High rates of forest loss and turnover obscured by classical landscape measures

Abstract

Loss of native forest is a key conservation concern globally, for reasons of biodiversity, climate change and ecosystem services. Landscape measures are used widely to characterize forest loss and associated landscape structure, but often without regard for structure imposed by the data used, and associated assumptions. Notably, forest loss is often expressed simply as net change in forest cover over time, but this approach does not account for turnover (i.e. the gross losses and gains of forest). It also ignores forest age (e.g. young regrowth forest or mature forest), which is significant in conservation terms. We investigate the effects of removing common data constraints on landscape characterization, as typically used in landscape dynamic analyses. We produced fine-resolution (0.0225 ha) classified maps from satellite imagery of the temperate Araucanía Region of Chile, for 1986, 1999 and 2008. We calculated areas of land-use classes and associated landscape indices. Landscape measures and trends through time varied markedly around the region, with forest loss and fragmentation confined to areas not designated as protected. Net (‘headline’) figures masked very large turnover through time, with about 30% of unprotected land switching land use each decade. Accounting for this, in unprotected areas the loss of established native forest was 2.4% and 3.5% per year in the two time periods, much higher than equivalent ‘standard’ figures. Using finer-resolution data increased estimates of native forest loss and reversed temporal trends in patch density and mean patch size, compared with the commonly-used National Vegetation Classification (6.25 and 4.5 ha resolution). Interestingly, mean patch size of native forest actually switched, from a decreasing trend to an increasing one, with continued deforestation. We conclude that landscape characterization can lead to effective conservation practices, but it is necessary to use appropriate data resolution, define the data domain carefully and examine change through time, including the degree of dynamism (turnover) within the landscape: our results suggest a strong need to consider continuity of forest cover as well as overall totals.