Genotype × environment interaction and yield stability in a cowpea-based cropping system

Twenty-four cowpea genotypes were evaluated under sole cropping or additive series intercropping with sorghum from 2004 to 2005 at four sites representative of the Guinea and Sudan savannah ecologies in Ghana. The aim was to determine whether cowpea breeding programs that emphasize selection under sole-crop conditions have the potential to produce cultivars that are effective under additive series intercropping. Genotype × cropping systems interaction was significant for days to 50% flowering but not for grain yield, biomass and other studied traits. Genotypic yield reaction to cropping systems indicated that bridging the yield gap between sole cropping and intercropping systems is best addressed by agronomic interventions that reduce stress on intercrop cowpea rather than by selecting for specifically adapted genotypes for intercropping. Significant genotype × environment interactions were observed for all traits when data was pooled over cropping systems. Partitioning of the genotype × environment interaction variance indicated that days to 50% flowering was dominated by heterogeneity of genotypic variance, whereas genotype × environment interactions for grain yield and biomass was mainly due to imperfect correlations. Large differences in genotypic yield stability were observed as estimated by the among-environment variance, regression of yield on the environmental index, Kataoka’s index, and by partitioning of genotype × environment interaction sum of squares into components attributable to each genotype. The results suggest that in regions where genotype × environment interaction for yield frequently causes re-ranking across environments, genotypes with the least contribution to the interaction sum of squares are likely to be most productive. On the whole, the results support the contention that breeding under sole-crop conditions has the potential to produce cultivars effective under intercropping conditions.     

Non-parametric measures of phenotypic stability in chickpea genotypes (<Emphasis Type="Italic">Cicer arietinum</Emphasis> L.)

Multi-environment trials (MET) play an important role in selecting the best cultivars and/or agronomic practices to be used in future years at different locations by assessing a cultivar's stability across environments before its commercial release. Objective of this study is to identify chickpea (Cicer arietinum L.) genotypes that have high yield and stable performance across different locations. The genotypes were developed by various breeders at different research institutes/stations of Iran and the International Center for Agricultural Research in Dray Areas (ICARDA), Syria. Several statistical methods were used to evaluate phenotypic stability of these test chickpea genotypes. The 17 genotypes were tested at six different research stations for two years in Iran. Three non-parametric statistical test of significance for genotype × environment (GE) interaction and ten nonparametric measures of stability analyses were used to identify stable genotypes across the 16 environments. The non-parametric measures (Kubinger, Hildebrand and Kroon/Van der) for G × E interaction were highly significant (P < 0.01), suggesting differential responses of the genotypes to the test environments. Based on high values of nonparametric superiority measure (Fox et al. 1990) and low values of Kang's (1988) rank-sum stability parameters, Flip 94-123C was identified as the most stable genotype. These non parametric parameters were observed to be associated with high mean yield. However, the other nonparametric methods were not positively correlated with mean yield and were therefore not used in classifying the genotypes. Simple correlation coefficients using Spearman’s rank correlation, calculated using ranks was used to measure the relationship between the stability parameters. To understand the nature of relationships among the nonparametric methods, a hierarchical cluster analysis based on non weighted values of genotypes, was performed. The 10 stability parameters fell into three groups.     

Multi-attribute responses of maize inbred lines across managed environments

In maize (Zea mays L.) breeding programs, selection among and within segregating progenies is based mainly on indirect selection criteria. A better understanding of the environment influence on physiological attributes of maize inbred lines is important to the identification and selection of superior inbred lines as well as to successful hybrid seed production. In this study, the size and form of genotype (G) and genotype × environment (G × E) interaction effects for plant grain yield (PGY) and several physiological attributes were examined for 12 maize inbred lines grown in four managed environments, represented by two seasons (Y) and two nitrogen levels. Mixed model analysis revealed that the G effect was relatively high for attributes related to light capture, phenology, early biomass production, and numerical components of PGY. The G × E interaction effect explained most of the variability for PGY, harvest index (HI), and biomass production at maturity. Three-mode principal component analysis allowed us to: (1) describe the associations among multiple attributes across environments, (2) reveal the form of the main patterns of G × E interaction, (3) establish the importance of the genotype × year (G × Y) interaction for kernel number, HI, and biomass at maturity in determining PGY, (4) identify promising genotypes of high-PGY across environments, and (5) detect genotypes of similar response patterns for PGY but with a contrasting relative behavior for other attributes, which may permit the simultaneous selection for grain yield and desired secondary traits. Such selection results would contribute greatly in the identification of superior inbreds than selecting for grain yield alone.

Diversity of white and narrow-leafed lupin genotype adaptive response across climatically-contrasting Italian environments and implications for selection

White lupin (Lupinus albus) and narrow-leafed lupin (L. angustifolius) have special interest as high-protein feed crops but their cultivation is limited by low grain yields. This study aimed to support breeding programs targeting Italy or other climatically variable south-European regions by investigating within-species adaptation patterns across contrasting Italian environments. An additional aim was comparing species for yielding ability. Eight narrow-leafed and six white lupin cultivars featuring different origin, phenological type (Mediterranean in both species; winter in white lupin; spring in narrow-leafed lupin) and plant architecture (determinate or indeterminate in both species; tall or dwarf in white lupin) were evaluated in a Mediterranean and a subcontinental-climate site under autumn and late-winter sowing. Additive main effects and multiplicative interaction was preferable to joint regression for modeling yield responses. In both species, cross-over GE interaction was observed (P < 0.05), autumn-sown Mediterranean and subcontinental environments were the most-contrasting for GE effects, and widely adapted material included cultivars of Mediterranean phenological type with indeterminate growth. Material with determinate growth was not among the best-yielding entries in any environment, whereas a dwarf winter-type white lupin entry was specifically adapted to autumn-sown subcontinental environments. White lupin displayed larger genetic variation than narrow-leafed lupin for phenology and other traits. Relationships of morphophysiological traits with grain yield were environment-specific and were locally high for some white lupin traits (early flowering, long reproductive phase, high aerial biomass, low proportion of pod wall). White lupin exhibited higher yielding ability than narrow-leafed lupin in all environments but the late-winter sown Mediterranean one, when comparing locally top-yielding cultivars.     

Yield stability in common bean (Phaseolus vulgaris L.) genotypes

An experiment was undertaken to determine the stability of seed yield in 21 common bean genotypes representing three growth habits. Seven genotypes in each growth habit (determinate bush, indeterminate bush and indeterminate prostrate) were evaluated in replicated trials at three locations for three years under rain fed conditions in Ethiopia. A combined analysis of variance, stability statistics and rank correlations among stability statistics and yield-stability statistic were determined. The genotypes differed significantly for seed yield and genotype × environment (year by location) interaction (GE). The different stability statistics namely Type1, Type 2 and Type 3 measured the different aspects of stability. This was substantiated by rank correlation coefficient. There were strong rank correlations among S_i ²d, W_i ²,σ_i ² and S_i ², where as there was weak correlation between b_iand R_i ², S_i ²d, W_i ², σ_i ² and S_i ². R² was significantly and negatively correlated with W_i ², σ_i ² and S_i ². σ_i ² is significantly correlated with W_i ².Yield is significantly correlated with b_i and R_i ².None of the statistics per se was useful for selecting high yielding and stable genotypes except the YS(yield-stability statistic). Most of the high yielding genotypes were relatively stable. Of the 21 genotypes, only 11genotypes were selected for their high yielding and stable performance. Genotypes with growth habit III and I (in determinate prostrate and determinate bush) were generally more stable than in determinate bush. This revised version was published online in July 2006 with corrections to the Cover Date.     

Evaluation of environments for preliminary testing of maize for yield and stover digestibility

Summary The identification of environments suitable for selection should enable plant breeders to test reliably a larger number of genotypes with given resources. This research was undertaken to evaluate discrimination ability (DA) and prediction ability (PA) of eight environments involving two locations. Eckartsweier (Eck) 1 and Voelkenrode (Voe); 2 yer. 1984 (84) and 1985 (85); and two harvests, silage harvest (I) and later (II). It was based on dry matter yield of plant (PDMY), ear (EDMY) and stover (SDMY) and on in vitro digestible organic matter (IVDOM), neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin (ADL) of stover of 12 inbred lines and their 66 diallel crosses in maize (Zea mays L.). Linear regression coefficient of the performance of genotypes in an environment on that averaged across all environments and its degree of determination were used as measures of the DA and PA of environments, respectively. In hybrids, the DA of environments differed significantly for all traits except SDMY and ADL. Environment Eck851 showed better PA (0.63) than other environments for PDMY, EDMY, IVDOM, NDF, and ADF. Among the 12 two-environment combinations. EckI was a better predictor for PDMY, EDMY, IVDOM and NDF (PA0.80). The more productive environments showed better DA and PA for PDMY and FDMY than less productive ones. For IVDOM and NDF the first harvest provided better discrimination and prediction than the second harvest. In inbred lines the DA of environments differed significantly for EDMY only. In the present study, PA and DA seemed to be functions of the environments, PA was improved by combining two environments, and some environments seemed to be suitable for preliminary selection of genotypes for such diverse traits as yield and digestibility.     

Differences in reciprocal crosses of maize inbred lines diverse for protein content

Summary Forty-two crosses and their reciprocals in maize (Zea mays L.) involving inbred lines highly diverse for protein content were evaluated in four environments. Data were recorded on crude protein content of grain, protein yield, grain yield, 1000 kernel weight, dry matter content of ear at harves,, days to 50% slking, plant height, ear height, and early vigor. No significant variation due to reciprocal differences was observed for protein content and early vigor. For all other traits the variance component due to reciprocal x environment interactions was significant while the variance component due to reciprocal differences was significant only for kernel weight, dry matter content of ear, plant height, and ear height. The variance components due to nuclear differences and their interactions with environments were always highly significant and larger than the components due to reciprocal differences and reciprocal x environment interactions. The instability and low magnitude of reciprocal differences indicated that it might be difficult to exploit them commercially. However, seeing the presence of reciprocal differences for most of the traits studied, the evaluation of breeding materials for these differences seems to be important.The research work was supported by the Deutsche Forschungsgemeinschaft, Sonderforschungsbereich 142.     

Stability and genetic improvement of maize yield in Argentina

Summary Yield stability and the genetic improvement of maize (Zea mays L.) grain yield in the humid pampas of Argentina were evaluated.Stability parameters were computed for 15 varieties grown in 35 environments. To analyze genetic improvement of grain yield, data were obtained from trials grown in four locations and twelve years. Two locations represented the typical maize area, while the remaining two were considered marginal. Simple correlation and linear regression coefficients were computed to study the relationships between yield, stability parameters, number of days to 50% tassel emergence, and year of release. Genetic improvement of grain yield was analyzed from linear regression of the average yield of the three highest-yielding varieties (as percentage of the average yield of five common checks) on year of trial.Yield differed significantly among varieties. Significant variety x environment (linear) interaction was also detected. Significant linear relationships were found between regression coefficient for yield (stability parameter 1) and days to tassel emergence, stability parameter 1 and yield, year of release and days to tassel emergence, and year of release and stability parameter 1. Thus, newer maize varieties tended to flower later and had greater responsiveness to favorable environments than did older varieties. Yields have increased for both the typical and marginal areas, with average yearly increases of 114 and 182 kg/ha, respectively.Breeding programs with nurseries located in the typical maize area have raised yield potential in both areas studied. However, if present trends continue, future releases could prove inadequate in low-yielding or short season environments.     

Expression of mature plant resistance to Striga hermonthica in maize

Twelve maize genotypes with different degrees of resistance were evaluated in plots inoculated with seeds of the phytoparasite Striga hermonthica (Del.) Benth. and in Striga-free (control) plots for three seasons between 1991 and 1993. Resistant genotypes, although showing similar levels of underground infection as susceptible 9 weeks after maize planting (WAP), had significantly fewer emerged parasitic plants and sustained lower damage. Host damage was not determined by the severity of infection. Percent height and dry matter reductions increased from 3 to 6 WAP for resistant and susceptible genotypes. While percent height and dry matter reductions for resistant genotypes declined at 9 and 12 WAP, susceptible genotypes either maintained or increased their levels of damage. Correlation between the severity of height and dry matter reductions at the vegetative (3 and 6 WAP) and reproductive (9 and 12 WAP) stages of maize growth were not significant, indicating that damage at early stages of growth cannot be used to predict mature plant response. Maize damage score (1–9) at 8 WAP, a non destructive and composite shoot indicator of host performance under Striga infestation, was significantly correlated (r = 0.88, r = 0.82; p < 0.01) to the level of shoot reduction at mature plant stages. Emerged Striga count at 8 WAP was significantly correlated (r = 0.98, p < 0.01) to the count at 10 WAP, the time when parasite emergence was highest. Under high and uniform infestation, mature plant resistance can be detected at 8 WAP, the onset of flowering in maize. This revised version was published online in July 2006 with corrections to the Cover Date.     

Phenotypic correlations between maize inbreds and their single cross hybrids in short season areas

Summary This study was conducted to determine the relationships among various agronomic traits in inbreds, hybrids and between inbreds and their progeny of maize in short season areas. Phenotypic correlations of each measured trait between as well as among hybrids and inbreds were not the same. The highest correlation coefficient (r=0.78) was obtained between days to anthesis and yield. The high yielding inbreds did not necessarily give rise to high yielding progeny. Given that days to anthesis plays an important role in two major traits (yield and moisture) for hybrids, an emphasis on optimum anthesis dates during inbred development could be a useful criterion for the prediction of hybrid performance in short season areas.     

Characterization of the effect of a QTL for drought resistance in rice, qtl12.1, over a range of environments in the Philippines and eastern India

A large-effect QTL for grain yield under drought conditions (qtl12.1) was reported in a rice mapping population derived from Vandana and Way Rarem. Here, we measured the effect of qtl12.1 on grain yield and associated traits in 21 field trials: ten at IRRI in the Philippines and 11 in the target environment of eastern India. The relative effect of the QTL on grain yield increased with increasing intensity of drought stress, from having no effect under well-watered conditions to having an additive effect of more than 40% of the trial mean in the most severe stress treatments. The QTL improved grain yield in nine out of ten direct-seeded upland trials where drought stress was severe or moderate, but no effect was measured under well-watered aerobic conditions or under transplanted lowland conditions. These trials confirm that qtl12.1 has a large and consistent effect on grain yield under upland drought stress conditions, in a wide range of environments.     

Selection of sites for preliminary maize yield trials in the rainforest zone of South-Western Nigeria

Summary Grain-yield data obtained from 49 varieties of maize (Zea mays L.), grown at each of five sites for two years, were analyzed with the objective of identifying two or three sites where mean performance adequately predicted the mean performance of all five sites combined. Regression of overall varietal mean on site or site-combination means was done and the predictive potential of a site or combination of sites was judged by four criteria: (i) r²0.9, (ii) b1.0, (iii) high yield (i.e., yield greater than or equal to overall mean yield) and (iv) consistent yield (i.e., yield did not fall below the mean yield by more than 10%). In addition, a trial differentiation index, D, was used to compare the site combinations selected on the basis of the above criteria with the differentiation potential of all five sites combined. None of the five individual sites and the ten two-site combinations satisfactorily predicted overall mean yield. However, five three-sites combinations had values of D similar to that of all the five sites. Rank-correlation coefficients of the three-site combinations with all five sites ranged from 0.90 to 0.79. It was concluded that certain three-site combinations could be satisfactorily used for preliminary maize yield trials in the rainforest ecological zone of S.W. Nigeria.     

Maturity interaction and black layer occurrence in opaque-2 and normal hybrids in maize (Zea mays L.)

Summary Six opaque-2 lines and their normal counterparts were crossed in diallel crosses. The crosses were grown in 1970 in a split plot randomized complete block design. Harvests were made at 7-day intervals starting at 28 days after pollination and countinuing through 63 days.The average kernel weight of opaque-2 hybrids was inferior to that of the normal. Nevertheless, the opaque-2 gene performed differently in different hybrids. In B14 x B37 single cross the mutant had similar kernel weight as its normal counterpart in the first and second harvests. In contrast a wide difference was found between the opaque-2 and the normal, both at early and late stages of development in W64A x A545 background.The normal hybrids had greater cob weight, ranging from 9'7 to 11.8% more than the opaque-2. The difference in cob weight of the opaque-2 and the normal remained constant over the different harvest dates.At physiological maturity, the opaque-2 hybrids averaged 3.5% higher moisture content than the normal. In general, a slower accumulation of dry matter in the kernels was accompanied by a retention of more moisture.Shelling percentage was higher for the normal hybrids. Black layer, an indicator of physiological maturity, was formed at about the same time in the opaque-2 and normal.Journal paper No. 6094.     

Six years of maize selection in ‘Kitale synthetic II’, ‘Ecuador 573’, and ‘kitale composite a’ using methods of the comprehensive breeding system

Summary Evaluation of the results of six years of selection in the Kitale maize breeding methods study proved reciprocal recurrent selection to be an effective interpopulation improvement method. Ear-to-row selection was effective in improving Kitale Composite A (KCA). Data from a diallel of the cycle-6 ear-to-row substrains showed that where less than 10 lines were selected, inbreeding appeared to depress gain, but differences among the various ear-to-row experiments were not significant. Intererossing the substrains was predicted to reduce the effect of inbreeding. Yield gains in reciprocal recurrent selection and ear-to-row selection were associated with increases in ears per 100 plants. The reciprocal recurrent selection variety cross yield gain was estimated at 3.5% per year (7%/cycle) and that of the best ear-to-row procedure at 2.6% per year. The variety cross syn-2 from reciprocal recurrent selection was predicted to improve at the same rate as the best ear-to-row procedure (E7) with one long growing season per year. Under a cropping system with two similar seasons per year, however, ear-to-row selection should be more effective than reciprocal recurrent selection or its derived syn-2. Mass selection at 10% selection intensity produced significant improvement, but not at 2% selection intensity.S₁ and three half-sib selection methods were ineffective in KCA. Inbreeding and the lack of linkage equilibrium in KCA were discussed as contributing. Theoretical substrain compositing required 30 to 40 lines selected to equal the effectiveness of ear-to-row selection with 10 lines. Full-sib selection was effective in improving yield at 1.2% per year. All the recurrent selection methods showed a significant increase in ears per 100 plants. Population size was shown to be important in reducing losses due to inbreeding.If a hybrid or variety cross is potentially useable, reciprocal recurrent selection provides the best population approach. Initially, the variety cross syn-2 could be released with no additional breeding effort. Transition to variety crosses or traditional hybrids from extracted lines could be done as farmer husbandry improves and seed production facilities develop. Otherwise, open-pollinated varieties superior to existing maizes can be effectively developed using ear-to-row selection. The other methods may be effective in breeding populations when larger effective population sizes are used, but they were not effective in KCA when only 10 lines were recombined each cycle.Contribution from the East African Agriculture and Forestry Research Organization of the East African Community, the U.S. Agency for International Development (Project 618-11-110-657, East African Food Crop Research), the Kenya Ministry of Agriculture, and the Agricultural Research Service of the U.S. Department of Agriculture.     

Genetic diversity evolution through participatory maize breeding in Portugal

Natural, and in particular, artificial (human) selection may pose a danger to the existing crop genetic diversity. Nevertheless, on-farm breeding systems seem to achieve phenotypic improvements even though preserving variability. Using SSR markers, we analysed several selection cycles, over a 20 years period, of a Portuguese on-farm participatory maize OPV-‘Pigarro’ breeding project. No significant differences in allelic richness (N_ar), observed heterozygosity (H_O), expected heterozygosity (or gene diversity; H_E) or inbreeding coefficient (f) were detected among the selection cycles. 58 out of 107 alleles were common to all the selection cycles studied. The analysis of molecular variance showed that the variation among selection cycles represented only 7% of the total molecular variation. However, the number of private alleles varied among the selection cycles, being the highest detected at the beginning of the selection project. These findings demonstrate that an allele flow took place during the on-farm selection process of ‘Pigarro’ but the level of genetic diversity was not significantly influenced. Since interesting phenotypic improvements were also achieved, on-farm breeding projects, like this one, should be valued as a way to preserve unique Portuguese maize landraces in risk of disappearing.     

Specific adaptation of barley varieties in different locations in Ethiopia

Barley is one of the most important cereal crops grown for the livelihoods of the poor farmers of Tigray region in northern Ethiopia. As many low input and marginal environments it has benefited less from the yield increases achieved by modern breeding. This has been largely attributed due to genotype × environment intraction (GEI). To investigate the causes of GEI, ten barley varieties including local checks (two farmers developed varieties, four modern varieties and three rare local varieties) were tested over 21 environments. Participatory methods were applied to sample an adequate number of environments spanning the regional diversity. The yielding ability and stability of the varieties was graphically depicted by GGE and PLSR biplot. There were two major groups of environments, the central and northern highlands, the latter with less rainfall and poorer soils. Rainfall per month and total nitrogen level were the environmental variables that differentiated these two groups. In Tigray, rainfall in June and July were negatively correlated with yield, reflecting waterlogging problems. The different varieties were either specifically or widely adapted across the two environments. The variety ‘Himblil’, originating in Tigray, was the highest yielding and also most stable in the region of origin. However, it was inferior to improved varieties (Shege and Dimtu) at high yield levels. The association of earliness with grain yield indicates that the trait can be effectively manipulated within the existing materials. We recommend breeding for drought/water logging resistance based on selection in the target environment as the best strategy to provide stable and high yielding varieties for Tigray.     

Mass selection for kernel depth in early maize

Summary Effectiveness of mass selection for kernel depth in an early maize (Zea mays L.) synthetic, i.e. NDSAB, was evaluated in eight North Dakato environments. Four cycles of selection for deeper kernels were evaluated for grain yield and associated traits by using corresponding randomly selected subpopulations of NDSAB to adjust out inbreeding effects on selected subpopulations. General combining ability (GCA) effects of selected subpopulations also were evaluated for these traits.Subpopulations selected from one to four cycles for deep kernels were reciprocally crossed to corresponding randomly selected subpopulations in a factorial mating design. Subpopulations, selfed subpopulations, and subpopulation crosses were evaluated for yield and associated traits. Subpopulations selected for deep kernels generally had deeper kernels, higher yields, heavier kernels, more kernel rows per ear, and lower test weights than corresponding randomly selected subpopulations. Differences in GCA effects among selected and unselected subpopulations generally corresponded to differences between selected and unselected parental subpopulations. Inbreeding effects from selfing were observed for all traits, but no average heterosis was detected when subpopulation crosses were compared to parental subpopulations. Reciprocal differences were observed for all traits except kernel depth, but these differences were inconsistent relative to whether deep kerneled females made superior hybrids to unselected females.Results supported the use of mass selection for kernel depth as a means of grain yield improvement in the short term for NDSAB. However, negative associations of kernel depth with test weight may restrict long term use of this procedure.Journal Article No. 1505 of the North Dakota Agric. Exp. Stn., Fargo     

A selection procedure for ear drying-rates in maize

Summary A selection procedure to change the drying rates of maize (Zea mays L.) ears was developed and tested. Results indicated that drying rates were affected by hybrid genotype, ear maturity at harvest, number of kernels per row, ear diameter, and moisture content at harvest.Mass selection was initiated in an early synthetic, NDSG, in an attempt to produce both fast and slow drying strains. After two cycles resultant substrains were evaluated in the laboratory for direct and in the field for correlated selection responses. Laboratory results indicated that selection effectively changed moisture loss rates in NDSG, and confirmed earlier observations that an ear's moisture content at harvest affects its drying rate.Data from field experiments grown at five locations in 1980 revealed that each of two selection cycles for slow laboratory drying rate significantly reduced ear moisture content at harvest, in the field. This lower harvest moisture content appeared to result from a lower moisture content at physiological maturity rather than a faster drying rate. Other correlated selection responses included lowered yield, plant height, and ear weight due to selection for fast laboratory drying, and lowered plant height and ear weight from selection for slow drying.Data indicated that this selection procedure can be used to change moisture loss rate and ear moisture content at harvest. Results also contributed to understanding of factors involved in ear drying rates which may lead to more effective selection procedures.Journal Article No. 1385 of the North Dakota Agri. Exp. Stn., Fargo.     

Genotype x environment interaction for root traits in maize,as analysed with factorial regression models

Summary The efficiency of genotopic and climatic characteristics in accounting for the interaction between genotypes and environments has been assessed in a three-year trial involving a set of genotypes presenting a range of root morphology characteristics (number and size). Climatic information on rainfall and temperature was recorded during the experiments, together with extra data on the growth and development of the genotypes. Their effects have been tested in factorial regression models.Climatic covariates were very powerful in accounting for the genotype by year interaction as well as the year main effect alone. For the number of adventitious roots on internode 7, the main effect of year could be described as a linear function of the average temperature and precipitation that occurred during the period of root initiation and growth. For internode 6, no clear conclusion was possible. For the root traits studied, 74 to 98% of the interaction could be explained by one climatic covariate. The regression coefficients can be considered as measures of genotypic stability.The genotypic covariates describing aerial development performed rather poorly, compared with environmental ones, even though the physiological and functional relationships between root and shoot are well known. Neither genotype main effect nor genotype by year interaction could be described sufficiently by factorial regression. Still, the genotypic covariates performing best clearly differed between root counts and size. Also the best genotypic covariates differed for main effect and interaction.