Acd. Ertsen et al., COMPARISON OF THE PERFORMANCE OF SPECIES RESPONSE MODELS IN SEVERAL LANDSCAPE UNITS IN THE PROVINCE OF NOORD-HOLLAND, THE NETHERLANDS, Ecological modelling, 109(2), 1998, pp. 213-223
In The Netherlands, empirical statistical response models are used in
the Dutch nature conservation policy to predict vegetation response to
management options. The development of empirical statistical models i
s time consuming and costly, due to the need for databases that extens
ively describe the biotic and environmental variation in the research
area. For this reason it is important to optimize the sampling procedu
re and thus reduce costs. Our research aimed at finding out to what ex
tent species response models will improve when an environmental gradie
nt is sampled outside the range optimal for a species, which may be es
pecially relevant where species presence is related to geographical la
ndscape units. We tested for 46 species and two landscape units in our
research area, a dune and a polder area in the province of Noord-Holl
and, if it would be worthwhile to extend sampling to the other landsca
pe unit. Herefore we compared the predictive values of response models
based on either a dune or polder subset with those of response models
based on a combined dune/polder set. In general the predictive values
of models based on the combined set were not significantly higher tha
n those of models based on the subsets. Thus, where time and financial
resources are limited, we recommend focusing sampling attention espec
ially on the regions of interest. Extrapolation of subset response mod
els to the other landscape unit was also tested. Considering the extre
me poor predictions, we advise against the extrapolation of response m
odels to regions with different environmental conditions. (C) 1998 Els
evier Science B.V. All rights reserved.