Regional risk shifts to monarch butterfly migration due to climate change
Description
Raster files of future land-use/land-cover maps, as well as ecological niche modeling current and projected distributions of Asclepias and monarch butterfly larvae and eggs under current and climate change scenarios.
Files
Steps to reproduce
Download the files for use. All maps are raster files in .tif format. The Asclepias raster is a multilayer file. Methods: We compiled georeferenced records of monarch butterfly eggs and larvae (ELM) in Mexico from Red Monarca and Journey North, two long-standing citizen science initiatives. ELM records were included due to their strict dependence on milkweeds (Asclepias spp.) during early life stages and their limited dispersal capacity. To model biological suitability, a taxonomically curated database of Asclepias was developed through expert revision and complemented with GBIF records for Central and South America. Records with taxonomic errors or located outside known species distributions were excluded. To reduce spatial autocorrelation, ELM records were filtered using correlograms, while Asclepias records were thinned following Assis’ method, retaining precipitation seasonality (bio15) as a key variable. Occurrences were spatially thinned based on minimum non-significant autocorrelation distances. A total of 46 Asclepias species with ≥20 records were included; species with fewer records were subjected to an alternative iterative thinning approach. All retained records were combined into a final dataset for model training. For ELM and Asclepias, 33 modeling experiments were tested using three algorithms, five pseudoabsence sets, and four partitioning strategies. Model performance was assessed using TSS, ROC, and Boyce index. Based on validation metrics and expert evaluation, the best-performing configuration used a random forest algorithm with 10,000 pseudoabsences, bootstrap partitioning (70/30), and five replicates. Extrapolation risks were assessed using ExDet. Calibration areas (M) were defined using RESOLVE ecoregions encompassing occurrence records, and 19 bioclimatic variables were constrained accordingly. Bio6 (minimum temperature of the coldest month) was retained for ELM and bio15 for Asclepias. Multicollinearity was addressed using Pearson correlations and VIF analysis. Climatic suitability was modeled for current conditions and future projections (2030, 2050, 2070) using WorldClim 2.1 data and two CMIP6 GCMs (CanESM5 and MPI-ESM1-2-HR) under SSP2-4.5. Models were projected to Mexico and converted to binary maps using the TSS-maximizing threshold. Biological suitability was estimated through weighted species richness maps of Asclepias, classified according to their relevance to ELM based on literature and expert knowledge. Environmental suitability was derived from land-use and land-cover change models under SSP2-4.5. Finally, climatic, biological, and environmental layers were overlaid to define low, medium, and high ELM suitability areas.
Institutions
- Universidad Nacional Autonoma de Mexico