How Accurate is Genomic Prediction Across Wild Populations?

Abstract

Evolutionary ecology seeks to understand causes and consequences of evolutionary changes across time and space, and genomic data present novel opportunities to investigate these processes. Genomic prediction - that is, predicting individual genetic values from high-density marker data - has revolutionized breeding programs and medical genetics. In wild populations, however, genomic prediction has been used in only a handful of studies, and only within populations. There is still a lack of applications that predict across wild populations, which could provide answers to questions related to spatially varying evolutionary processes, such as local adaptation. A severe challenge for across-population genomic prediction, however, is the decrease in accuracy when models are trained on data from one population and predict genetic values in another. Here, we used genomic prediction to predict across wild house sparrow populations, and compared the accuracy to within-population prediction. Predictions across populations were less accurate and more variable than within populations. We also highlighted limitations of the current theory for general genomic prediction accuracy, and related across-population accuracy to several population differentiation measures. Our results underline the necessity of understanding why genomic prediction currently performs poorly across populations, and of developing methods that exploit genomic data in novel ways.