Data: Using environmental DNA and occupancy modeling to estimate rangewide metapopulation dynamics
Kinziger, Andrew (2020), Data: Using environmental DNA and occupancy modeling to estimate rangewide metapopulation dynamics, Dryad, Dataset, https://doi.org/10.5061/dryad.vdncjsxs9
We demonstrate the power of combining two emergent tools for resolving rangewide metapopulation dynamics. First, we employed environmental DNA (eDNA) surveys to efficiently generate multi-season rangewide site occupancy histories. Second, we developed a novel dynamic, spatial multiscale occupancy model to estimate metapopulation dynamics. The model incorporates spatial relationships, explicitly accounts for non-detection bias and allows direct evaluation of the drivers of extinction and colonization. We applied these tools to examine metapopulation dynamics of endangered tidewater goby, a species endemic to California estuarine habitats. We analyzed rangewide eDNA data from 190 geographically isolated sites (813 total water samples) surveyed from two years (2016 and 2017). Rangewide estimates of the proportion of sites that were occupied varied little between 2016 (0.52) and 2017 (0.51). However, there was evidence of extinction and colonization dynamics. The probability of extinction of an occupied site (0.106) and probability of colonization of an unoccupied site (0.085) were nearly equal. Stability in site occupancy proportions combined with nearly equal rates of extinction and colonization suggests a dynamic equilibrium between the two years surveyed. Assessment of covariate effects revealed that colonization probability increased as the number of occupied neighboring sites increased and as distance between occupied sites decreased. We show that eDNA surveys can rapidly provide a snapshot of a species distribution over a broad geographic range, and when these surveys are paired with occupancy modeling, can uncover metapopulation dynamics and their drivers.
We analyzed rangewide eDNA data from 190 geographically isolated sites (813 total water samples) surveyed from two years (2016 and 2017) using a spatial, dynamic occupancy model.
Site occupancy, covariate data, neighborhood structure, and data file definitions are provided in the files:
The R code for implementation of the novel dynamic, spatial multiscale occupancy model is provided in:
To run place all files (e.g., data files and R code) in your working directory and open and execute "AnalysisOfGobyData.R" in R.