Lava-fragmented forests

A study system I have recently been developing involves the communities of native plants and animals in forest fragments on the Island of Hawaii. These fragments once comprised a large expanse of contiguous forest, but are now isolated as a result of a lava flow from Mauna Loa Volcano in 1855, resulting in more than 300 forest fragments. The fragments range in size from <0.1 to >100 ha, but share the same 3000- to 5000-year-old soils, within a single vegetation life zone between 1500 and 1800 m. Thus, this system represents a highly replicated gradient of forest size. As with elsewhere in Hawaii and other Pacific islands, these forests have been invaded by introduced mammals (e.g., rats, mice, mongoose), which exert wide-ranging effects as omnivorous predators. Designated as a forest reserve, these naturally fragmented forests have otherwise been little affected by human disturbance, making them valuable for the conservation and restoration of the unique Hawaiian biota.

Building on my previous research in New Zealand, I am working with entomological, ornithological, botanical and biogeochemical colleagues to use this Hawaiian landscape as a model system for studying interactive effects of omnivorous rats and forest fragment size on arboreal arthropod food webs. Currently, we are testing the hypothesis that rats as arboreal predators alter the structure of arthropod food webs in the forest canopy and that they do so more severely in smaller forest fragments. To accomplish this, we are using a combination of experimental removal of rats, experimental exclusion of insectivorous birds, stable isotope analysis of arthropod food webs, and remote sensing analysis of habitat characteristics.

In the future, I plan to expand this project by experimentally removing multiple species of introduced predators (rats, mice and mongoose) in different orders to examine how the disassembly history of non-native species affects the restoration of native plants, invertebrates and birds across a gradient of forest size and isolation. This work provides a unique opportunity to test theoretical hypotheses on community assembly and disassembly in a real-world, conservation-relevant setting.