Modeling habitat suitability of wild rice (Zizania spp.) to inform restoration efforts in Michigan, U.S.A.

Abstract

Wild rice (Zizania aquatica and Zizania palustris) has been an integral component of Michigan wetlands for thousands of years, but is disappearing today due to habitat loss. Identifying suitable habitats and associated environmental conditions for wild rice viability is a critical step toward guiding ongoing restoration efforts statewide. To this end, we used maximum entropy (Maxent), a presence‐only species distribution model, to characterize habitat preferences of wild rice in Michigan and identify suitable habitats for restoration. Extant wild rice occurrences were mapped alongside 11 environmental variables using Maxent to determine probabilities of occurrence throughout the state. The Maxent model (AUC > 0.9) predicted the highest probabilities of wild rice occurrence in the western and southwestern regions of the state. Climatic variables, biotic interactions, and anthropogenic disturbances were all found to influence wild rice occurrence, with land cover, lower May precipitation, and minimum winter temperatures having the highest contributions to model predictions. The results of this study suggest that statewide restoration of wild rice in Michigan may be possible, although localized differences in environmental conditions should be accounted for when selecting restoration sites. With further refinement and the inclusion of more key environmental predictors of wild rice viability, habitat suitability models such as the one created in this study may be powerful and effective tools in guiding wild rice restoration efforts nationally.