness, varieties clearly display differential responses towards environmental characteristics. Estimation of variety specific regression terms ({alpha}'ih) adjusted for differences in earliness should be particularly interesting for evaluating the adaptation of varieties to water deprivation. Interaction modeling may be further improved by considering other covariates and (or) non-linear interaction modeling. In particular, site-specific information on, e.g., soil characteristics (fertility or depth) had not been collected for the field trials involved in this study and deserve consideration in the future.

These results also confirm the importance of testing hybrids under representative environmental conditions to identify best the most stable and the highest yielding cultivars. Covering a wide range of situations for water balance, solar radiation, and mean temperatures over key phases of the plant cycle appeared particularly important. Two years of testing over numerous environments, which is the current practice within the AGPM network, increases the probability of reaching this goal more than do 1-yr trials. However, some pairs of years with similar climatic characteristics may not provide sufficiently representative conditions. Examination of environmental covariates allows the detection of such situations, for which additional testing may be advisable.


   ACKNOWLEDGMENTS
 
We are grateful to B. Nicoullaud, A. Gallais and A. Bouchez for helpful discussion. We also thank P. Guellier (AGPM) for computing the covariates used in this study. We are grateful to L. Moreau for critical reading of the manuscript.

Received for publication August 18, 1999.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Agron. J.Home page
T. Sarlangue, F. H. Andrade, P. A. Calvino, and L. C. Purcell
Why Do Maize Hybrids Respond Differently to Variations in Plant Density?
Agron. J., June 5, 2007; 99(4): 984 - 991.
[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]

Home page
 Crop Sci.Home page
R. A. Malvar, P. Revilla, A. Butron, B. Gouesnard, A. Boyat, P. Soengas, A. Alvarez, and A. Ordas
Performance of Crosses among French and Spanish Maize Populations across Environments
Crop Sci., May 6, 2005; 45(3): 1052 - 1057.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
A. Butron, P. Velasco, A. Ordas, and R. A. Malvar
Yield Evaluation of Maize Cultivars across Environments with Different Levels of Pink Stem Borer Infestation
Crop Sci., May 1, 2004; 44(3): 741 - 747.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
E. A. Lee, T. K. Doerksen, and L. W. Kannenberg
Genetic Components of Yield Stability in Maize Breeding Populations
Crop Sci., November 1, 2003; 43(6): 2018 - 2027.
[Abstract] [Full Text] [PDF]

Home page
 Plant Physiol.Home page
M. Reymond, B. Muller, A. Leonardi, A. Charcosset, and F. Tardieu
Combining Quantitative Trait Loci Analysis and an Ecophysiological Model to Analyze the Genetic Variability of the Responses of Maize Leaf Growth to Temperature and Water Deficit
Plant Physiology, February 1, 2003; 131(2): 664 - 675.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
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 HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Signor, C.E.-L.
Right arrow Articles by Charcosset, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Signor, C.E.-L.
Right arrow Articles by Charcosset, A.
Agricola
Right arrow Articles by Signor, C.E.-L.
Right arrow Articles by Charcosset, A.
Related Collections
Right arrow Maize
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 Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome