Pearl millet genomic vulnerability to climate change in West Africa highlights the need for regional collaboration

Abstract

Climate change is already affecting agro-ecosystems and threatening food security by reducing crop productivity and increasing harvest uncertainty. Mobilizing crop diversity could be an efficient way to mitigate its impact. We test this hypothesis in pearl millet, a nutritious staple cereal cultivated in arid and low-fertility soils in sub-Saharan Africa. We analyze the genomic diversity of 173 landraces collected in West Africa together with an extensive climate dataset composed of metrics of agronomic importance. Mapping the pearl millet genomic vulnerability at the 2050 horizon based on the current genomic-climate relationships, we identify the northern edge of the current areas of cultivation of both early and late flowering varieties as being the most vulnerable to climate change. We predict that the most vulnerable areas will benefit from using landraces that already grow in equivalent climate conditions today. However, such seed-exchange scenarios will require long distance and trans-frontier assisted migrations. Leveraging genetic diversity as a climate mitigation strategy in West Africa will thus require regional collaboration.

A high-resolution downscaled CMIP5 projections dataset of essential surface climate variables over the globe coherent with the ERA5 reanalysis for climate change impact assessments

“…CDF-t was first developed for wind values and is now referenced in dozens of peer-reviewed publications to downscale different sets of data and variables (e.g. [11] , [12] , [13] ). QM methods relate the cumulative distribution function of a climate variable at large scale (e.g., from the GCM) to the CDF of the same variable at a local scale (e.g., from the reanalysis).…”

High-resolution downscaled CMIP5 projections dataset of essential surface climate variables over the globe coherent with ERA5 reanalyses for climate change impact assessments

“…CDF-t was first developed for wind values and is now referenced in dozens of peer-reviewed publications to downscale different sets of data and variables (e.g. [10][11][12]). QM methods relate the cumulative distribution function of a climate variable at large scale (e.g., from the GCM) to the CDF of the same variable at a local scale (e.g., from the reanalyses) and are increasingly popular in climate applications although bias correction methods have received criticism (e.g.…”

What processes must we understand to forecast regional-scale population dynamics?

“…Researchers have begun to incorporate genetic variation into regional population prediction. Some researchers have fit models that identify genotypes associated with specific current environments, assuming local adaptation, and then used predicted future environments to assess mismatch between local genotypes and their future environments [38][39][40][41] (reviewed by Capblancq et al [15]). Additionally, researchers have fitted environmental response models using geographically restricted subsets of populations, presumably accounting for geographic genetic variation, to project distributions under future conditions [42][43][44].…”

“…After all, these models are being developed for prediction in a future of novel conditions. Recently, genetic models of environmental response and adaptation have been used in out-of-sample predictions of individual or population performance in response to environmental stressors in both wild and agricultural species [38,41,[45][46][47][48]. These have generally had modest success, for example, in predicting relative genetic variation in change in performance [46,47] or population change [38] with accuracy (predicted versus observed correlation) ≈ 0.3.…”