Using experimental reintroductions to resolve the roles of habitat quality and metapopulation dynamics on patch occupancy in fragmented landscapes.

Declines of species in fragmented landscapes can potentially be reversed either by restoring connectivity or restoring local habitat quality. Models fitted to snapshot occupancy data can be used to predict the effectiveness of these actions. However, such inferences can be misleading if the reliability of the habitat and landscape metrics used is unknown. The only way to unambiguously resolve the roles of habitat quality and metapopulation dynamics is to conduct experimental reintroductions to unoccupied patches, allowing habitat quality to be measured directly from data on vital rates. We therefore conducted a 15-year study that involved reintroducing a threatened New Zealand bird species to unoccupied forest fragments to obtain reliable data on their habitat quality, and re-assess initial inferences made by modelling occupancy against habitat and landscape metrics. While reproductive rates were similar among fragments, subtle differences in adult survival rates resulted in λ (finite rate of increase) being estimated to be < 0.9 for 9 of the 12 fragments that were previously unoccupied whereas this was the case for only 1 of 14 naturally occupied fragments. This variation in λ largely explained the original occupancy pattern, reversing our original conclusion that this occupancy pattern was isolation driven, and suggesting that it would be detrimental to increase connectivity without improving local habitat quality. These results illustrate that inferences from snapshot occupancy should be treated with caution and subjected to testing through experimental reintroductions in selected model systems. This article is protected by copyright. All rights reserved.