Predicting hydrologic disturbance of streams using species occurrence data

Area of Science:

• Ecohydrology
• Conservation Science
• Geospatial Analysis

Background:

• Aquatic organisms are adapted to natural river flow regimes.
• Human activities and climate change are altering these flow regimes globally.
• Increased hydrologic variability pressures aquatic ecosystems.

Purpose of the Study:

• To model and map spatial patterns of hydrologic disturbance in streams.
• To predict hydrologic disturbance for ungaged streams using machine learning.
• To identify regions for targeted conservation efforts.

Main Methods:

• Utilized a machine learning approach (Random Forest) with georeferenced species occurrence data.
• Trained models on fish community data, hydrologic, and landscape metrics from gaged streams (NHDPlusV2).
• Predicted a hydrologic disturbance index (HDI) for ungaged streams, incorporating topographic variables to improve accuracy.

Main Results:

• Fish presence/absence data showed high prediction accuracy (77%), comparable to flow variables (76%).
• Including topographic variables significantly increased model accuracy for both fish (90%) and flow (86%) predictions.
• Low HDI streams were concentrated in the Ozark Highlands, Boston Mountains, and Ouachita Mountains, often with stable groundwater and forested cover.
• High HDI streams were associated with large rivers, intermittent runoff, and agricultural land use.

Conclusions:

• Long-term biological data are valuable for predicting hydrologic disturbance in ungaged streams.
• Distinct ecohydrological regions with varying disturbance levels were identified.
• Findings can inform focused conservation strategies to mitigate hydrologic disturbance impacts on aquatic ecosystems.