Ranking populations for genetic conservation using a new index

Background/Question/Methods

Individual taxa can be prioritized for conservation based on their expected contribution to the total diversity of some set (e.g. populations to species or species to higher taxa). One such measure from cooperative game theory is called the Shapley index (formally equal to “evolutionary distinctiveness”), which has been applied to phylogenetic trees (see, e.g. edgeofexistence.org) and genetic networks (see, e.g., work by Volkmann et al. in 2014). Very recent theoretical work (by Wicke et al. in 2021) has formally extended the Shapley index to the underlying measured features directly. We applied this new “Feature Diversity” index to genomic data by treating derived SNPs as features and population frequencies of these SNPs as feature weights, and then compared the Shapley values using this direct method with the pre-2021 approach of considering networks inferred from genetic distances.

Results/Conclusions

As one initial test case, we considered 437 SNPs across 14 populations of Sitka Spruce on the west coast of North America from Alaska to California (from a study by Holliday et al., 2010). The ranked Shapley index based on genetic networks and the ranks using the Feature Diversity index were only weakly correlated across the 14 populations (Kendall’s Tau = 0.32), though they did agree on 3 of the top 5 most distinct populations from preliminary analysis. Future work will probe this difference and expand the comparison to other species and alternative characterizations of population differentiation, e.g. by comparing prioritizations using SNPs linked to local adaptation vs. random SNPs (following the work by Fernandez-Fournier et al., 2021). Our goal is to offer conservation geneticists a transparent and simple-to-use tool for prioritization.