Combining Quantitative Trait Loci Analysis and an Ecophysiological Model to Analyze the Genetic Variability of the Responses of Maize Leaf Growth to Temperature and Water Deficit

Abstract

Ecophysiological models predict quantitative traits of one genotype in any environment, whereas quantitative trait locus (QTL) models predict the contribution of alleles to quantitative traits under a limited number of environments. We have combined both approaches by dissecting into effects of QTLs the parameters of a model of maize (Zea mays) leaf elongation rate (LER; H. Ben Haj Salah, F. Tardieu [1997] Plant Physiol 114: 893-900). Response curves of LER to meristem temperature, water vapor pressure difference, and soil water status were established in 100 recombinant inbred lines (RILs) of maize in six experiments carried out in the field or in the greenhouse. All responses were linear and common to different experiments, consistent with the model. A QTL analysis was carried out on the slopes of these responses by composite interval mapping confirmed by bootstrap analysis. Most QTLs were specific of one response only. QTLs of abscisic acid concentration in the xylem sap colocalized with QTLs of response to soil water deficit and conferred a low response. Each parameter of the ecophysiological model was computed as the sum of QTL effects, allowing calculation of parameters for 11 new RILs and two parental lines. LERs were simulated and compared with measurements in a growth chamber experiment. The combined model accounted for 74% of the variability of LER, suggesting that it has a general value for any RIL under any environment.In an agricultural context, a plant that tolerates water deficit can produce a maximum harvested biomass under moderate water deficits. Involved mechanisms are not necessarily common with those underlying the ability of cells to survive tissue dehydration (e.g. Cushman and Bohnert, 2000; Seki et al., 2001). In maize (Zea mays), moderate water deficits usually cause no appreciable decrease in leaf water status because of an efficient stomatal control combining hydraulic and chemical messages (Tardieu and Davies, 1993;Wilkinson et al., 1998). A similar combination of messages allows maize plants to dramatically reduce leaf elongation rate (LER) under moderate water deficits sensed either in the soil or in the air, before that leaf water status is appreciably altered (Sharp et al., 2000;Tardieu et al., 2000). Reductions in stomatal conductance and in leaf expansion decrease transpiration rate, thereby saving soil water and maintaining leaf water potential at high values. They also reduce photosynthesis, growth, and yield, so optimum tolerance strategies cannot be common to different climatic scenarios. Plants with steepest responses might be most adapted to scenarios with most severe water deficits, whereas maintenance of growth and photosynthesis under deficit might be appropriate for scenarios with milder deficits. Therefore, identification of sources of variability in the responses to water deficit is necessary for designing plants adapted to a given climatic scenario.We aimed to identify and analyze the genetic variability of responses of leaf elongation to water deficits caused eit...

Multiple cropping systems as drivers for providing multiple ecosystem services: from concepts to design

“…In this second strategy, the development of ideotypes can also benefit from QTL detection or genomic selection methodologies, with markers used to tag the genomic areas controlling targeted traits or parameters in ecophysiological models, e.g., vernalization or photoperiod parameters in models predicting plant earliness (Bogard et al 2014). In this way, marker-based modeling can be used to predict key parameters of ecophysiological models and then build genotypes with optimum parameter combinations (Letort et al 2008;Reymond et al 2003;Prudent et al 2011) through the selection of the best allelic combinations (Hoogenboom et al 2004;Chenu et al 2009). For all these reasons, we believe that a trait-based approach is a promising strategy for building ideotypes for multiple cropping systems, even though the main bottlenecks are still the availability of cultivar trait values and dedicated ecophysiological models.…”

Toward a Bayesian procedure for using process-based models in plant breeding, with application to ideotype design

“…So modellers may want to translate plant parameters into equivalent genes or collections of genes. This was not attempted, as discussed in the Introduction, because for most parameters the link with DNA sequences or QTLs is still tenuous, with some notable exceptions (Reymond et al 2003;Yin et al 2004). Messina et al (2009) andCooper et al (2014) sketch a future of plant breeding in which unravelling the genetic basis of modelled plant properties will support high-throughput phenotyping, which they consider essential for the scaling up of breeding programmes to larger numbers of genotypes.…”

“…Similarly, Reymond et al (2003) were able to relate QTLs to the parameters that govern response of maize leaf expansion to temperature and water status, and Dong et al (2012) developed a regulatory network model for over 30 genes controlling flowering time in maize. These various studies were largely based on recombinant inbred lines, thus simplifying the genetic differences between genotypes, and the variety of environmental conditions was also limited.…”

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

“…When z is time related, the error term δ ij demands careful modelling of possible autocorrelations between observations at short intervals. Two illustrative examples of physiological modelling of response curves followed by QTL mapping of the estimated curve parameters are Reymond et al (2003) for linear parameters and Yin et al (2005) for non-linear parameters.…”

“…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…”