Prediction of historical forest habitat patterns using binomial distributions and simple Boolean logic from high spatial resolution remote sensing

The identification of forest habitat. its spatial pattern and use by select ed taxa is a vital step for the protection of biodiversity. The use of airb orne videography and frequency distribution models based on historical habi tat complexity data can provide detailed information on the spatial and tem poral variation of habitat, respectively. The two techniques, however, have not been jointly applied to link the temporal variation in habitat to the spatial variation of habitat over the landscape to provide a complete histo rical picture of the variation of habitat quality of a forest estate. In th is paper, a processing methodology is developed which allows the current sp atial distribution of habitat quality to be used as a base to make retrospe ctive predictions of the spatial extent and pattern of habitat quality over the landscape. This is achieved by projecting the spatial distribution of habitat complexity scores derived from the videography, backward in time us ing a combination of simple Boolean logic, estimated binomial distributions , and the use of random fluctuations to mimic natural forest dynamics that are likely to have occurred over the modeling period. The simulations provi de information on the type and condition of habitat in recent history and c an be linked to models predicting the abundance of a variety of common and endangered taxa. (C) 2001 Elsevier Science Ltd. All rights reserved.