Methods
I created my predictive extinction risk models using a multi-criteria evaluation (MCE) framework, specifically utilizing the Suitability Analysis tool in ArcGIS Pro. This process allowed me to incorporate all of my anthropogenic variable datasets in a single model to identify regions of most and least likely mammalian extinction risk.
I first conducted an Analytic Hierarchy Process (AHP) to determine the relative factor weights used in the initial MCE model. The ranking decisions were based on a comprehensive review of existing literature detailing the relative importance of various anthropogenic variables on mammalian extinction risk.
Table 2 and Figures 2 and 3 below include the aggregated results of this literature review in order of final factor significance, as well as the subsequent results from the AHP for both isolated and non-isolated environments.
Table 2. Explanations for relative weighting assigned to each anthropogenic input variable inputted to the initial MCE analysis. Human population was assigned the highest factor priority, agricultural conversion was given the next highest preference, and invasive species counts were weighted the lowest given extensive review of prior research comparing these variables.
Figure 2. Numerical assignments given to each pairwise comparison of anthropogenic input factors in the analytical hierarchy process.
I then conducted several steps for data preparation, transformation, and analysis in ArcGIS Pro to create five different models of extinction risk using the same anthropogenic variable datasets representing human population, agricultural land conversion, and invasive species counts.
The full ArcGIS Pro workflow is visualized in the figure to the right (Figure 4), detailing each processing and analytical step by input and function.
To summarize the steps in the figure, my primary workflow proceeded as follows:
Add all datasets (IUCN mammal ranges, human population, agricultural conversion, and invasive species counts) to a blank base map of the world
Assign relevant data to standardized country codes to streamline interpretation
Convert all anthropogenic variable layers to raster format
Use Suitability Analysis tool to create a predictive model including the three anthropogenic variable layers and factor weights described above
Assign all variables to MSLarge membership model, as variable effects can be expected to increase until a certain threshold is met, at which point compounding impacts on extinction risk would become nominal
Conduct a sensitivity analysis by creating four additional models using the same steps, but varying factor weights and membership models
Use the Locate tool to identify regions with highest predicted extinction risk for each model
Determine the total proportion of overlap between these hotspot regions and the ranges of threatened mammals described by the IUCN validation data
Figure 3. Results of the analytical hierarchy process, including the final decision weights, relative significance given to each contextual criteria, and a final consistency ratio (CR) of 0.0735.
Figure 4. Model of ArcGIS Pro workflow executed for this study, including a data preparation phase, a transformation stage, and several steps for final analysis.
The maps resulting from the data preparation stage can be found below (Figures 5 and 6). These depict the validation data of currently threatened mammal species ranges, as well as each of the anthropogenic variables in global raster format.
Figure 5. Initial map visualizing the IUCN validation data of current threatened mammal ranges, bucketed into color-coded regions by risk category (severity of risk).
Figure 6. Set of three maps visualizing each anthropogenic factor (top: human population, middle: agricultural land conversion, bottom: invasive species) on a standardized global scale and transformed into raster format for analysis.