Search

The use of nonlinear functions provides a concise measure of a variety of physiological traits associated with growth and development that would otherwise be difficult to observe. Maize kernel growth and development is a complex process, and the description of the processes involved benefits from the application of a nonlinear function to the process over growing degree days (GDD), a measure of heat accumulation over a period of development. The objective of this study was to compare and contrast canonical models of growth and development to determine which provided the best description of maize kernel biomass accumulation. Observations of kernel dry weights starting shortly after pollination through maturity were regressed onto a measure of thermal time. Observations from differing maize hybrids taken in two years with significantly different weather patterns were used to construct the model. Of the four nonlinear functions described, the Weibull and Gompertz functions were found to describe the pattern of biomass accumulation best. The Gompertz function was selected to describe kernel growth based on information criteria, biological interpretation of the parameters, and computational ease. Tests of autocorrelation and homogeneity of errors determined that the errors were heteroscedastic but autocorrelation was not an issue. The application of a variance function, which models the residuals as a function of the variance, was used to account for heteroscedastic errors. While the Gompertz function was selected as the best fit for use with this data set, it is suggested that future studies use the selection process described herein to determine the most appropriate function.

Genotype-by-environment interactions (GEI) have been identified as an important component of the genotypic variation for grain and stover yield traits of rabi sorghum varieties and hybrids in India. It has been argued that obtaining an understanding of the causes of these GEI is an important step to identify the scope for genetic improvement of grain and stover yield by conventional breeding. Pattern analysis was used to investigate the regional structure of the GEI for the four traits grain yield, stover yield, days-to-flower, and plant height, measured on the varieties and hybrids tested over 10 years in the All India Coordinated Sorghum Improvement Program (AICSIP). There was evidence that regional differences accounted for part of the GEI for grain yield and days-to-flower but not for stover yield and plant height. Cluster analysis was used to group the locations included in the ACSIP experiments. The five-group level was chosen to examine the causes of GEI among the location groups. Hypotheses were proposed for the observed regional grouping of trials for grain yield. One group of trials consisted of predominantly irrigated trials, the other four groups differed in the timing and intensity of drought stress imposed on the entries. This retrospective analysis provides a basis for testing the hypothesised contributions of environmental variation in water availability to regional GEI for grain and stover yield. If these hypotheses are substantiated, the current multi-environment testing strategy used for the AICSIP trials could be modified to ensure adequate sampling of the five regional groups identified by the retrospective pattern analysis.

Grain and stover yield are key traits for the improvement of rabi sorghum varieties and hybrids in India. Large genotype-by-environment interactions (GEI) were identified for both grain and stover yield based on 10 years’ data available from the All India Coordinated Sorghum Improvement Program (AICSIP) experiments conducted from 1986/87 to 1996/97. From the estimates of the genotypic, GEI and residual components of variance for the four traits, a multi-environment testing strategy based on 15–20 locations, 2–3 years, and two replicates per trial would be required to achieve estimates of repeatability of approximately 0.70 for grain yield, 0.45 for stover yield, 0.70 for flowering time, and 0.85 for plant height. The results of these experiments indicate that over the 10-year period there was little genetic progress demonstrated for both yield traits and greater grain yields were associated with earlier flowering and greater stover yields were associated with greater plant height. Selection for greater grain and stover yield, while holding flowering time and plant height constant will be difficult. Unless the physiological trade-off between grain and stover yield and the causes of the large GEI for both traits can be better understood, conventional selection on phenotypic performance for grain and stover yield in the AICSIP trials is likely to continue to be slow.

While a large genotype-by-environment (G x E) interaction component of variance for grain yield (GY) has been widely reported for rainfed lowland rice, the reasons for such large interactions are not well known. A random reference population of 34 genotypes taken from the Cambodian rice improvement program was used to examine the magnitude and nature of G x E interactions for GY in Cambodia. These genotypes were evaluated in a multi-environment trial (MET) conducted across 3 years (2000-2002) and eight locations in the rainfed lowlands. The G x E interaction was partitioned into components attributed to genotype-by-location (G x L), genotype-by-year (G x Y) and genotype-by-location-by-year (G x L x Y) interactions. The G x L x Y interaction was the largest component of variance for GY. The G x L interaction was also significant and comparable in size to the genotypic component (G). The significant G component was partly explained by a group of four genotypes that were broadly adapted to different environmental conditions represented by three environmental groups. The three environmental groups were identified from a pattern analysis, and the grouping was partly related to the time of sowing, and hence water availability at flowering. A major factor contributing to the large G x L x Y interactions for GY was late maturing genotypes being affected greatly when soil water availability at flowering was reduced greatly, compared to earlier maturing genotype groups. While the differential genotypic responses to the water availability environment explained part of a large G x E interaction for GY, other non-water related environmental conditions also appeared to have contributed to the interaction. Three target environments were identified for focusing efforts of the breeding programs in Cambodia, and four putative genotypes were selected for their high yield and wide adaptation in the rainfed lowlands. (c) 2006 Elsevier B.V. All rights reserved.

Drought is a major constraint for rice production in the rainfed lowlands in Southeast Asia and Eastern India. The breeding programs for tainted lowland rice in these regions focus on adaptation to a range of drought conditions. However, a method of selection of drought tolerant genotypes has not been established and is considered to be one of the constraints faced by rice breeders. Drought response index (DRI) is based on grain yield adjusted for variation in potential yield and flowering date, and has been used recently, but its consistency among drought environments and hence its usefulness is not certain. In order to establish a selection method and subsequently to identify donor parents for drought resistance breeding, a series of experiments with 15 contrasting genotypes was conducted under well-watered and managed drought conditions at two sites for 5 years in Cambodia. Water level in the field was recorded and used to estimate the relative water level (WLREL) around flowering as an index of the severity of water deficit at the time of flowering for each entry. This was used to determine if DRI or yield reduction was due to drought tolerance or related to the amount of available water at flowering, i.e. drought escape. Grain yield reduction due to drought ranged from 12 to 46%. The drought occurred mainly during the reproductive phase, while four experiments had water stress from the early vegetative stage. There was significant variation for water availability around flowering among the nine experiments and this was associated with variation in mean yield reduction. Genotypic variation in DRI was consistent among most experiments, and genotypic mean DRI ranged from -0.54 to 0.47 (LSD 5% = 0.47). Genotypic variation in DRI was not related to WLREL around flowering in the nine environments. It is concluded that selection for DRI under drought conditions would allow breeders to identify donor lines with high drought tolerance as an important component of breeding better adapted varieties for the rainfed lowlands; two genotypes were identified with high DRI and low yield reduction and were subsequently used in the breeding program in Cambodia. (c) 2006 Elsevier B.V. All rights reserved.

Screening for drought resistance of rainfed lowland rice using drought score (leaf death) as a selection index has a long history of use in breeding programs. Genotypic variation for drought score during the vegetative stage in two dry season screens was examined among 128 recombinant inbred lines from four biparental crosses. The genotypic variation detected for drought score in the dry season was used to examine the reliability of the dry season screening method to estimate relative grain yield of genotypes under different types of drought stress in the wet season. Large genotypic variation for drought score existed in two experiments (A and B). However, there was no relationship between the drought scores of genotypes determined in these two experiments. Different patterns of development and severity of drought stress in these two experiments, i.e. slow development and mild plant water deficit in experiment A and fast development and severe plant water deficit in experiment B, were identified as the major factors contributing to the genotypes responding differently. Larger drought score in the dry season experiments was associated with lower grain yield under specific drought stress conditions in the wet season, but the association was weak to moderate and significant only in particular drought conditions. In most cases, a significant phenotypic and moderate genetic correlation between drought score in the dry season and grain yield in the wet season existed only when both drought score and grain yield of genotypes were affected by similar patterns and severity of drought stress in their respective experimental environments. The dry season environments used to measure genotypic variation for drought score should be managed to correspond to relevant types of drought environment that are frequent in the wet season. The efficiency of using the drought score as an indirect selection criterion for improving grain yield for drought conditions was lower than the direct selection for grain yield, and hence wet season screening with grain yield as a selection criterion would be more efficient. However, using drought score as a selection index, a larger number of genotypes can be evaluated than for wet season grain yield. Therefore, it is possible to apply higher selection intensities using the drought score system, and the selected lines can be further tested for grain yield in the wet season. (C) 2004 Elsevier B.V. All rights reserved.

A sample of recombinant inbred lines (RILs) was derived from a bi-parental cross between Lemont and BK88-BR6, which contrasted in maintenance of leaf water potential (LWP) and expression of osmotic adjustment (OA). Genotypic variation for LWP and OA, and their associations with yield determination under water deficit, was studied in a series of five field experiments. Genotypic variation in the maintenance of high LWP was consistent across water deficit experiments. In the determination of genotypic variation in the maintenance of LWP, rate of water deficit was not an important factor influencing ranking, but degree of water deficit, and phenological development stage were important, particularly around heading. Genotypic variation in expression of OA was also observed under water deficits during both vegetative and flowering stages but ranking was inconsistent across experiments. This was in part because of large experimental errors associated with its measurement, but also because the expression of OA was associated with extent of decline of LWP. The relationship between OA and LWP was demonstrated when data were combined across experiments for vegetative and flowering stages. Under water-limited conditions around flowering, grain yield reduction was mainly due to a increased spikelet sterility. Variation in OA was not related to grain yield nor yield components. There were however, negative phenotypic and genetic correlations between LWP and percentage spikelet sterility measured at flowering stage on panicles at the same development stage during a water deficit treatment. This suggests that traits contributing to the maintenance of high LWP minimized the effects of water deficit on spikelet sterility and consequently grain yield. (C) 2002 Elsevier Science B.V. All rights reserved.

Responses of rice genotypes to drought stress may be different when characteristics of the drought stress environments differ. The performance of 128 genotypes was examined under irrigation and four different types of drought stress, to determine genotypic consistency in yield and factors determining yields under different drought stress conditions. The different drought conditions were mild drought during grain filling, short and severe drought at flowering, prolonged severe drought during the reproductive to grain filling, and prolonged mild drought during vegetative and grain filling. Genotypic grain yield under mild stress conditions was associated with yield under irrigated conditions, indicating the importance of potential yield in environments where the yield reduction was less than 50%. However, yields under irrigated conditions differed over time and locations. Under prolonged or severe drought conditions, flowering time was an important determinant of grain yield. Earlier flowering genotypes escaped the severe stress and had higher grain yields indicating large genotype by environment (G x E) interactions which have implications for plant breeding even for mild stress. It is suggested that variations in flowering time, potential yields and drought patterns need to be considered for development of drought-resistant cultivars using specific physiological traits. (C) 2002 Elsevier Science B.V. All rights reserved.

Drought frequently reduces grain yield of rainfed lowland rice. A series of experiments were conducted in drought-prone northeast Thailand to study the magnitude and consistency of yield responses of diverse, rainfed lowland rice genotypes to drought stress environments and to examine ways to identify genotypes that confer drought resistance. One hundred and twenty-eight genotypes were grown under non-stress and four different types of drought stress conditions. The relationship of genotypic variation in yield under drought conditions to genetic yield potential, flowering time and flowering delay, and to a drought response index (DRI) that removed the effect of potential yield and flowering time on yield under stress was examined. Drought stress that developed prior to flowering generally delayed the time of flowering of genotypes, and the delay in flowering was negatively associated with grain yield, fertile panicle percentage and filled grain percentage. Genotypes with a longer delay in flowering time had extracted more water during the early drought period, and as a consequence, had higher water deficits. They were consistently associated with a larger yield reduction under drought and in one experiment with a smaller DRI. Genotypes, however, responded differently to the different drought stress conditions and there was no consistency in the DRI estimates for the different genotypes across the drought stress experiments. The results indicate that with the use of irrigated-control and drought test environments, genotypes with drought resistance can be identified by using DRI or delay in flowering. However, selections will differ depending on the type of drought condition. The inconsistency of the estimates in DRI and flowering delay across different drought conditions reflects the nature of the large genotype-by-environment interactions observed for grain yield under various types of drought in rainfed lowland conditions. (C), 2002 Elsevier Science B.V. All rights reserved.

The magnitude of genotype-by-management (G x M) interactions for grain yield and grain protein concentration was examined in a multi-environment trial (MET) involving a diverse set of 272 advanced breeding lines from the Queensland wheat breeding program. The MET was structured as a series of management-regimes imposed at 3 sites for 2 years. The management-regimes were generated at each site-year as separate trials in which planting time, N fertiliser application rate, cropping history, and irrigation were manipulated. irrigation was used to simulate different rainfall regimes. From the combined analysis of variance, the G x M interaction variance components were found to be the largest source of G x E interaction variation for both grain yield (0.117 +/- 0.005 t(2) ha(-2); 49% of total G x E 0.238 +/- 0.028 t(2) ha(-2)) and grain protein concentration (0.445 +/- 0.020%(2); 82% of total G x E 0.546 +/- 0.057%(2)), and in both cases this source of variation was larger than the genotypic variance component (grain yield 0.068 +/- 0.014 t(2) ha(-2) and grain protein 0.203 +/- 0.026%(2)). The genotypic correlation between the traits varied considerably with management-regime, ranging from -0.98 to -0.31, with an estimate of 0.0 for one trial. Pattern analysis identified advanced breeding lines with improved grain yield and grain protein concentration relative to the cultivars Hartog, Sunco and Meteor. It is likely that a large component of the previously documented G x E interactions for grain yield of wheat in the northern grains region are in part a result of G x M interactions. The implications of the strong influence of G x M interactions for the conduct of wheat breeding METs in the northern region are discussed. (C) 2001 Elsevier Science B.V. All rights reserved.

The physical environment of the rainfed lowland ecosystem is often characterised and grouped according to the surface hydrology of rice paddies and rice cultivars have been developed for each subecosystem. Rainfall is an important determinant of the yield of rainfed lowland rice, but other factors such as topography and soil fertility also affect grain yield and choice of cultivars. The growing environment and also rice yield vary greatly within small areas as well as across seasons. This causes great difficulty in determining the target population of environments for a rice breeding program. This paper reviews past work on characterising the variability in the physical environment, and rice production in the rainfed lowland ecosystem. It examines possible connections between this variability and slow progress in developing new cultivars that are widely adapted to the rainfed lowland rice ecosystem.

In this review, reasons for the slow progress in development of drought resistant cultivars for rainfed lowland rice are considered first. Recent advances in screening methods for development of drought resistant cultivars from mostly our research in Thailand in the 1990s, are then described for drought that develops early or late in the wet-season. There now appears to be a good prospect for developing drought resistant cultivars that produce higher yield than existing cultivars when drought develops late in the wet-season. Appropriate phenology to escape late-season drought and high potential yield under well-watered conditions are important characters for cultivars adapted to rainfed lowland conditions. In addition, ability to maintain higher leaf water potential when drought develops late in the season is another desirable character. Maintenance of leaf water potential just prior to flowering is associated with higher panicle water potential, reduced delay in flowering time, and reduced spikelet sterility, and hence contributes to higher yield. Genotypes that are adapted to areas of late-season drought should also have high harvest index, intermediate height and rather small total dry matter compared to existing traditional cultivars, under well-watered conditions. This combination of characters would ensure high potential yield under favourable conditions and also contribute to resistance against late-season drought. Screening against early-season drought that develops during the vegetative stage is more difficult, as the genotype’s ability to recover from the stress appears more important than drought tolerance during the stress period. Other than appropriate phenology, high potential yield and ability to maintain high leaf water potential, no specific physiological and morphological characters appear to contribute directly to higher yield under drought conditions in rainfed lowland rice in Thailand, where drought develops rather rapidly due to the prevailing coarse textured soils. It is thus appropriate to develop a breeding program that is primarily based on selection for grain yield. There are large genotype by environment interactions for yield in rainfed lowland rice and hence it is important that genotypes are selected for yield under appropriate target environments. Addition of a drought screening program that is conducted in the field in the wet-season to the overall breeding program would enhance the opportunity to select for drought resistance within the breeding materials and increase the chance of developing high yielding cultivars adapted to the drought-prone rainfed lowland environments.

Stable and high yields of rainfed lowland rice are important for food security in many of the subsistence farming systems in Asia. Progress in breeding for higher yields has been slow. Improving the adaptation of rainfed lowland rice to the range of abiotic stresses encountered in these drought-prone systems has been advocated as the basis of a breeding strategy to increase the rate of genetic improvement of yield. Putative adaptive traits have been nominated and studied. However, there is little consensus on the definition of appropriate trait-based selection criteria to complement direct selection for yield. The complexity of the target population of environments, and the incidence of large genotype-by-environment (G x E) interactions, has stimulated interest in analysis and interpretation of yield variation in multi-environment trials as part of the process of examining the adaptation of rainfed lowland rice. The rationale underlying analysis of adaptation is examined from the perspective of facilitating selection decisions within rainfed lowland rice breeding programs. (C) 1999 Elsevier Science B.V. All rights reserved.

The magnitude and nature of genotype-by-environment (G x E) interactions for grain yield, days-to-flower and plant height of rainfed lowland rice in Northeast Thailand were examined using random F-7 lines from seven crosses developed by the Thai breeding program. A total of 1116 lines and checks were evaluated in a multi-environment trial conducted across three years (1995-1997) and eight sites. The G x E interaction was partitioned into components attributed to genotype-by-site (G x S), genotype-by-year (G x Y) and genotype-by-site-by-year (G x S x Y) interactions. The G x S x Y interaction was the largest G x E interaction component of variance for all three traits. There was little G x S interaction for grain yield and days-to-flower. The G x S interaction was significant for plant height, but was the smallest component of variance for that trait. The G x Y interaction component of variance was significant for all three traits, but was small relative to the genotypic component for days-to-flower and plant height. For grain yield the G x Y interaction component was comparable in size to the genotypic component. Partitioning the genotypic and G x E interaction components of variance into among-cross and within-cross components indicated that there was significant variation both among and within the crosses for each trait. The relationships between the three traits differed among the crosses and the environments. A major factor contributing to the large G x S x Y interactions for grain yield was the genotypic variation for days-to-flower in combination with environmental variation for the timing and intensity of drought. Some of the interactions associated with timing of drought were repeatable across the environments sampled in the multi-environment trial, and to some extent the environments were characterised on the basis of whether there was pre-flowering, intermittent, or terminal drought. There was genotypic variation for grain yield after taking into consideration the influences of timing of drought in relation to plant development, Three of the seven crosses involved the Thai cultivars KDML105 and RD6 as parents. These crosses produced an array of progeny with lower yield than the Thai cultivars, suggesting it would be difficult to improve on the yield of these cultivars. In contrast three of the remaining four crosses, which did not have a Thai cultivar as a parent, produced progeny that had higher yield than KDML105 and RD6 indicating that yield of rainfed lowland rice could be improved above that of the these popular cultivars. (C) 1999 Elsevier Science B.V. All rights reserved.

An experiment was conducted to quantify the effect of rate of development of water deficit (Rψ) on the expression of maximum osmotic adjustment (OA) and desiccation tolerance (DT) in three grain sorghum lines, TAM422, Tx2813 and QL27. DT was determined as the lethal relative water content (RWCL) and lethal leaf water potential (ψL) of leaves, i.e. the value of these traits just prior to tissue death. Five rates of development of water deficit were generated by using different soil volumes in which the decrease in leaf water potential ranged from 0.093 MPa day−1 to 0.153 MPa day−1. The expression of maximum OA increased markedly with a reduced Rψ, for all three lines. In contrast, DT decreased only slightly with a reduced Rψ. The three lines differed for expression of maximum OA and DT. Both TAM422 and Tx2813 had higher maximum OA, higher RWCL and lower ψL than QL27. There was a significant line-by-Rψ interaction for the expression of OA and DT. The interaction was much smaller than the effect of lines for the expression of maximum OA. Therefore, the absolute level of OA for the three sorghum lines and the magnitude of the difference between them changed with the Rψ without affecting their ranking. As with maximum OA, the interaction for RWCL was smaller than the line effect. The RWCL of TAM422 and QL27 was stable across rates of development of water deficit. The results indicated that low Rψ increased the expression of OA and decreased ψL but had little effect on RWCL. While line-by-soil-volume interactions were observed, these were largely explained by the increase in magnitude of the variation for the traits as Rψ decreased and were not the result of changes in rank of the lines among soil volumes. Therefore, the current screening strategy is considered to be appropriate for examining genetic variation for OA and DT in grain sorghum. Opportunities exist for using smaller pots and soil volumes where large numbers of genotypes are examined.

Drought is a major problem for rice grown under rainfed lowland and upland conditions, but progress in breeding to improve drought resistance has been slow. This paper describes patterns of water-stress development in rice fields, reviews genetic variation in physio-morphological traits for drought resistance in rice, and suggests how knowledge of stress physiology can contribute to plant breeding programmes that aim to increase yield in water-limiting environments. To provide a basis for integrating physiological research with plant-breeding objectives we define drought resistance in terms of relative yield of genotypes. Therefore, a drought-resistant genotype will be one which has a higher grain yield than others when all genotypes are exposed to the same level of water stress. A major reason for the slow progress in breeding for drought resistance in rice is the complexity of the drought environment, which often results in the lack of clear identification of the target environment(s). There is a need to identify the relative importance of the three common drought types; early-season drought which often causes delay in transplanting, mild intermittent stress which can have a severe cumulative effect, and late stress which affects particularly late-maturing genotypes. In addition, in rainfed lowland rice, flooded and non-flooded soil conditions may alternate during the growing season, and affect nutrient availability or cause toxicity. Several drought-resistance mechanisms, and putative traits which contribute to them, have been identified for rice; important among these being drought escape via appropriate phenology, root characteristics, specific dehydration avoidance and tolerance mechanisms, and drought recovery. Some of these mechanisms/traits have been shown to confer drought resistance and others show potential to do so in rice. The most important is the appropriate phenology which matches crop growth and development with the water environment. A deep root system, with high root length density at depth is useful in extracting water thoroughly in upland conditions, but does not appear to offer much scope for improving drought resistance in rainfed lowland rice where the development of a hard pan may prevent deep root penetration. Under water-limiting environments, genotypes which maintain the highest leaf water potential generally grow best, but it is not known if genotypic variation in leaf water potential is solely caused by root factors. Osmotic adjustment is promising, because it can potentially counteract the effects of a rapid decline in leaf water potential and there is large genetic variation for this trait. There is genotypic variation in expression of green leaf retention which appears to be a useful character for prolonged droughts, but it is affected by plant size which complicates its use as a selection criterion for drought resistance. There is a general lack of drought related research for rice in rainfed lowland conditions. This needs to be rectified, particularly considering their importance relative to upland conditions in Asian countries. We suggest that focussing physiological-genetic research efforts onto clearly defined, major target environments should provide a basis for increasing the relevance of stress physiology and the efficiency of breeding programmes for development of drought-resistant genotypes.

Osmotic adjustment has been widely proposed as a plant attribute that confers adaptation to water stress. Genetic variation for osmotic adjustment among ten inbred sorghum lines was investigated by diallel analysis. Both general combining ability (GCA) and specific combining ability (SCA) were found to be significant (P < 0.05). The differences in patterns of GCA and SCA among the parents strong evidence that genetic variation for osmotic adjustment among the ten parents was controlled by more than one gene. Four parents (TAM422, Tx2813, Ajebsido and 40019), which phenotypically expressed high osmotic adjustment, expressed different patterns of GCA and SCA. This suggests that the genetic control for high osmotic adjustment differed among the four parents. Two major patterns of combining ability inheritance were identified among the high osmotic adjustment parents, one of these based on positive GCA inheritance (TAM422 and 40019) and the other on negative GCA inheritance (Tx2813 and Ajebsido). However, SCA influenced the expression of these patterns of inheritance and their expression was modified in specific parental combinations.

Pattern analysis, involving the joint use of classification and ordination, has been applied in plant breeding trials to find relationships among environments according to their discrimination among test genotypes (clones). However, rationalisation of the number of test environments from such analyses has often been subjective. A method which alleviates this subjectivity involves considering the blocks (experimental replicates) within environments as different environments and conducting a pattern analysis of blocks. The method is demonstrated using three sugarcane (Saccharum spp. hybrids) genotype × environment data sets, which contained clones representative of specific selection stages in the breeding program. The first data set contained 54 relatively unselected clones evaluated in two randomised complete blocks (replicates) at each of nine environments in south-eastern Queensland (Qld.) and north-eastern New South Wales (N.S.W.), Australia. The second data set contained 52 moderately selected clones, evaluated in two replicates at each of six environments in south-eastern Qld. The last data set contained information on six highly selected clones present in the same trials from which the second data set was collected. The two blocks within each environment were considered to be different environments and a pattern analysis of the blocks present in each data set was conducted. Results confirmed that an appropriate group number following hierarchical classification could be found for the first data set by truncating the hierarchy when both blocks from each environment were grouped together. For the other two data sets an appropriate group number was found when the blocks from five of the six environments grouped together. Reasons for the non-grouping of the blocks from that environment are discussed. Similarly, an indication of the proximity among blocks necessary for them to be considered similar was made by examining the distance between blocks from the same environment on the first three vectors from a principal component analysis.

Australian wheat breeding programs are currently evaluating and using advanced breeding lines from the International Centre for Wheat and Maize Improvement (CIMMYT). A more coordinated and efficient procedure for identifying useful lines for Australian breeding programs was evaluated by assessing the potential for exploiting correlated response to selection for grain yield between CIMMYT international and Australian evaluation trials. The genetic correlation, based on 40 common lines, between average line performance across 18 international and six Australian environments was high (0.813±0.049), indicating considerable potential for exploiting correlated response to selection for broad grain yield adaptation to the Australian environments for selection on broad adaptation in the international environments. The potential to select for specific grain yield adaptation to the individual Australian environments was investigated and individual international environments were identified which expressed a higher genetic correlation with the Australian environments than did average performance over all of the international environments. Therefore, opportunity exists for Australian wheat breeders to use the principles of indirect selection to select among advanced CIMMYT wheat lines on grain yield performance in international performance trials for both broad and specific grain yield adaptation to Australian environments prior to the introduction of this material to Australia. Since genotype by environment interaction for yield in the international trials is large and its causes are not well understood, selection for specific adaptation on the results from individual trials will require further investigation of the individual relationships.

Variation in maximum osmotic adjustment and desiccation tolerance were determined before anthesis for 21 lines of Sorghum bicolor (L.) Moench, selected on the basis of their capacity for osmotic adjustment and for their putative drought resistance. A reproducible and controlled imposition of gradual water stress was achieved by withholding water from plants grown in a controlled environment with a constant evaporative demand. Even though the rate of imposition of stress was lower in the second of two experiments, the maximum level of osmotic adjustment expressed by the lines was fairly consistent across both. The highest osmotic adjustment was 1.71±0.06 MPa for TAM422 in Exp. 2, while the lowest was 0.78±0.09 MPa for QL27 in Exp. 1. The difference in maximum osmotic adjustment between the highest and the lowest lines was 0.75 and 0.87 MPa, respectively, for Exp. 1 and 2. There was also variation in desiccation tolerance among the 20 lines in Exp. 2; 58% to 68% for lethal relative water content and −3.1 to − 3.9 MPa for lethal leaf water potential. Mean values of lethal realtive water content and lethal leaf water potential were, respectively, 62% and −3.4 MPa. Maximum osmotic adjustment was inversely related to desiccation tolerance; lethal relative water content and lethal leaf water potential increased linearly as maximum osmotic adjustment increased. Thus lines with high osmotic adjustment died at a higher relative water content and lower leaf water potential, than those with low osmotic adjustment. Despite their reduced tolerance of desiccation, lines with high osmotic adjustment survived 10 days longer. In both experiments, a high level of repeatability of line mean discrimination was identified for osmotic adjustment (0.75±0.15 and 0.96±0.09, for Exps. 1 and 2, respectively). The repeatability on a line mean basis across experiments was also high (0.84±0.07), as was the genetic correlation between the line means in both experiments (0.86±0.08). The high levels of repeatability suggest that the screening procedure developed in this study in robust, and that it could be used to evaluate the inheritance of osmotic adjustment in breeding populations.

Classification methodology is widely used by plant breeders to group environments on the results of regional evaluation trials to assist in selection among genotypes. To be effective, this strategy must be integrated with the theory of indirect selection. Environments which group together should reflect commonality of genotypic discrimination and therefore give rise to similar selection among genotypes. Four strategies for classifying environments were compared. These were based on untransformed and three forms of transformed data (coded, standardised and rank). The comparison assessed how effectively the groups of environments formed by using each transformation maximised the opportunity for exploiting indirect selection between environments within the same group relative to environments in other groups. The objective in this study was to identify groups of international environments, used by CIMMYT in its international nursery program, which gave high indirect response to selection for grain yield in six Australian environments. Generally the four classification strategies identified subsets of international environments for which selection gave a greater indirect response than that for selection on average performance across all of the international environments (35% to 94% on average over all Australian environments). Environmental classifications based on the standardised and rank transformations were generally superior to those based on the untransformed and coded transformations (46% on average over all Australian environments). The magnitude of this advantage differed between the Australian environments but was substantial for the two environments which expressed the greatest opportunity for exploiting indirect selection. These results have obvious and large implications for the use of classification methodology to structure regional testing regimes for plant breeding programs.

The principles of indirect selection can be used to assess the impact of genotype by environment interaction on selection strategies within breeding programs. The methodology can also be used to assess the opportunity for more efficient acquisition of new germplasm for a breeding program where common genotypes are evaluated in environments sampled by two breeding programs. However, the large number of environmental comparisons that is required complicates this process. To assist environmental comparison, the theory for indirect selection is related to classification methodology for grouping of environments in plant breeding experiments. To demonstrate the application of this methodology, classification was used to jointly group 18 international wheat evaluation environments conducted by the International Centre for Wheat and Maize Improvement (CIMMYT) as part of the 17th International Bread Wheat Screening Nursery (17th IBWSN) and six environments from a trial series conducted in Queensland, Australia using a set of 40 common genotypes. This process was used to identify a subset of international environments which improved indirect response for average grain yield performance in the Australian environments over that for selection on average grain yield performance over all 18 international environments. Similarly, classification was used to identify subsets of international environments which improved indirect response to selection in five of the six Australian environments. More efficient strategies for acquiring useful wheat germplasm for Australian wheat breeding programs from the CIMMYT program by extending this methodology are discussed.

To simplify the investigation of genotypic performance in large plant breeding trials, clustering techniques can be applied to reduce the number of responses that must be considered. However, the determination of an appropriate number of groups to represent such data is often subjective. Previous work suggested a procedure to determine the appropriate number of groups that should be selected to represent the data. The variability in patterns of performance that could be expected through experimental error at each environment was assessed using repeats of a check line (cultivar). The hierarchical classification was truncated when the repeats of the check line were partitioned into different groups. That procedure is assessed further by application to a data set consisting of five cultivars of sugarcane, each repeated six times within each of three blocks in six environments in south-east Queensland, Australia. In such a data set it is reasonable to assume that five groups, each consisting of all of the repeats of a particular cultivar, would be an appropriate summary of the data. As a more general approach, a method in which the information contained in the partitions, obtained from classifying different combinations of repeats within checks across environments, is used to infer the similarity in patterns of performance of the repeated check lines. The appropriate number of groups to summarize the data is determined from this information. Application of this approach identified the assumed underlying group number and group composition within the data set.

The investigation of genotypic performance in plant breeding trials is often complicated by genotype × environment (G×E) interaction. Clustering techniques can be applied to facilitate the investigation of genotypic performance when G×E interaction is a major contributor to the variation. However, the determination of an appropriate number of groups in such applications is often subjective. A simple and practical guideline for determining an appropriate group number in clustering performance-trial data is outlined. The method utilizes repeats of check lines (genotypes) which are grown in each block in each environment within the trial. Application of the method is demonstrated using a relatively small data set consisting of the grain yield of 15 wheat lines and four repeats of a check cultivar, each evaluated in 14 environments in Queensland, Australia. This guideline for group number determination can be used as a stopping rule in hierarchical classification. This approach also highlights the confidence that can be drawn from any particular partition and dramatizes the effects experimental error has on group composition.

Rainfed lowland rice is defined as rice growing in bunded paddies that do not have any access to irrigation water. This rice ecosystem is important, particularly in South and Southeast Asia. The area occupied by this ecosystem in Asia was 36.4 million ha and the average grain yield was 2.30 t ha-1 in 1991 (IRRI, 1997). The area was about 25% of the total cultivated area for rice. Fischer (1996) argued that by the year 2025 the production of rice must be increased by 70% over current levels in order to meet growing food demands. Scobie et al. (1993) estimated that of the extra rice production required by the year 2030, 21% would have to come from the rainfed lowlands. Sustainable genetic and agronomic improvements of the yield of rainfed lowland rice are required if this projected increase in productivity is to occur.