Trait physiology and crop modelling as a framework to link phenotypic complexity to underlying genetic systems

Abstract

New tools derived from advances in molecular biology have not been widely adopted in plant breeding for complex traits because of the inability to connect information at gene level to the phenotype in a manner that is useful for selection. In this study, we explored whether physiological dissection and integrative modelling of complex traits could link phenotype complexity to underlying genetic systems in a way that enhanced the power of molecular breeding strategies. A crop and breeding system simulation study on sorghum, which involved variation in 4 key adaptive traits-phenology, osmotic adjustment, transpiration efficiency, stay-green-and a broad range of production environments in north-eastern Australia, was used. The full matrix of simulated phenotypes, which consisted of 547 location-season combinations and 4235 genotypic expression states, was analysed for genetic and environmental effects. The analysis was conducted in stages assuming gradually increased understanding of gene-to-phenotype relationships, which would arise from physiological dissection and modelling. It was found that environmental characterisation and physiological knowledge helped to explain and unravel gene and environment context dependencies in the data. Based on the analyses of gene effects, a range of marker-assisted selection breeding strategies was simulated. It was shown that the inclusion of knowledge resulting from trait physiology and modelling generated an enhanced rate of yield advance over cycles of selection. This occurred because the knowledge associated with component trait physiology and extrapolation to the target population of environments by modelling removed confounding effects associated with environment and gene context dependencies for the markers used. Developing and implementing this gene-to-phenotype capability in crop improvement requires enhanced attention to phenotyping, ecophysiological modelling, and validation studies to test the stability of candidate genetic regions.

Accounting for Variability in the Detection and Use of Markers for Simple and Complex Traits

“…For the wheat examples, we simulated selection using QuLine, a breeding module used to simulate wheat-breeding programmes (Wang et al 2003), and to predict cross performance for quality traits (Wang et al 2005). For sorghum, the original simulations were done using a proprietary breeding module (Hammer et al 2005). More details of the examples are given in the Results section.…”

“…Chapman et al (2003) quantified how bias in the sampling of environments by the breeding programme (because of variability in rainfall between successive seasons) reduced the efficiency of selection, through the generation of substantial genotypeby-environment interaction. Using the same dataset, Hammer et al (2005) showed how even 'simple' combinations of traits across genotypes and environments could easily confound detection of QTL associated with yield.…”

“…In this situation, many sources of error exist, which include: experimental error in measuring the phenotype during the QTL study (trait heritability); error in selection of the marker or markers for the QTL (poor linkage); lack of observation (or knowledge) about how 'sub-traits' combine physiologically to affect the trait of interest; and, most critically, lack of knowledge of the gene action of the 'unexplained' variance in the QTL study. Using simulation, Chapman et al (2003) and Hammer et al (2005) illustrated that when simple additive gene action was defined for four sub-traits ('trait parameters') in a sorghum-cropping system, complex gene-by-gene-by-environment interactions could still be generated for expression of yield. Models of gene action (i.e.…”

“…While there has been some application of simulation approaches to examine the value of QTL for complex traits in Australian sorghum (Hammer et al 2005), marker technology is still being developed for application in that breeding programme, but is focusing on utilization for QTL associated with complex traits such as midge resistance and stay-green (see Hammer et al,Chapter 5). Markers for single-gene/single-trait applications have been used in wheat-breeding programmes in Australia for over 10 years, e.g., Ogbonnaya et al (2001) and Eagles et al (2001).…”

Modelling the Genetic Basis of Response Curves Underlying Genotype × Environment Interaction

“…Even the inclusion of response surfaces in various dimensions does not present statistical-technical problems, although the number of environments necessary for sufficiently precise estimation of the increasing number of regression parameters will not often be reached in plantbreeding programmes. When good explicit environmental characterizations are available, it is often preferable to model the genotypic responses by parametric linear and non-linear regression functions based on physiological insights, control equations (Reymond et al 2003;Tardieu 2003;Tardieu et al 2005) or meta-mechanisms (Hammer et al 2005), instead of working with polynomial approximations to these non-linear functions. A general expression for non-linear genotypic responses in one dimension is…”