Dakota skipper distribution model for North Dakota, South Dakota, and Minnesota aids conservation planning under changing climate scenarios

Abstract

Species distribution models are useful conservation planning tools for at-risk species, especially if they are linked to planning efforts, conservation delivery, and a changing environment. The Dakota skipper (Hesperia dacotae) is an endemic butterfly of mixed and tallgrass prairie of the northern Great Plains that is listed as federally threatened in the United States and Canada. We modeled broad-scale habitat suitability for the Dakota skipper by relating occurrence observations collected via non-probabilistic population surveys and a stratified sample of pseudo-absences to environmental predictors using a machine learning approach (i.e. Random Forest classification model). Predictors were summarized at two local scales and one landscape scale to reflect a potential spatial hierarchy of settlement responses. We used recursive feature elimination to select the top 25 covariates from a suite of predictor variables related to climate, topography, vegetation cover, biomass, surface reflectance, disturbance history, and soil characteristics. The top model included six bioclimatic, one soil, and 18 local- and landscape-scale vegetation variables and indicated an association with undisturbed grasslands with higher perennial grass and forb cover and biomass. The model performed well, with kappa and AUC estimates of 0.92 and 0.99, respectively, for 20% of data withheld for validation. To understand how climate change might affect Dakota skipper distribution, we applied the model using future 30-year bioclimatic predictions. Predicted suitable habitat declined and the climate envelope associated with Dakota skipper occurrence shifted north into Canada. While it is unknown to what degree the bioclimatic relationships in the model are biologically meaningful or are simply correlative with our non-probabilistic sample of occurrences, our results present an urgency to improve data collection for Dakota skipper populations and better understand climatic relationships, as climate change could have profound effects on populations and conservation planning. Regardless of climate or model uncertainty, our results demonstrate the importance of maintaining sufficient quantities and quality of grass on the landscape to support populations of Dakota skipper.

Comparing Two Species Distribution Models Using Satellite Only and Ready-Made Environmental Variables for the Dakota Skipper (Hesperia dacotae), Interlake Region of Manitoba, Canada

“…Four recent SDMs have been created for the Dakota skipper, however, all four used a mixture of common and satellite imagery derived environmental variables (Table 1). Three of the studies were situated in regions outside of the Manitoba Interlake area (Post van der Burg et al 2020;Seidle et al 2020;Barnes et al 2024). The study by Seidle et al (2020) was created for a region in south-western Saskatchewan and the study by Barnes et al (2024) Post van der Burg et al (2020) was applied to three and four US states, respectively.…”

“…Three of the studies were situated in regions outside of the Manitoba Interlake area (Post van der Burg et al 2020;Seidle et al 2020;Barnes et al 2024). The study by Seidle et al (2020) was created for a region in south-western Saskatchewan and the study by Barnes et al (2024) Post van der Burg et al (2020) was applied to three and four US states, respectively. In both cases the habitat is likely substantially different than that found in the Interlake area of Manitoba and therefore they cannot be directly comparable.…”

“…Past Dakota skipper research in Manitoba, Saskatchewan and the northern U.S. has generally focused on describing local biotic and abiotic site characteristics (Dana 1991;Rigney 2013;Seidle et al 2018) as well as some preliminary SDM studies to determine potential habitat range of Dakota skipper and Poweshiek skipperling, Oarisma poweshiek (Westwood et al 2019;Post van der Burg et al 2020;Seidle et al 2020;Dearborn et al 2022) including additional modelling response to climate change (Barnes et al 2024).…”