Environmental data provide marginal benefit for predicting climate adaptation

Populations of natural and cultivated plant and animal populations will be
affected by more extreme climate events such as drought and flooding in
the future. We explore whether characterization of the
environment-of-origin of each accession in a large sample of traditional
maize germplasm can be used to accelerate conservation and breeding
efforts for adaptation. We compare the utility of genotype and
environmental data for predicting fitness of individuals in a number of
common garden trials. We find that environment-of-origin data and genome
scans for loci that associate with abiotic environmental variables provide
surprisingly little benefit to prioritizing accessions for improvement,
despite clear evidence of environmental adaptation in these accessions.
These results provide important practical insight into the use of gene
banks for climate adaptation. Methods include prediction of environmental
variables from genotyping-by-sequencing data, environmental GWAS (envGWAS)
to identify loci associated with climatic gradients, and genomic
prediction of yield component traits. Data includes metadata associated
with phenotypic field trials, transformed climate-of-origin data
associated with accessions used for study, output from MegaLMM for genomic
prediction of environment (GPoE) and covariance matrices used for envGWAS,
results from envGWAS and phenotypic GWAS, models run for transfer plots,
and genomic prediction results. Additional raw data available via CIMMYT
and reproducible code via Github can be found under Access information or
at the associated publication "Environmental data provide marginal
benefit for predicting climate adaptation."