Monitoring biodiversity & genetic variation
Studying species records from the last centuries, we have learned that many species have already lost the race against extinction. Some of these extinctions are direct influence from us, e.g. habitat destruction or hunting, but others are rather indirect, i.e. species cannot keep up adapting to changing atmospheric and oceanic conditions. Species adapt and evolve by the process of natural selection over genetic variation. However, this is harder as species shrink geographically and lose genetic diversity, commencing a vicious extinction circle. To work against this, we need to have a good understanding on the past and current status of a species genetic diversity and geographic distributions. In the lab, we work with public databases of genetic information (NCBI), species sighting records (GBIF.org, inaturalist.org, Fig. right, top), and images of the Earth and maps of weather records (Landsat missions, Google Earth Engine, Fig. right, bottom). We integrate these data using deep learning approaches to generate geographic visualizations of species composition and diversity and their changes over time, to find genetic coldspots and recent diversity losses.
Records of historic specimens such as herbaria are great "genetic snapshots" in time that allow us to track such changes in time and learn about evolutionary principles. For example, in the past we have studied a 400-year-old lineage of A. thaliana that was isolated in North America. We identified over 5,000 new mutations, some of which generated novel morphological differences likely related to adaptation (Exposito-Alonso et al. 2018 PLOS Genetics). This shown that even large organisms such as plants could, even at contemporary time-scales, rapidly evolve and adapt from new mutations.
Predicting genetic evolution to climate change
The field of evolutionary biology is transforming from a historical science, where one studies past patterns of diversification to learn biological principles, to a predictive science, where we apply those principles to forecast the future. In the past, we have pioneered the development of bioinformatic tools to study how much climate will change locally for different populations of a species, and how many potentially useful genetic variants they harbor to face new climates (Exposito-Alonso et al. 2018 Nature Ecology Evolution, 2019 bioRxiv). The next challenge to realistically project the fate of populations is to account for the stochasticity of demography during adaptation, and for the myriad of complex combinations of adaptive genetic variants that give the wide gradient of plant fitness values observed in nature.
In order to empirically test models of short-term responses of organisms to climate, we created a community of about 100 researchers to conduct large-scale real-time evolution experiments in 45 locations ranging from the Mediterranean to Scandinavia, and California to Israel. This crowd-sourced project, called “Genomics of rapid Evolution to Novel Environments” (GrENE-net.org, Fig. left) follows an evolve and resequence approach for live monitoring of genetic evolution over time. Combining this data with predictive population genetic models will allow me to robustly assess the predictability of rapid plant evolution to climate change.
Polygenic engineering of adaptive variants
Genetically rescue plants from extinction is challenging. I think the reasons are twofold: Overall plant performance in natural conditions is expected to be determined by hundreds, if not thousands, of genetic players (most of which we don’t know!). In addition, the physiological pathways relevant for climate adaptation known for one plant species might may be different in other species. This realization lead us to use our evolutionary and probabilistic understanding of adaptation to develop molecular tools to speed up adaptation. Because new mutations are the ultimate fuel of adaptation, we are leveraging CRISPR as a mutagenesis technology informed by powerful deep learning models that can tell us what mutations in the genome would be most likely to provide an advantage under future stressful environments. By growing engineered plants in drought conditions in the lab, we can test our ability to predict fitness effects of mutations as well as evaluate whether genetic engineering strategies could be potentially used in the future to prudently aid the conservation of crops or wild plants in critical danger.