Finlay-Wilkinson Regression Research Articles

Factor-Analytic Variance-Covariance Structures for Prediction Into a Target Population of Environments.

Finlay-Wilkinson regression is a popular method for modeling genotype-environment interaction in plant breeding and crop variety testing. When environment is a random factor, this model may be cast as a factor-analytic variance-covariance structure, implying a regression on random latent environmental variables. This paper reviews such models with a focus on their use in the analysis of multi-environment trials for the purpose of making predictions in a target population of environments. We investigate the implication of random versus fixed effects assumptions, starting from basic analysis-of-variance models, then moving on to factor-analytic models and considering the transition to models involving observable environmental covariates, which promise to provide more accurate and targeted predictions than models with latent environmental variables.

Kernel number and kernel weight stability can vary across corn hybrids

AbstractThe stability of a trait refers to the extent to which its expression in a given genotype varies across environments. The more stable a trait, the less variable its expression. Grain yield stability is a central consideration in corn production to ensure that hybrids perform consistently across environments and is frequently quantified given its importance. Little attention has been paid to the stability of corn yield components, kernel number per m2 (KN), and kernel weight (KW). Our hypothesis is that while previous research suggests that yield stability of commercial corn hybrids is generally consistent, the stabilities of KN and KW may exhibit significant differences, even when overall yield stability remains constant. This study evaluated the yield and yield component stabilities of 23 commercial corn hybrids conducted on‐farm at five location‐years in Ontario, Canada, using Finlay–Wilkinson regression. Most (61%) hybrids exhibited average yield stability with β1‐values close to 1.0. But seven hybrids displaying average yield stability had KN and/or KW stabilities significantly different than average. While in absolute terms, KW was always more stable than KN across environments, the data indicate that hybrids have different mechanisms to achieve stable yields in terms of relative yield component adjustments. Overall, 14 hybrids had yield component β1‐values significantly more or less stable than average. The instances where yield component β1‐values differed significantly from 1.0 were almost equally divided between KN and KW. These findings support the potential for hybrid‐specific corn management, that is, tailoring management practices to take advantage of hybrid variation in yield component stabilities.

Revisiting superiority and stability metrics of cultivar performances using genomic data: derivations of new estimators

The selection of highly productive genotypes with stable performance across environments is a major challenge of plant breeding programs due to genotype-by-environment (GE) interactions. Over the years, different metrics have been proposed that aim at characterizing the superiority and/or stability of genotype performance across environments. However, these metrics are traditionally estimated using phenotypic values only and are not well suited to an unbalanced design in which genotypes are not observed in all environments. The objective of this research was to propose and evaluate new estimators of the following GE metrics: Ecovalence, Environmental Variance, Finlay–Wilkinson regression coefficient, and Lin–Binns superiority measure. Drawing from a multi-environment genomic prediction model, we derived the best linear unbiased prediction for each GE metric. These derivations included both a squared expectation and a variance term. To assess the effectiveness of our new estimators, we conducted simulations that varied in traits and environment parameters. In our results, new estimators consistently outperformed traditional phenotype-based estimators in terms of accuracy. By incorporating a variance term into our new estimators, in addition to the squared expectation term, we were able to improve the precision of our estimates, particularly for Ecovalence in situations where heritability was low and/or sparseness was high. All methods are implemented in a new R-package: GEmetrics. These genomic-based estimators enable estimating GE metrics in unbalanced designs and predicting GE metrics for new genotypes, which should help improve the selection efficiency of high-performance and stable genotypes across environments.

Extending Finlay–Wilkinson regression with environmental covariates

AbstractFinlay–Wilkinson regression is a popular method for analysing genotype–environment interaction in series of plant breeding and variety trials. It involves a regression on the environmental mean, indexing the productivity of an environment, which is driven by a wide array of environmental factors. Increasingly, it is becoming feasible to characterize environments explicitly using observable environmental covariates. Hence, there is mounting interest to replace the environmental index with an explicit regression on such observable environmental covariates. This paper reviews the development of such methods. The focus is on parsimonious models that allow replacing the environmental index by regression on synthetic environmental covariates formed as linear combinations of a larger number of observable environmental covariates. Two new methods are proposed for obtaining such synthetic covariates, which may be integrated into genotype‐specific regression models, that is, criss‐cross regression and a factor‐analytic approach. The main advantage of such explicit modelling is that predictions can be made also for new environments where trials have not been conducted. A published dataset is employed to illustrate the proposed methods.

Mega-environment analysis to assess adaptability, stability, and genomic predictions in grain sorghum hybrids

Multi-environment trials (MET) are fundamental for assessing genotype-by-environment interaction (GxE) effects, adaptability and stability of genotypes and provide valuable information about target regions. As such, a MET involving grain sorghum hybrid combinations derived from elite inbred lines adapted to diverse sorghum production regions was developed to assess agronomic performance, stability, and genomic-enabled prediction accuracies within mega-environments (ME). Ten females and ten males from the Texas A&M and Kansas State sorghum breeding programs were crossed following a factorial mating scheme to generate 100 hybrids. Grain yield, plant height, and days to anthesis were assessed in a MET consisting of ten environments across Texas and Kansas over two years. Genotype plus Genotype-by-block-of-environment biplot (GGB) assessed ME, while the "mean-vs-stability" view of the biplot and the Bayesian Finlay–Wilkinson regression evaluated hybrid adaptability and stability. A genomic prediction model including the GxE effect was applied within ME to assess prediction accuracy. Results suggest that grain sorghum hybrid combinations involving lines adapted to different target regions can produce superior hybrids. GGB confirmed distinct regions of sorghum adaption in the U.S. Further, genomic predictions within ME reported inconsistent results, suggesting that additional effects rather than the correlations between environments are influencing genomic prediction of grain sorghum hybrids.

Genetic characterization of agronomic traits and grain threshability for organic naked barley in the northern United States

AbstractNaked barley (Hordeum vulgare L.) grain threshes freely from its hull during harvesting and cleaning. Much of the available naked barley germplasm is unadapted to U.S. barley growing regions, and few genotypes have been selected to thrive under organic systems. The goal of this research was to characterize a set of spring naked barley genotypes for agronomic traits in the northern United States under organic conditions and estimate the degree of genotype × environment interaction (GEI). To achieve this goal, a multienvironment trial was conducted over 3 yr. Experiments were grown under organic conditions where traits including grain yield, test weight, plant height, heading date, and threshability were evaluated. Contributions of environment, genotype, and GEI to the phenotypic variance were calculated. In the tested germplasm, test weight, plant height, heading date, and threshability were found to have higher variance attributed to genotypic effects compared to GEI. Grain yield variance attributed to GEI was five times greater than that of genotype alone. Genotype + GEI (GGE) and location‐grouping (LG) biplots showed two sets of environments where naked barley genotypes performed similarly. Sensitivity analysis by Finlay–Wilkinson regression found that grain yield was highly sensitive to changes in environment mean yield. These results show that selection for grain yield should be conducted within mega‐environments to leverage GEI patterns. Heading date, plant height, test weight, and threshability can be selected across mega‐environments.

Genomic Selection and Genome-Wide Association Studies for Grain Protein Content Stability in a Nested Association Mapping Population of Wheat

Grain protein content (GPC) is controlled by complex genetic systems and their interactions and is an important quality determinant for hard spring wheat as it has a positive effect on bread and pasta quality. GPC is variable among genotypes and strongly influenced by the environment. Thus, understanding the genetic control of wheat GPC and identifying genotypes with improved stability is an important breeding goal. The objectives of this research were to identify genetic backgrounds with less variation for GPC across environments and identify quantitative trait loci (QTLs) controlling the stability of GPC. A spring wheat nested association mapping (NAM) population of 650 recombinant inbred lines (RIL) derived from 26 diverse founder parents crossed to one common parent, ‘Berkut’, was phenotyped over three years of field trials (2014–2016). Genomic selection models were developed and compared based on predictions of GPC and GPC stability. After observing variable genetic control of GPC within the NAM population, seven RIL families displaying reduced marker-by-environment interaction were selected based on a stability index derived from a Finlay–Wilkinson regression. A genome-wide association study identified eighteen significant QTLs for GPC stability with a Bonferroni-adjusted p-value < 0.05 using four different models and out of these eighteen QTLs eight were identified by two or more GWAS models simultaneously. This study also demonstrated that genome-wide prediction of GPC with ridge regression best linear unbiased estimates reached up to r = 0.69. Genomic selection can be used to apply selection pressure for GPC and improve genetic gain for GPC.

Characterization of Groundnut (Arachis hypogaea L.) Test Locations Using Representative Testing Environments With Farmer-Preferred Traits.

In this study, the differential rankings of 36 groundnut genotypes under varying environmental conditions were studied at various levels of phenotype. Locations that are generally accepted by the crop- and soil-based research community to represent the entire Guinea and Sudan Savanna agro-ecological zones in Ghana were characterized, this time using a crop. The characterization was done based on four farmer-preferred traits (early and late leaf spot disease ratings, and haulm and pod yields) using three models (i.e., AMMI, GGE, and Finlay–Wilkinson regression). These models were used to capture specific levels of phenotype, namely, genotype-by-environment interaction (GE), genotype main effect plus GE (G+GE), and environment and genotype main effects plus GE (E+G+GE), respectively. The effect of three major environmental covariables was also determined using factorial regression. Location main effect was found to be highly significant (p < 0.001), confirming its importance in cultivar placement. However, unlike genotypes where the best is usually adjudged through statistical ranking, locations are judged against a benchmark, particularly when phenotyping for disease severity. It was also found that the locations represent one complex mega-environment, justifying the need to test new technologies, including genotypes in all of them before they can be approved for adoption nationally. Again, depending on the phenotypic level considered, genotypic rankings may change, causing environmental groupings to change. For instance, all locations clustered to form one group in 2017 for early and late leaf spot diseases and pod yield when GE was considered, but the groupings changed when G+GE was considered for the same traits in the same year. As a result, assessing genotypic performance at the various levels to arrive at a consensus decision is suggested. Genotypes ICGV-IS 141120 and ICGV-IS 13937 were found to be the best performing.

Management duration controls the synergistic effect of tillage, cover crop, and nitrogen rate on cotton yield and yield stability

Resilient agroecosystems are foundational for stable and profitable food, feed, and fiber production in the face of increasing climatic perturbations and environmental stresses. Enhanced soil and environmental benefits of cover crops applied to conservation tillage systems has been well documented. However, little is known about the timespan for interactions of no tillage and cover crops to achieve enhanced yield and yield stability while lowering N fertilization. Using a long-term continuous cotton experiment in southeastern USA, we analyzed yield data collected from 1986 to 2018 from 32 management systems to identify how management duration controls the synergistic effect of applied mineral N rates (0, 34, 67, and 101 kg ha−1), cover crops (no cover [NC], hairy vetch [HV], crimson clover [CC], and winter wheat [WW]), and tillage practices (no tillage [NT] and conventional tillage [CT]) on cotton yield and yield stability. Yield stability was analyzed using Finlay—Wilkinson regression model, Wricke’s Ecovalance, and coefficient of variation, and a mixed model approach was used to compare the yield and yield stability within three time phases (1−10 years, 11–20 years, and 21−33 years) at the 95% confidence level. During the initial 10 yr period (phase 1) CT resulted in greater cotton yield (7%) and yield stability than NT. However, in phase 2 (11–20 years) and phase 3 (21−33 years) NT led to greater yield (7%) and yield stability for almost all cover crop × N interactions, except for zero N following NC and WW. Increased N during the initial phase reduced both yield and yield stability under legume cover crops. During phases 2 and 3, however, the higher N rates (67 and 101 kg N ha−1) increased cotton yield, although management systems with 0 and 101 kg N ha−1 showed the largest temporal yield variability. Legume cover crops increased yield and yield stability under low N rates. The maximum combined yield and yield stability benefit was obtained from HV cover on NT with additional application of 34 kg N ha−1. Our results suggest that after the initial phase NT delivers the most consistent yield benefits while enhancing yield stability against unfavorable environmental conditions. Long-term integration of legume cover crops (particularly HV) to NT systems was effective in maintaining high yield and increasing yield stability while lowering N rates.

Genetic analysis of drupelet count in black raspberry (Rubus occidentalis)

U.S. black raspberry (BR) production is currently limited by narrowly adapted, elite germplasm. Improved understanding of genetic control and genotype-by-environment interactions (GEI) of pomological traits will inform the development of improved BR germplasm and cultivars. To this end, analysis of a multi-environment trial of two BR mapping populations derived from crosses of commercial cultivars with wild accessions is providing insights into genetic variation, quantitative trait loci (QTL), and GEI related to traits of horticultural value. Twenty-six horticultural and fruit quality traits were evaluated in two mapping populations over three years at four distinct locations representative of current U.S. BR production. One fruit quality trait, drupelet count, has been analyzed, and nine additional fruit quality traits are being analyzed. Random factor analysis of genetic effect was used to account for missing data. GEI analysis was performed using a Bayesian approach to Finlay-Wilkinson regression and genotyping-by-sequencing (GBS)-derived genomic relationship matrices, with the purpose of verifying relatedness and identifying genotypes with stable performance over environments. Genetic maps were reconstructed using GBS-derived markers and the latest genome assembly for BR. Among relevant results, pedigree verification led to construction of reliable linkage maps, and a QTL spanning a one to five megabase region containing several hundred genes was identified on chromosome 1. This region has a stable influence over drupelet count over environments (location by year). These and related findings are expected to guide further genetic characterization of BR fruit quality, management of breeding germplasm, and development of improved BR cultivars for U.S. production.

The Genetic Architecture for Phenotypic Plasticity of the Rice Grain Ionome.

The ionome of the rice grain is crucial for the health of populations that consume rice as a staple food. However, the contribution of phenotypic plasticity to the variation of rice grain ionome and the genetic architecture of phenotypic plasticity are poorly understood. In this study, we investigated the rice grain ionome of a rice diversity panel in up to eight environments. A considerable proportion of phenotypic variance can be attributed to phenotypic plasticity. Then, phenotypic plasticity and mean phenotype were quantified using Bayesian Finlay-Wilkinson regression, and a significant correlation between them was observed. However, the genetic architecture of mean phenotype was distinct from that of phenotypic plasticity. Also, the correlation between them was mainly attributed to the phenotypic divergence between rice subspecies. Furthermore, the results of whole-genome regression analysis showed that the genetic loci related to phenotypic plasticity can explain a considerable proportion of the phenotypic variance in some environments, especially for Cd, Cu, Mn, and Zn. Our study not only sheds light on the genetic architecture of phenotypic plasticity of the rice grain ionome but also suggests that the genetic loci which related to phenotypic plasticity are valuable in rice grain ionome improvement breeding.

Genetic variation of phenotypic plasticity in Bangladesh rice germplasm

Active selection of phenotypic plasticity has the potential to increase agricultural crop productivity. We screened 208 Bangladeshi rice genotypes in Bangladesh in 2015 for their phenotypic plasticity with respect to panicle dry weight per hill (PDW) relative to planting geometry, which alters solar radiation captured by leaves due to differences in mutual shading. The plasticity of PDW relative to planting geometry (ratio of panicle dry weight at half density/normal density) varied by genotypes ranging from 0.82 to 2.56 times with an average of 1.59 times. Most of the variation in PDW was explained by the plasticity of total dry weight and panicle number, rather than individual panicle weight and harvest index. We selected 10 genotypes with the best and worst phenotypic plasticity for PDW. These candidate genotypes were tested for their plasticity in six different environments in 2016. The Finlay-Wilkinson regression coefficient for PDW was 1.09 for the best group compared to 0.87 for the worst group although large variation existed within each group. The mean PDW averaged for all environments was 8% greater for the best group compared to the worst one. Two genotypes “Chini sagar” and “Koha binni” with highest plasticity and productivity could be candidate donor varieties for future breeding programs. These results indicated that phenotypic plasticity is a heritable trait, and therefore a useful indicator for improving crop productivity for any expected climate change.

Selection of superior genotype of sweet-potato in Indonesia based on stability and adaptability

Sweet-potato (Ipomoea batatas [L.] Lam.) is the 3rd most important root and tuber plant in the world, after potatoes and cassava. The knowledge of the structure of Genotype × Environment interaction (GEI) could assist in the development of cultivars of sweet-potato. The objective of this research was to select superior sweet-potato genotypes in Indonesia based on Finlay-Wilkinson regression, additive main effect and multiplicative interaction (AMMI) analysis, AMMI stability value (ASV), and Genotype plus Genotype × Environment (GGE) biplot methods. Genetic materials evaluated in this research were 10 Indonesian superior genotypes developed by the Plant Breeding and Seed Technology Lab., Faculty of Agriculture, University of Padjadjaran, Bandung. The research was arranged in randomized block design, with 16 genotypes as treatments. Data analysis of the yield characters was performed by using combined variance analysis: Finlay-Wilkinson regression, ASV, AMMI, and GGE Biplot. The GEI significantly influenced the tested genotypes at three sites. However, they did not show any significant differences in the GEI. Ayamurasaki, Beniazuma, Awachy2, 15(112), Awachy4, Awachy5, 80(109), 54(160), 35(180) were specific adaptation genotypes of marginal land based on Finlay-Wilkinson analysis; whereas, Rancing, AC Putih, Ayamurasaki, Naruto Kinotoki, Awachy1, Awachy2, 15(112), Awachy4, 57(97), 80(109), 54(160), 68(120) and 35(180) performed as stable genotypes based on ASV. Ayamurasaki, Naruto Kinotoki, Awachy1, Awachy2, 57(97), 80(109) and 68(120) were stable genotypes based on AMMI analysis. However, based on GGE Biplot analysis Ayamurasaki, Awachy1, 80 (109) and 68 (120) were stable genotypes. ASV analysis provided effective stable genotype selections of 87.50% greater than other stability methods. Ayamurasaki, Awachy1, 80(109) and 68(120) were stable genotypes based on AMMI, ASV, and GGE biplot analysis.

Mineral NPK and manure fertilisation affecting the yield stability of winter wheat: Results from a long-term field experiment

Abstract It is well known that a balanced nutrient supply ensures high crop yields and stable cropping systems. In comparison, little is known about the influence of different fertilisation regimes on the yield stability of crops. Therefore, the objective of this study was to identify the impact of mineral and organic fertilisation supplied in different combinations (P + K/N + P/N + K/N + P + K) and levels (50%/100%/100% + manure) on the stability of the winter wheat grain yield. A long-term fertilisation field experiment at Giessen University (Germany, temperate climate, silty clay), established in 1956 with a multifactorial set up, was used. For an accurate stability analysis that matches the experimental design, we subjected the data to a mixed model analysis based on the restricted maximum likelihood (REML) method calculating Shukla‘s stability variance and Finlay–Wilkinson regression and performing Eskridge's risk analysis. It was found that without any nutrient supply, the wheat yields exhibited high annual variability and only modest stability. A similar pattern was identified for the fertiliser combination of the minerals P + K. The combinations containing mineral N showed a significant improvement in yield stability in which N + P and N + K are almost identical. In all fertilisation variants, the additional supply of manure stabilised the wheat yields. The mineral supply of N + P + K with additional manure provided the best yield stability and lowest agronomic risk for yield failure.

Root Traits Enhancing Rice Grain Yield under Alternate Wetting and Drying Condition.

Reducing water requirements and lowering environmental footprints require attention to minimize risks to food security. The present study was conducted with the aim to identify appropriate root traits enhancing rice grain yield under alternate wetting and drying conditions (AWD) and identify stable, high-yielding genotypes better suited to the AWD across variable ecosystems. Advanced breeding lines, popular rice varieties and drought-tolerant lines were evaluated in a series of 23 experiments conducted in the Philippines, India, Bangladesh, Nepal and Cambodia in 2015 and 2016. A large variation in grain yield under AWD conditions enabled the selection of high-yielding and stable genotypes across locations, seasons and years. Water savings of 5.7–23.4% were achieved without significant yield penalty across different ecosystems. The mean grain yield of genotypes across locations ranged from 3.5 to 5.6 t/ha and the mean environment grain yields ranged from 3.7 (Cambodia) to 6.6 (India) t/ha. The best-fitting Finlay-Wilkinson regression model identified eight stable genotypes with mean grain yield of more than 5.0 t/ha across locations. Multidimensional preference analysis represented the strong association of root traits (nodal root number, root dry weight at 22 and 30 days after transplanting) with grain yield. The genotype IR14L253 outperformed in terms of root traits and high mean grain yield across seasons and six locations. The 1.0 t/ha yield advantage of IR14L253 over the popular cultivar IR64 under AWD shall encourage farmers to cultivate IR14L253 and also adopt AWD. The results suggest an important role of root architectural traits in term of more number of nodal roots and root dry weight at 10–20 cm depth on 22–30 days after transplanting (DAT) in providing yield stability and preventing yield reduction under AWD compared to continuous flooded conditions. Genotypes possessing increased number of nodal roots provided higher yield over IR64 as well as no yield reduction under AWD compared to flooded irrigation. The identification of appropriate root architecture traits at specific depth and specific growth stage shall help breeding programs develop better rice varieties for AWD conditions.

Finlay-Wilkinson's regression coefficient as a pre-screening criterion for yield responsiveness to elevated atmospheric CO2 concentration in crops.

The rising atmospheric CO2 concentration ([CO2 ]) can increase crop productivity, but there are likely to be intraspecific variations in the response. To meet future world food demand, screening for genotypes with high [CO2 ] responsiveness will be a useful option, but there is no criterion for high [CO2 ] responsiveness. We hypothesized that the Finlay-Wilkinson regression coefficient (RC) (for the relationship between a genotype's yield versus the mean yield of all genotypes in a specific environment) could serve as a pre-screening criterion for identifying genotypes that respond strongly to elevated [CO2 ]. We collected datasets on the yield of 6 rice and 10 soybean genotypes along environmental gradients and compared their responsiveness to elevated [CO2 ] based on the regression coefficients (i.e. the increases of yield per 100 µmol mol-1 [CO2 ]) identified in previous reports. We found significant positive correlations between the RCs and the responsiveness of yield to elevated [CO2 ] in both rice and soybean. This result raises the possibility that the coefficient of the Finlay-Wilkinson relationship could be used as a pre-screening criterion for [CO2 ] responsiveness.

FW: An R Package for Finlay-Wilkinson Regression that Incorporates Genomic/Pedigree Information and Covariance Structures Between Environments.

The Finlay–Wilkinson regression (FW) is a popular method among plant breeders to describe genotype by environment interaction. The standard implementation is a two-step procedure that uses environment (sample) means as covariates in a within-line ordinary least squares (OLS) regression. This procedure can be suboptimal for at least four reasons: (1) in the first step environmental means are typically estimated without considering genetic-by-environment interactions, (2) in the second step uncertainty about the environmental means is ignored, (3) estimation is performed regarding lines and environment as fixed effects, and (4) the procedure does not incorporate genetic (either pedigree-derived or marker-derived) relationships. Su et al. proposed to address these problems using a Bayesian method that allows simultaneous estimation of environmental and genotype parameters, and allows incorporation of pedigree information. In this article we: (1) extend the model presented by Su et al. to allow integration of genomic information [e.g., single nucleotide polymorphism (SNP)] and covariance between environments, (2) present an R package (FW) that implements these methods, and (3) illustrate the use of the package using examples based on real data. The FW R package implements both the two-step OLS method and a full Bayesian approach for Finlay–Wilkinson regression with a very simple interface. Using a real wheat data set we demonstrate that the prediction accuracy of the Bayesian approach is consistently higher than the one achieved by the two-step OLS method.

Genotypic stability of weed competitive ability for bread wheat (Triticum aestivum) genotypes in multiple environments

Weed control in broadacre cropping systems is becoming increasingly difficult owing to widespread evolution of herbicide resistance in major weed species. The importance of crop competition in weed management is often overlooked but it can play an important role in cropping systems. Competitive ability of 86 wheat (Triticum aestivum L.) genotypes varying for early vigour was investigated at two sites over two growing seasons against cultivated oats (Avena sativa L.) as a weed mimic. There were significant (P &lt; 0.001) treatment effects of weed, wheat genotype and weed × genotype interaction at the different sites. Mature crop height and early crop vigour were strongly correlated with improved weed suppression and tolerance. Negative correlation between early vigour (normalised difference vegetation index and visual score) and weed-free yield indicates the presence of some yield penalty in high-vigour (HV) lines. Wheat genotypes with high grain yield under weed-free conditions tended to suffer high yield loss from weeds (low tolerance) and allowed greater production of weed seed (low weed suppression). However, many of the HV lines produced significantly higher grain yield than the tested commercial cultivars under weedy conditions. The use of the Finlay–Wilkinson regression approach for assessing cultivar stability revealed a strong association between genotype mean weed suppression and stability across the four environments. Several HV lines showed consistently greater weed suppression than the wheat cultivars investigated. Genotypic variation was much greater for weed suppression than weed tolerance, suggesting greater opportunity for the selection of improved weed suppression in wheat. However, strong positive correlation between weed suppression and tolerance (r = 0.79, P &lt; 0.001) suggests that wheat lines selected on the basis of high weed suppression may also exhibit improved weed tolerance.

Genotype × environment interaction studies highlight the role of phenology in specific adaptation of canola (Brassica napus) to contrasting Mediterranean climates

While genotype (G)×environment (E) interaction (G×E) complicates broad crop adaptation, understanding its causes facilitates breeding for specific adaptation. South-western Australia captures a broad range of Mediterranean climates, from a very warm short season with low rainfall in the north and east to a longer season with high rainfall in the southwest, and provides a unique opportunity to investigate G×E interaction. In this study, we evaluate G×E interaction for seed yield and oil content of canola genotypes with wide ranging phenology across south-western Australia.Environments were separated into year (Y) and location (L) and a factor analytic (FA) model used to partition G×E interactions into G×Y, G×L and G×Y×L across four years (2006–2009). G×E interaction contributed 34% to total variance of seed yield compared with 9% for G. An additive main effects and multiplicative interaction (AMMI) model was used to further evaluate the significance of G×L, and delineate mega-environments (ME) and determine the best performing cultivar in each year. AMMI identified two MEs with different seasonal climates. ME1 combines >330mm seasonal rainfall with a cooler, longer post-anthesis growing period. ME2 is more terminally drought-prone, with higher temperatures and <300mm rainfall, resulting in a short growing season. There were significant crossover yield responses to location changes: the medium flowering genotypes produced significantly higher yield than the early flowering genotypes in ME1 but yielded poorly in ME2, and vice versa. In contrast to yield, G effects were very strong in oil content, accounting for 53% of total variance, compared with 14% for G×Y, and negligible G×L effects. Finlay–Wilkinson regression showed little crossover interaction in oil content with E. The key outcome of this G×E interaction study is the importance of phenology to the adaptation of canola in south-western Australia. Therefore, it is suggested that breeding for specific adaptation to each mega-environment should be targeted with a breeding strategy focusing on drought and heat tolerance in ME2 and high yield potential in ME1.

A comparison between Joint Regression Analysis and the Additive Main and Multiplicative Interaction model: the robustness with increasing amounts of missing data

This paper joins the main properties of joint regression analysis (JRA), a model based on the Finlay-Wilkinson regression to analyse multi-environment trials, and of the additive main effects and multiplicative interaction (AMMI) model. The study compares JRA and AMMI with particular focus on robustness with increasing amounts of randomly selected missing data. The application is made using a data set from a breeding program of durum wheat (Triticum turgidum L., Durum Group) conducted in Portugal. The results of the two models result in similar dominant cultivars (JRA) and winner of mega-environments (AMMI) for the same environments. However, JRA had more stable results with the increase in the incidence rates of missing values.