Aim

Improving our understanding of the drivers of forest fragmentation is fundamental to mitigating the consequences of anthropogenic fragmentation for biodiversity. Moreover, the impacts of fragmentation on biodiversity depend on the spatial scale at which fragmentation occurs. Therefore, understanding how the effect of land use on fragmentation patterns varies across scales is critical to ensure that fragmentation is managed at scales relevant to the ecology of target species or to land management. Here, we quantified the influence of land use on patterns of forest fragmentation at different scales using Queensland, Australia, as a case study.
Location

North-eastern Australia.
Methods

We combined fractal analysis with piecewise linear regression to measure patterns of forest fragmentation across a range of scales in 5309 landscapes of c. 50 km2, with different proportions of land used for cropping and grazing. A significant change in fragmentation patterns occurred at approximately 1 km2. We used beta regression to quantify the impact of land use on the degree of fragmentation at scales finer and coarser than 1 km2.
Results

The use of land for grazing tended to create more fragmented forest patterns than use of land for cropping. This difference was more pronounced at coarser than finer scales.
Main conclusions

Our finding suggests that the choice of land use where conservation actions, such as revegetation and retention of forest patches, are to be prioritized depends on the scale at which we measure fragmentation. This information contributes to reducing the risk of mismatches between the scale at which fragmentation is managed and the scale at which fragmentation is measured, which is often dictated by the scale of species movements or the scale of land management. Our finding also improves our capacity to discern between fragmentation patterns that are typical of land-sharing and land-sparing conservation strategies, as spatial scale varies, thus aiding the implementation of land sparing and land sharing at scales relevant to biodiversity conservation and land management.