Classification of aquatic bioregions through the use of distributional modelling of freshwater fish

Abstract

Bioregional classifications are used to manage natural resources including the conservation of biota. There are a variety of ways to define bioregions and mostly a combination of data analysis and subjective expert judgement is used, mainly because data on the distributions of biota are sparse or uneven. We trialled a method of using distributional modelling of individual freshwater fish species to produce a classification of rivers in New South Wales, Australia. Distributional modelling was done for 44 fish taxa using the genetic algorithm for rule set production (GARP) and a classification was done using non-hierarchical clustering. The data used was a combination of museum records (presence only records) and data from designed surveys. The natural distributions of seven fish species could not be modelled due to insufficient records. The models for the majority of remaining species displayed substantial to almost perfect model accuracy. The classification produced similar bioregions as had been previously defined for freshwater fish in New South Wales. Our study demonstrates that distributional modelling of individual species is a feasible and practical approach to defining regions using data derived from a variety of sources. The potential benefits of the method would be that a description of the potential “natural” fish assemblage could be described for any given site, separation between zones can be clearly delineated and it is independent of the actual fish sampling locations.