Full text in pdf format

MEPS 224:251-266 (2001)  -  doi:10.3354/meps224251

Modelling spatial variations in spawning habitat suitability for the sole Solea solea using regression quantiles and GIS procedures

Paul D. Eastwood1,*, Geoff J. Meaden1, Alain Grioche2

1Marine Fisheries GIS Unit, Department of Geography and Tourism, Canterbury Christ Church University College, Canterbury, Kent CT1 1QU, United Kingdom
2Laboratoire d¹ichtyoécologie marine de l¹Université du Littoral‹Côte d¹Opale, 17 avenue Blériot, BP 699, 62228 Calais, France

ABSTRACT: In order to quantify spatial variations in spawning habitat suitability for the sole Solea solea (L.), spatial models were developed within a Geographical Information System (GIS) using data on the distribution of sole eggs within the eastern English Channel and southern North Sea, collected during March to June 1991. Raster maps of the physical environmental variables formed the basis of the models, with regression quantiles used to predict the spatial variation in sole egg density according to changes in physical habitat factors. The regression quantile parameter estimates were used to re-code the raster environmental maps so as to produce spatial models reflecting variations in both spawning activity and habitat suitability. Models developed for each month were then combined to produce a map of overall habitat suitability for a single spawning season. The use of regression quantiles allowed linear model parameters to be estimated for any part of the biological response, in this case sole egg density, including near to the upper bounds of the distribution. Changes near to the upper bounds would be expected when habitat factors included in the model actively limit the biological response, with the variation below the maxima caused by the influence of unmeasured factors. Spatial models developed using regression quantiles therefore yield predictions of the upper limits imposed by the measured habitat factors, i.e. the potential biological response. Spatial variations in sole egg density were found to be limited by depth, temperature, salinity and sediment type, with highest densities found in the shallower regions over sediments consisting of <30% gravel. The final models predicted different scores of spawning habitat suitability at each spatial location, with the model developed from upper regression quantiles generally predicting higher suitability scores. Models performed well when tested with data describing the distribution of sole eggs along the French coast of the eastern English Channel in April 1995. The use of regression quantiles allowed for the development of spawning habitat models that portrayed the upper limits imposed by the measured habitat factors. Underestimation of the relative importance of the habitat in terms of its suitability for spawning was therefore minimised using this technique.
Erratum


KEY WORDS: Habitat suitability models · Solea solea · Regression quantiles · Eastern English Channel · Southern North Sea · Geographical Information Systems


Full text in pdf format