Western Monarch Habitat Suitability Model USFWS

To better understand the distribution of key breeding areas for the declining western population of monarch butterflies, the USFWS, in collaboration with the Xerces Society, developed a set of relative habitat suitability models. This effort, which took place between Feb. 2015 and Feb. 2016, involved compilation of all available western milkweed and monarch species observation data, field surveys to acquire new data in summer 2016, and Maximum Entropy MaxEnt modeling in R. Products include relative habitat suitability models of five milkweed species thought to be important to western monarchs that enough data points to allow for creation of reasonably robust models. The species are Asclepias speciosa showy milkweed, Asclepias fascicularis narrowleaf milkweed, Asclepias eriocarpa woollypod milkweed, Asclepias asperula antelope horn or spider milkweed, and Asclepias cordifolia heartleaf milkweed. These models were developed by correlating a set of environmental covariates to the known occurrence points. A different subset of uncorrelated variables was selected for each species from a set of 25 possible covariate rasters. A model of monarch breeding habitat suitability was also developed, by correlated known monarch breeding habitat points to as an uncorrelated subset of the environmental covariates as well as the habitat suitability models already developed for the 5 milkweed species. All of these models are being distributed via ServCat in their original 270m pixel resolution for use by the USFWS and its partners. These models are imperfect because they are based on limited and spatially biased data, but it is appropriate to use them to help plan prioritize habitat improvement actions at a regional scale, and to identify areas for 2016 surveys. Locations that show up as having high relative suitability but do not have known occurrences would be especially valuable to survey. The models include embedded metadata and Esri layer files. Full documentation of the methods used to develop these models will be provided in a report that is currently in draft and will be submitted for peer review in early 2016.