Dissecting drought tolerance in winter wheat using phenotypic and genetic analyses of agronomic and spectral traits

Worldwide, wheat (Triticum aestivum L.) is cultivated on more land than any other food crop. In 2013 wheat was grown on more than 220 million hectares worldwide, which is a larger area than the entirety of Mexico. Part of the global success of wheat can be attributed to its adaptability to diverse environmental conditions, including regions with limited water availability.;The United States is the largest exporter of wheat, and in recent years has exported 20--30% or more of its total production. Much of the wheat grown in the United States is cultivated under rainfed conditions, including regions across the Great Plains that are primarily planted to hard winter wheat. However, grain yield can be severely affected by water stress, and future climate projections predict drought will become more frequent and more severe. Therefore, it is important to characterize drought response and better understand genetic variation and genetic mechanisms of drought tolerance in winter wheat present in the U.S. Great Plains hard winter wheat.;This study used a collection of 299 hard winter wheat entries, designated the Triticeae Coordinated Agricultural Project Hard Winter Wheat Association Mapping Panel (HWWAMP), representative historic lines, recent cultivars, and experimental breeding lines present across the U.S. Great Plains. The entries were evaluated at a total of 11 Great Plains environments during 2011--2012 and 2012--2013. These environments include four Colorado environments (paired water-stressed and non-stressed treatments in Greeley in 2011--2012 and Fort Collins in 2012--2013) with detailed phenotypic data for many agronomic traits, and seven environments in other states with data limited to heading date and grain yield.;The objectives of this study were to 1) determine allelic variation present in major developmental genes known to affect the timing of the developmental sequence and therefore adaptability, and estimate the effects of variants on heading date; 2) estimate the extent of variation and phenotypic plasticity of heading date in a range of environments representative of the U.S. Great Plains; 3) evaluate effects of water stress on grain yield and other agronomic traits, and identify underlying genomic regions affecting these drought responses, using side-by-side water-stressed and well-watered environments grown in two years; and 4) evaluate the effectiveness of water-based spectral indices calculated from hyper-spectral canopy reflectance measurements to characterize drought stress in the field.;In wheat and other small grain cereals, heading refers to the developmental stage where the spike has fully emerged from the flag leaf sheath. Heading date reflects genotypic 'earliness' and is important for regional adaptability of wheat. At heading, the developing spikelets and their sensitive reproductive structures become more exposed to changing environmental conditions, such as periods of cold, heat, or drought stress. Stress at heading and anthesis (which follows several days later) can have severe effects on grain yield.;The developmental sequence, including heading date, is affected by the vernalization and photoperiod pathways. Semi-dwarf alleles at the reduced-height genes also have an effect on the timing of plant development. We genotyped candidate genes at vernalization (Vrn-A1, Vrn-B1, and Vrn-D1), photoperiod (Ppd-B1 and Ppd-D1 ), and reduced-height (Rht-B1 and Rht-D1 ) loci using polymerase chain reaction (PCR) and Kompetitive Allele Specific PCR (KASP) assays. The main effects and two-way interactions of alleles at these loci explained an average of 44% of variation in heading date across nine environments. Most of the variation was explained by Ppd-B1, Ppd-D1, and their interaction. The photoperiod sensitive and insensitive alleles were present in our germplasm in large proportions for both Ppd-B1 and Ppd-D1, however, the sensitive alleles have been decreasing over time and are more common in germplasm from the northern than central or southern regions of the U.S. Great Plains.;There was significant (P < 0.001) genotype-by-environment interaction for heading date and growing degree-days to heading among all 11 environments. Phenotypic plasticity describes the range of possible phenotypes observed for one genotype, given different environmental conditions. We estimated phenotypic plasticity of growing degree-days to heading (GDDP) and yield for each entry, and found there was variation in our germplasm for both. We found GDDP to be negatively associated with yield (r=-0.58, P<0.001), and thus detrimental in the germplasm and environments evaluated. Greater yield plasticity was associated with increased maximum (r=0.80, P <0.001) and minimum ( r=0.33, P<0.001) grain yield across environments, indicating it was a favorable trait. Over time GDDP has decreased and yield plasticity has increased, which suggests these are possible traits that could be targeted for selection.;In the Colorado environments, grain yield was reduced by similar amounts under water stress in 2011--2012 (48%) and 2012--2013 (46%), even though water stress occurred during different periods of the two growing seasons. In 2011--2012 stress occurred before anthesis and primarily reduced grain yield by limiting biomass and tillering, and producing fewer total spikelets, fewer fertile spikelets, and fewer kernels per spike. In 2012--2013 stress occurred during grain filling and affected yield primarily by reducing kernel size. We conducted genome-wide association studies on agronomic traits in individual environments, and combined across different combinations of environments, and detected nearly 250 significant marker--trait associations for 15 agronomic traits. Most significant marker--trait associations were only detected for a single trait in one environment, had modest allelic effects, and explained a small proportion of total phenotypic variation. However, associations for kernel number explained up to 29% of variation in one environment, and associations for the proportion of fertile spikelets were stable across multiple environments.;We measured canopy spectral reflectance using a hyper-spectral radiometer in 2012--2013 and calculated spectral indices previously shown to be associated with plant water status. There was substantial spatial--temporal variation across each sampling date, which contributed to a lack of significant differences in index values among genotypes. However, values of normalized water indices 1 (NWI-1), 3 (NWI-3), and 4 (NWI-4) varied gradually among developmental stages. Changes in index values were especially pronounced under water stress, when the most extreme values coincided with the period of most water stress.;In summary, there is substantial variation for agronomic and phenological traits that affect drought tolerance or susceptibility. Variation in the timing of developmental stages, such as heading date, can confer regional adaptability. Introducing additional allelic diversity at photoperiod loci could enable finer adaptation under current or future climate scenarios. Alternatively, selecting for reduced GDDP and/or increased grain yield plasticity could result in greater yields under varying environmental conditions. We did not find evidence supporting use of canopy spectral reflectance as a selection tool, but spectral traits might be useful to monitor changes in plant water status, especially if sources of spatial and temporal variation are reduced, such as by using a sensor with an active light source or taking simultaneous estimates from an aerial vehicle.