COS 215-2 - Species distribution modeling of Texas SGCN fishes to guide landscape-level conservation

Abstract: The Native Fish Conservation Areas (NFCA) are a network of 20 freshwater systems across Texas, identified as priority areas for the conservation of fish species of greatest conservation need (SGCN) and the habitats they need to thrive. Conservation networks such as NFCAs are a primary tool in multi-species stream conservation at large spatial scales, provided that they are based on accurate predictions of species occurrence.

Here we develop new species distribution models (SDMs) to update predicted occurrences of species, especially for SGCN. We chose an ensemble modeling approach because it reduces the sensitivity of predictions to model class, by maintaining lower variance and higher accuracy than single models, which is especially important when producing predictions across multiple species. We generate an ensemble prediction of probability of occurrence derived from average results across model classes such as generalized linear models, generalized additive models, and machine learning methods (boosted regression trees, random forest models, etc.). Individual model fits are performed using occurrence data from the Fishes of Texas biodiversity database, which compiles information on species occurrence from museum collections, fish community field surveys, detailed biodiversity assessments (i.e., Bioblitzes) and citizen-based monitoring.



In this presentation, we will talk about which environmental factors best predict the occurrence of fishes in Texas streams. We will describe the ensemble modeling approach we chose, with model validation steps, and detail resulting predictions, as well as accuracy and performance among species. This project is the first step in aligning conservation planning objectives and priority projects within NFCAs to the distributions of SGCN, with a goal to support the upcoming updating of the Texas Conservation Action Plan (i.e., the State Wildlife Action Plan for Texas)