Integrating spatial data for predicting the influence of altered hydrologic and thermal conditions on fish assemblage traits and taxa across stream flow regimes
June 2020 - June 2024
Stream hydrology and temperature are among the most influential regulators of life-history traits and community structure of aquatic organisms (Buisson et al. 2008; Bruckerhoff et al. 2019). Hydrologic and thermal gradients strongly affect individual fitness and ultimately species success by imposing fundamental constraints on behavior, metabolic rates, reproduction, growth and ecological interactions. Stream hydrology and water temperature are also among the most frequently altered components of lotic systems due to human activities and other environmental disturbance (Poff et al 2010; Bae et al. 2015). Despite their critical role in sustaining native aquatic biodiversity, few studies have examined the cross-scale influence of hydrology and water temperature on freshwater biota using a multi-species and flow regime analytical framework.
The Ozark and Ouachita Interior Highlands and Gulf Coastal Plains regions are characterized by high biological diversity and species endemism (Matthews and Robison 1998), in addition to a rapidly growing human population dependent on freshwater resources (Northwest Arkansas Council, 2017). Human activities leading to hydrologic and temperature alteration are common in this region. Additionally, climate change scenarios predict more frequent temperature and precipitation extremes leading to increased warming, increased flooding in winter and spring, and increased drought during summer and fall in streams in the Interior Highlands and Gulf Coastal Plain (Diffenbaugh et al. 2005). Due to these changes, a growing percentage of streams are predicted to experience flow and temperature regime shifts over the next decades.
We propose to leverage high performance computing resources at the University of Arkansas and through the Google Earth Engine to quantify the influence of hydrology, temperature and landscape change on fish and aquatic macroinvertebrate assemblages. To do so, we will link large species taxonomic and functional trait databases with hydrologic metrics derived from the USGS national stream gage network, and satellite remote-sensing data including daily NASA Integrated Multi-satellite Retrievals for Global Precipitation Measurements (IMERG-GPM) and daily land surface temperature (LST) and emissivity from the Moderate Resolution Imaging Spectroradiometer (MODIS). We will use a new machine learning approach, Gradient Forest modeling, that is based on Random Forest models to examine non-linear environmental thresholds. Gradient Forests split values of a predictor variable and evaluate where species composition or traits change along an environmental gradient leading to the identification of environmental thresholds, or not, dependent on the underlying data. Results will indicate significant hydrologic, temperature and land use thresholds for individual fish species and functional traits. This approach can focus on entire assemblages, species of greatest conservation need or those of management concern. We have recently presented our approach to examining hydrologic thresholds of fish assemblages and species of greatest conservation need at multiple venues to great interest by researchers, managers and decision makers, and we feel there is a great opportunity to continue to develop and expand this approach to address important natural resource questions at local, regional, and national scales.