Smith Fellows
Smith Fellow


Freshwater ecosystems teem with life and provide a vital resource for both surrounding wildlife and migrating species. Yet, freshwater ecosystems must also support human populations with drinking water, crop irrigation, waste processing, transportation and recreation. As global demands on these ecosystems increase, ecological processes are impacted, and aquatic communities placed at risk. However, the very forces threatening aquatic ecosystems create an opportunity to identify key resource factors and processes essential to maintaining freshwater ecosystem integrity.

By comparing aquatic communities across a range of habitat variables, from relatively pristine reference sites to impacted, hydrologically altered sites I will identify key resource factors and processes essential to maintaining the integrity of freshwater ecosystems. Using an artificial neural net program, I will model the mechanistic relationship between watershed indices and focal species. Coupling Geographic Information Systems land cover data with an artificial neural net program I will create a spatially explicit predictive model capable of determining thresholds of hydrologic alteration sustainable by freshwater ecosystems.

The proposed model represents a breakthrough for ecological modeling and conservation management. Artificial neural net models are a cutting-edge approach for modeling the complex mechanistic relationship between key ecological resources and focal species. Due to their capacity to handle nonlinear, nonparametric data, artificial neural nets show promise for modeling the complex interconnected relationship between indices such as water chemistry, flow rates, benthic macroinvertebrates and a focal species or assemblage.

By linking spatial and temporal habitat information with abundance of focal species this model can address questions about ecological processes such as dispersal, meta-population and source-sink population dynamics. In addition, by using hydrologic alteration and land use as watershed indices, the model can determine critical thresholds at which focal species can no longer sustain viable populations, providing conservation organizations with a tool for ecologically sustainable water management.