USING STATIC AND DYNAMIC HABITAT MODELS TO GUIDE RECOVERY OF THE THREATENED STEPHENS’ KANGAROO RAT | |||
| Wayne D Spencer; Conservation Biology Institute; wdspencer@consbio.org; Heather Rustigian-Romsos | |||
Statistical habitat value or species distribution models are valuable tools that should be tailored to specific purposes. We produced static and dynamic (annual) habitat-value models for Stephens’ kangaroo rat (SKR; Dipodomys stephensi) using multi-spectral satellite imagery to establish a spatial sampling frame for rangewide monitoring and to help managers understand how weather and management actions affect SKR habitat and populations. The static model used environmental indices generated from LANDSAT imagery and the annual models from Sentinel-2 imagery generated in the Google Earth Engine environment. The static model used MaxEnt and 20 years of SKR locality data. We averaged the environmental indices over that same 20-year period to smooth out annual variability and used a relatively low probability-of-occurrence threshold to identify where SKR might occur in the long term and therefore where monitoring is warranted. The annual models used Random Forest methods implemented in R with the flexSDM package to investigate annual variation in habitat value and populations using each year's monitoring results. Each year, the models used that year’s environmental variables and SKR captures to investigate weather- and disturbance-driven annual variation, including from habitat management actions. The models show fine-scale habitat changes that may correlate with variations in SKR occupancy and local density. We illustrate the methods and results for the first 5 years of monitoring and discuss recommendations for similar efforts. | |||
| |||
Speaker Bio: Wayne Spencer is Chief Scientist Emeritus at the Conservation Biology Institute. He works with skilled modelers and analysts to study mammals of concern, including martens, fishers, and kangaroo rats. |