Temporal changes in Tasmanian devil genetic diversity at sites with and without supplementation

Management interventions for threatened species are well documented with genetic data now playing a pivotal role in informing their outcomes. However, in situ actions, such as supplementations (the release of individuals into an existing population of conspecifics) are often restricted to a single site to advise species-level management decisions. A considerable amount of research and management effort has been dedicated to conserving the endangered Tasmanian devil (Sarcophilus harrisii), offering a unique opportunity to investigate the temporal genetic consequences of supplementation at multiple sites, in comparison to outcomes observed in the absence management interventions. Using 1,778 genome-wide SNPs across 1,546 individuals, we compared four wild supplemented sites to four monitoring-only sites (not supplemented; control sites) over nine years (2014-2022). By the end of the study, genetic differentiation among supplemented sites had significantly decreased compared to among not-supplemented sites. We found statistically significant variation in patterns of genetic change over time between sites using linear mixed-effects modelling with random slopes. Investigating this among-site variation showed that three of the supplemented sites conformed to predictions that supplementations would have a positive impact on the genetic diversity of devils. We predicted there would be no change over time at our fourth site due to relatively high gene flow previously observed there, however, that site did not align with predictions, instead showing a decrease in genetic diversity and increase in relatedness. Amongst not supplemented sites, there was no consistent pattern of temporal genetic change, suggesting that devil sites across Tasmania are highly heterogeneous, likely reflecting variation in connectivity among sites, genetic drift, and non-genetic factors. Our study demonstrates that long term concurrent monitoring of multiple sites, including controls, is necessary to contextualize the influence of management interventions against natural fluctuations within a species system.

Methods
There is once dataset included in the Excel workbook:

A set of SNPs generated using DArTSeq, a form of reduced representation sequencing. Genotype SNP data were quality filtered using the dartR package (v 2.9.7) in Rstudio (v 4.3.1). We filtered out monomorphic loci and SNPs with low (lower threshold = 5) and high (upper threshold) mean read counts. Further filtered on a reproducibility of ≥95%, loci with a call rate of ≥70%, 0.001 minor allele frequency, individual call rate of ≥70%. Loci were filtered out if the SNP position was outside the trimmed sequence tag, secondary SNPs in the same sequence, or predicted to be sex linked. Data was subset to include individuals for the current study, monomorphic loci were removed here and refiltered with a call rate of ≥70%.