Description: As part of the Kansas Aquatic Gap project, researchers at Kansas State University coordinated with Kansas Department of Wildlife and Parks Stream Biologists and utilized years of data from KDWP’s annual stream monitoring program, and historical records from KU and Sternberg natural history museum records, the Sam Nobel Museum of Natural History, Vaughn Weaver, and some KSU related projects to map occurrences of 119 fish species. These current and historical occurrence records were then analyzed against a suite of stream bio-geo physical variables and processed with Maximum Entropy (Maxent) models to generate a predictive model of potential habitat suitability for each species. Maxent is a general-purpose machine learning method with a simple and precise mathematical formulation, and it has a number of aspects that make it well-suited for species distribution modeling (Phillips et al. 2006). Because Maxent only relies on presence only data, we used our entire database on fish species distributions (approximately 3,400 community collections). Mean variable importance across fish species was greatest for geographic coordinates and stream order, reflecting regional differences in fish species distributions and variation with stream size. The relative importance of stream habitat variables in these models is assessed by evaluating the performance of the model when the variable of interest is left out. Caution is suggested in interpreting these variable weightings because this method does not account for potential interactions among variables. Model results were initially mapped to a stream network with fewer stream segments than the NHD datset, so processing was conducted to relate the model results from the original stream segments to the corresponding NHS stream segments. Due to the higher level of detail in the NHD dataset (included 361, 406 segments) Maxent predictive results were only transferred to stream segments identified as being Strahler order 2 or greater (82,997 segments).
Copyright Text: Phillips, S.J., R.P. Anderson, R.E. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecological Modeling 190: 231–259.
Mark Van Scoyoc, Stream Survey Program Coordinator, Kansas Department of Wildlife, Parks, and Tourism, 512 SE 25th Ave. Pratt, KS 67124; 620-672-5911.
Fish collections from Kansas Department of Wildlife and Parks Stream Assessment Program, University of Kansas and Sternberg Museum of Natural History collection records.
Maxent processing performed by Dr. Keith B. Gido, Kansas State University, 208 Bushnell Hall, Manhattan, KS 66506; 785-532-5088. GIS processing to integrate Maxent results to NHD stream segments conducted by Michael Houts, Kansas Applied Remote Sensing Program.
Web GIS development conducted by Jorgina Ross, Kansas Applied Remote Sensing Program.
When you use this data, we request that you acknowledge the Kansas Department of Wildlife and Parks, Kansas State University, and the Kansas Applied Remote Sensing Program, KARS, at the Kansas Biological Survey at the University of Kansas as appropriate.