Climate variables are not the dominant predictor of Arctic shorebird distributions

200m, approximately covering the plot itself, to 10km, increasing by increments of 100m up to 2500m and by increments of 500m thereafter. We created a series of logistic regression models relating the occupancy of each of the 17 shorebird species to each predictor at each spatial scale. Then for each predictor, we identified the scale of effect for that species as the extent where the model had the lowest AIC value. In the models described below, we entered the snowmelt timing and land cover predictors at their scales of effect. We used path analysis to test whether the effects of the additional variables on habitat suit- ability for shorebirds were largely explainable as indirect effects of climate. Path models were fit using the R package piecewiseSEM [37]. We fit a path model for each of the 17 shorebird species, following the causal structure illustrated in Fig 2. We also included correlated error terms among all climate variables, and among all land cover variables. All models included an offset for plot size. We used a test of directed separation (d-sep) to evaluate the goodness-of-fit for our hypothesized path model, assessing the conditional independence between pairs of var- iables in the model. Significant d-sep test results indicated associations between substrate and climate, and substrate and shorebird presence, so we included these links as correlated error terms as well. For each species model, the median standardized effects and 95% confidence intervals were estimated using nonparametric bootstrapping from 10,000 resamples using the

Fig 3. Map of survey locations. 2336 plots were surveyed for breeding shorebirds across northern Canada, 1994–2018. Reprinted from ArcGIS under a CC BY license, with permission from Esri, original Copyright 2022 Esri (Basemaps supported by Esri, HERE, Garmin, OpenStreetMap contributors, and the GIS User Community). https://doi.org/10.1371/journal.pone.0285115.g003 PLOS ONE Climate variables are not the dominant predictor of Arctic shorebird distributions PLOS ONE | https://doi.org/10.1371/journal.pone.0285115 May 17, 2023 6 / 14