Predictive modelling and spatial mapping of freshwater fish and decapod assemblages using GIS and neural networks

Abstract

1. We used stream fish and decapod spatial occurrence data extracted from a national database and recent surveys with geospatial landuse data, geomorphologic, climatic, and spatial data in a geographical information system (GIS) to model fish and decapod occurrence in the Wellington Region, New Zealand. 2. To predict the occurrence of each species at a site from a common set of predictor variables we used a multi-response, artificial neural network (ANN), to produce a single model that predicted the entire fish and decapod assemblage in one procedure. 3. The predictions from the ANN using this landscape scale data proved very accurate based on evaluation metrics that are independent of species abundance or probability thresholds. The important variables contributing to the predictions included the latitudinal and elevational position of the site reach, catchment area, average air temperature, the vegetation type, landuse proportions of the catchment, and catchment geology. 4. Geospatial data available for the entire regional river network were then used to create a habitat-suitability map for all 14 species over the regional river network using a GIS. This prediction map has many potential uses including: monitoring and predicting temporal changes in fish communities caused by human activities and shifts in climate, identifying areas in need of protection, biodiversity hotspots, and areas suitable for the reintroduction of endangered or rare species.

Relative influence of local and landscape-scale features on the density and habitat preferences of longfin and shortfin eels

“…Davey et al 2011). At landscape scales interactions between variables such as slope and distance inland may be likely RootV Water depth (log(depth*100)) LogD Presence of cover for large eels (0 to 10) COVER to influence migratory fish, such as eels (Joy & Death 2004). We therefore repeated all partial constrained ordinations after having included second order interactions between all predictor variables.…”

“…Our work suggested that, to adequately represent patterns in densities, models will need to incorporate both local-scale and landscapescale variables that include interactions and nonlinear relationships. Further studies are in progress, but it seems likely that best approaches will require relatively long-term data and sophisticated modelling tools such as mixed-effects models (Beˆche et al 2009), regression trees (Elith et al 2008) or artificial neural networks (Joy & Death 2004). Consideration of diel variations in local-scale habitat preferences, as has been investigated for other New Zealand native fishes (Davey et al 2010), may also be an area which would benefit from further research.…”

“…Our results also correspond with the findings of Joy & Death (2001) who stated that caution is required when associating habitat characteristics with fish distribution in New Zealand, because the observed distribution may not be the result of the proximal habitat but rather of access to that habitat. Both Levin (1992) and Joy & Death (2004) stated that factors influencing fish distributions occur over a range of scales, but there is likely to be a hierarchy of effects so that the finer scale variables are constrained by the landscape-scale factors. Whilst our methods were not designed to test this hypothesis explicitly, our results did support the theory that the finer scale variables are constrained by the landscape-scale factors.…”

“…Since changes in habitat preferences ) and upstream penetration (McDowall 1990;Joy & Death 2004) are known to occur with size and species of eel, we separated the data by species and into four size-classes. These size-classes grouped individuals into sizes of 0Á149 mm, 150Á299 mm, 300Á399 mm, ]400mm.…”

“…These models are produced by empirical comparisons between reach-scale abundance or presence-absence of a species (or several species) and landscape-scale variables such as distance to coast or maximum downstream slope (Graynoth & Niven 2004;Leathwick et al 2005;Leathwick et al 2008a;Graynoth & Booker 2009). These models may be used for stock assessment, identification of conservation or restoration areas ) and understanding the impact of dams as barriers to migration (Joy & Death 2004). Cross-validation techniques can be used to quantify uncertainties and, once devised, they have broad spatial coverage and are relatively cheap to utilise.…”