Kīpuka as a model for anthropogenic fragmentation
Kīpuka are a well-suited natural experiment we can use to explore the effect of anthropogenic habitat fragmentation on tropical forest arthropod communities. Indeed, the stark contrast in conditions provided by lava relative to moist tropical forest makes kīpuka systems an interesting model for studying urbanization and road-building. The parallels are particularly striking in terms of the abiotic conditions created by lava versus paved surfaces. Of course, we must note that the parallelism to urbanization and road-building is limited at the biotic level. Whereas lava flows are themselves sterile and merely create conditions suited to the persistence of non-natives that passively arrive to the kipuka system, urbanization and road-building are in fact conduits for the active introduction of non-natives to the disturbed zone.
While tropical forests harbor the majority of global terrestrial biodiversity (Ghazoul & Sheil 2010), the resident taxa appear to be extremely sensitive to the effect of habitat fragmentation, at least according to what we learn from examining Hawaiian arthropods in kīpuka. Isolation by lava flows and resulting species-area effects means that small kīpuka harbor significantly reduced species richness. Even the forest edge of kīpuka is entirely avoided by many forest-specialist species, which are dominated by a few edge-tolerant native taxa living aside non-native colonists. The near complete isolation of kīpuka communities across very short geographic distances, supported by accelerated species turnover over similar distances in the kipuka compared to the continuous forest system, suggests that even a wide road may suffice to prevent dispersal for many arthropod taxa. Intriguingly, the species-area effect we observed in the kipuka system was even more pronounced at the haplotype level; i.e., genetic diversity may be affected more strongly by isolation in small fragments than taxonomic diversity. Hence very small kīpuka might have less evolutionary value as the haplotype diversity they contain for any given species dwindles under powerful genetic drift, in the process accumulating high extinction debt relative to wet forest taxa. In this light, our results suggest that even a network of relatively close forest fragments might not promote the persistence of viable metapopulations of some specialized forest taxa.