Spatially Explicit Models for Freshwater Fish for Conservation Planning

Abstract

In conservation biology, spatially explicit models, sometimes known as predictive habitat distribution models (Guisan and Zimmermann 2000), have been widely used to predict distributions of plants (Leathwick 1998), insects (Lobo and Martin-Piera 2002; Gutiérrez et al. 2005), amphibians (Ray et al. 2002), reptiles (Fisher et al. 2002), fishes (Joy and Death 2001; Leathwick et al. 2005), birds (Peterson and Robins 2003), and mammals (Carroll et al. 1999). This modeling technique has been applied to Japan’s freshwater ecosystems: for example, to reconstruct historical global distribution of an endangered fish species with an already diminished distribution range (Fukushima et al. 2011) and to predict potential areas susceptible to invasion by exotic fish species (Iguchi et al. 2004). Because spatially explicit modeling is a correlative statistical technique, it basically requires only two sets of data: response and predictor variables. What makes this modeling technique different from classic techniques is that the variables (1) are spatial in nature, (2) are most often observational rather than designed, and (3) are not repetitive in space because the earth is the only unit (Cressie 1993). In a field-based ecological study, it practically means that the data are geo-referenced in a given landscape (e.g., by a global positioning system, or GPS).