Pollinators of Mexico

Exploring the dynamic between pollinators and conservation land. This is a new and reduced version of the processing in polinizadores.

Process overview

1. For each species, create a model that predicts likelihood of species occupancy based on the environment variables (altitude, landuse, WorldClim).
2. Use each model to predict species occupancy across Mexico.
3. Sum the likelihood rasters into a richness raster.

Inputs:

• species points from GBIF provided by Dr. Quesada
• predictor variables: all 19 WorldClim bioclimatic variables, elevation, land use/landcover, biomes
• administrative boundaries for context
• polygons of natural protected areas (areas protegidas naturales, ANPs)

Files necessary to re-run the process (stored in the LANASE cluster)

• Mexico states: input_data/.../dest18gw.shp
• Stacked predictor variables with accompanying GRD file: all files in tidy/environment_variables
• If the above does not exist, then: all files in input_data/environment_variables/cropped

Outputs

Data

• species distribution models (SDMs) for all species
• dataframe with species modeling attributes
• richness layers by species groupings

Figures

• richness maps
• richness distribution by ANP zone, ecoregion, and pollinator group

Workflow

Using the scripts

1. Modify parameters in 00_initialize.R, such as parameters for the filtering of species observation points and parameters for the random forest model.
2. If the point data needs to be re-processed, run prep_Quesada_GBIF_data.R. The function add_taxon_info requires manual input when taxize needs help in identifying the best match.
3. To rerun the model, run process_SDM_for_cluster.R.
4. To perform spot-check and sum the SDMs into richness layers, run process_SDM.R.
5. To perform analysis on the richness results, run analyze_richness.R.