Time and matrix quality increase the relative habitat value of smaller patches in fragmented landscapes

Abstract The landscape‐scale impacts of habitat subdivision (fragmentation per se) on biodiversity are not fully resolved. While smaller patches usually contain more species for equal total area, many implications of this remain unclear. For example, do equal areas of smaller and larger patches provide equivalent habitat value? How might this change over time and under differing matrix conditions? To help address these knowledge gaps, we developed an indicator of relative habitat value based on a species‐individual null model. We used the indicator to compare sets of patches ordered in small‐to‐large and large‐to‐small orders, building hierarchical Bayesian regression models to test the role of time since patch creation and contrasting matrix conditions. This allowed us to assess habitat value for 85 metacommunities inhabiting fragmented landscapes (1354 patches, >4500 species). We expected comparable habitat value following patch creation due to unpaid extinction debts, and the matrix would determine the direction of change over time, a harsher matrix increasing the relative value of larger patches. Averaged over time and matrix quality, the indicator probability density was mostly negative, suggesting slightly greater habitat value among larger patches. This pattern was consistent across taxonomic groups, although amphibians and reptiles were the most affected, and invertebrates least so. Larger patches were of greater value within 20 years of patch creation, but over time smaller patches increased in value under any matrix type, whereas value remained constant for larger patches. Matrix conditions mediated the relative difference: after 100 years under a light matrix, patches of all sizes were essentially of equivalent value, while larger patches were still favoured under a harsh matrix. Policy implications . In long‐fragmented, light matrix landscapes, small and large patches apparently offer comparable per‐unit‐area habitat value. Conversely, under a harsh matrix, larger patches retain slightly greater habitat value. We speculate that this reflects a ‘colonization credit’, which might occur within smaller patches after the initial loss of fragmentation‐sensitive species in disturbed landscapes. Overall, analyses support the need to maintain and enhance total habitat area—regardless of configuration—for biodiversity conservation, especially in long‐fragmented landscapes with light matrices.