Characterizing Drought Tolerance and Agronomic Traits in Wheat Using Canopy Spectral Reflectance.

Plant breeding are benefiting from impressive advances in marker and data technologies that are simultaneously increasing efficiency and reducing costs of genotyping. These resources include genetic markers, sequence data, and many other genetic and genomic resources. However, ability to reliably, rapidly, and inexpensively phenotype germplasm in the field is lagging behind and is a major constraint. Additionally, most genetic gain in yield has been made from empirical selections based on yield per se, and a handful of other key agronomic traits. Genetic gain is typically highest under favorable conditions, although many crops are grown in stressed environments. Canopy spectral reflectance is a tool to rapidly, inexpensively, and non-destructively phenotype germplasm; and could be a useful technique for indirect selection of yield in stressed conditions. 

We measured canopy spectral reflectance on 299 winter wheat varieties in 2013 using an OceanOptics Jaz spectrometer. The population consists of the Triticeae Coordinated Agricultural Project (TCAP) hard winter wheat association mapping panel and contains varieties derived in the U.S. Great Plains between 1874 and 2010. The experiment was grown in side-by-side water stressed (“dry”) and non-stressed (“wet”) treatments in Fort Collins, CO using a modified augmented design. CSR measurements were taken with a sensor optimized for the near infrared (590 - 1240 nm) on five dates in the dry treatment, and seven occasions in the wet treatment. Detailed notes on plant phenology were also taken throughout the season, so we were able to assign each plot to one of seven developmental stages (tillering and stem expansion, booting, heading, anthesis, grain filling, late grain filling, and maturity) on each sampling date. 

We calculated several published spectral indices previously reported to be associated with plant water status from each spectral measurement. The water index (WI) and normalized water indices 1, 3, and 4 (NWI-1, NWI-3, NWI-4) had significant correlations with grain yield in the dry treatment when data was pooled across all five sampling dates. The normalized water index 2 (NWI-2) was not significantly associated with yield when pooled across sampling dates. Spectral indices were not strongly correlated to yield in the wet treatment. The strength of the phenotypic correlation between spectral indices and agronomic traits varied gradually with plant development. The strongest correlations for all indices in the dry treatment were observed when spectral measurements were collected at flowering. The strongest correlation with yield was from NWI-4 sampled at flowering, which explained 58% of the variability in yield.

Genetic correlations were also strong between spectral indices across sampling dates in the dry treatment, and averaged 0.52 for WI, -0.53 for NWI-1, -0.43 for NWI-2, -0.53 for NWI-3, and -0.55 for NWI-4. Broad-sense heritabilities were also calculated for each index and sampling date, and ranged from 0% (when sampled on 5/31/13) to 64% (for WI sampled on 6/19/13), but averaged between 32 - 55% for each index. High genetic correlation and moderate heritability suggest using canopy spectral reflectance as an indirect selection method might be effective at increasing yield under some stressed conditions.