Climate variables are not the dominant predictor of Arctic shorebird distributions

One hypothesis suggests that climate is the dominant factor influencing species’ distribu- tions at large spatial scales [6, 7]. Climate influences species’ distributions directly, for example through thermal tolerances, and indirectly, for example through climate’s influence on land cover (Fig 1). Interestingly, many of the papers that invoke this hypothesis are studies of future species distribution based on climate-only models [i.e. 8]. One reason for this practice is prag- matic: often there are no future predictions available for the other potential covariates of distri- bution such as land cover [9]. However, these papers also often argue that at large spatial extents, such as entire ranges for widespread species, the direct and indirect effects of climate encompass the majority of the influences on species’ distributions [10]. ‘Climate-only’ models are often evaluated using a threshold, such as a model with an AUC of at least 0.7 has “good” explanatory power, and usually climate-only models exceed the threshold [i.e. 11, 12]. A competing hypothesis suggests that habitat suitability models need to be more compre- hensive because there can be additional important factors shaping species distributions that may not be well described as an indirect effect of climate, as is often assumed in climate-only models. In contrast to the future-oriented models described above, papers estimating current species distributions often include a wide range of predictors, many of which are non-climatic,

relating instead to resources and habitat structure [13]. Studies aimed at estimating current habitat suitability typically compare multiple models to identify a best model, which is rarely a climate-only model [i.e. 14–16]. In particular, biotic interactions may be important even at large spatial extents, not just at local scales [17]. For example, many species require the pres- ence of specific plants as food or hosts to facilitate their presence in a region [18]. Precipitation is often a poor proxy for the water available to plants, as topography and soil substrate control how precipitation translates into soil moisture [19]. Soil predictors such as pH and nutrients Fig 1. Conceptual relationships between the drivers of species occupancy. Illustration of the common assumption that, over large spatial extents, climate predictors drive the probability of species occupancy. Such climate-only models imply that the total effects of climate predictors, i.e., both their direct effects (a) and their indirect effects through their effects on additional predictors such as land cover (b*c), are much greater than the direct effects of these additional predictors (c). The underlying assumption is that climate is the main driver of variation in the predictors that influence species occupancy, and so it is only necessary to include climate predictors in species occupancy models. https://doi.org/10.1371/journal.pone.0285115.g001 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 2 / 14 Engineering Research Council of Canada - https:// www.nserc-crsng.gc.ca/index_eng.asp PAS and JR - Institutional funding - Environment and Climate Change Canada - https://www.canada.ca/ en/environment-climate-change.html CA - 518225- 2018 - National Science and Engineering Research Council of Canada - https://www.nserc-crsng.gc. ca/index_eng.asp CA - W. Garfield Weston Award in Northern Research (Doctoral) 2019 – 2020 - The W. Garfield Weston Foundation. - https:// westonfoundation.ca/northern-science-and- knowledge/ The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.