| Deforestation is the primary cause of species loss in the tropical forests that harbor much of the world's biodiversity, and understory bird species are particularly sensitive to resulting habitat fragmentation. Ongoing extinctions long after initial fragmentation suggest that demographic consequence of habitat loss are persistent, but the particular vital rates most affected and the range of demographic responses among species are not well understood. Demographic analyses can be useful in identifying the mechanisms by which fragmentation drives extinctions, assessing extinction risk for remaining populations, and guiding effective conservation planning and reserve design. Through analysis of a long-term capture-recapture dataset for understory birds of the Usambara Mountains, Tanzania, I estimated the effects of habitat fragmentation on apparent survival, recruitment, and population growth rate for 22 species in a highly fragmented sub-montane forest. I also estimated landscape-scale effects through analysis of two disjunct communities in adjacent mountain ranges. To assess the role of depressed demographic rates on the long-term persistence of the avian community on remaining forest fragments, I used count-based population viability analysis to estimate extinction risk for eight species on small (2 ha), medium (34 ha), and (704) large forest fragments. Finally, I used multi-site population viability analysis to estimate extinction risk for 14 species given four, real-world reserve scenarios. I found that across feeding guilds and dispersal abilities, apparent survival was lower on small forest fragments relative to large. Because the effect of fragmentation on recruitment was not different than zero across species, I conclude that depressed survival was the primary demographic mechanism of lower population growth rate on small fragments. Fragmentation effects were consistent across landscapes but were more pronounced in the East Usambara Mountains relative to the West. Population viability analysis showed high extinction risk for five common species on small forest fragments and for two species on medium and large fragments. These results confirm a long faunal relaxation time and suggest continued future extinctions, particularly of less abundant species, even on the largest remaining forest fragments. Population viability analysis for four reserve scenarios showed an average delay in extinction risk of almost three decades under conditions of maximum reserve connectivity and a slight increase in protected area. Increasing dispersal among reserves was generally beneficial to population persistence, with the greatest benefit accruing through increased connectivity among the larger reserves. Many species had high extinction risk regardless of the conservation scenario, suggesting that improvements to habitat quality may also be required to maintain populations. Finally, a substantial difference in the results from demographic and species-area approaches suggests that projections of species losses from even well-protected biodiversity hotspots may be even greater than previously anticipated. |
