Habitat fragmentation, degradation and habitat loss are leading drivers of the biodiversity crisis worldwide. While theoretical work predicts negative effects of fragmentation on biodiversity, empirical work suggests that fragmentation can result in positive, negative and neutral effects on local biodiversity. Fragmentation includes reduced connectivity and changes in movement patterns between habitats. Further, connectivity depends on the spatial scale evaluated. Fragmented landscapes may still be connected for species that disperse over long distances, while species that operate on smaller scales and disperse below a critical threshold will be isolated on individual fragments. However, dispersal ability is often unknown, especially at biological hierarchy levels above the single population or species scale. Here, we present a multiscale analysis of landscape connectivity in a historically fragmented forest landscape. We apply a graph theory approach to a dataset of canopy arthropods from a naturally fragmented forest on Hawaii Island. We set out to address the following question: How do connectivity and available area across spatial scales predict richness, abundance and biomass in canopy arthropods?
Results/Conclusions
Arthropod samples were collected on Hawai’i Island from the canopy of Metrosideros polymorpha trees in the center and edge of 34 forest patches. We used a graph theoretical approach to incorporate area, connectivity and spatial configuration. Surprisingly, we find that rarefied richness does not increase with fragment area. A multiscale analysis across threshold distances indicates that the landscape is connected in three components at 350m, and is fully connected in a single graph at 700m. This is supported by the fact that at small spatial scales (50 m threshold distance), rarefied richness was significantly correlated with landscape connectivity. With increasing spatial scales, connectivity was no longer a significant predictor. Arthropod abundance and biomass showed similar patterns, with connectivity as a significant predictor only at small spatial scales. For arthropod abundance, tree height was a strong predictor across scales. This is likely because arthropod abundances are limited at the hyperlocal scale by resource availability. The lack of effect of area and connectivity on richness suggests that canopy arthropods do not experience the landscape as fragmented. The nature of the landscape is such that multiple patches are combined in a connected component. This study highlights the importance of understanding habitat use across spatial scales in fragmentation studies.