The extent of forest fragmentation in New Zealand and its effects on arthropod biodiversity
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Historically, New Zealand was almost completely forested below the alpine treeline, but 1000 years of Polynesian and European colonisation has resulted in the destruction of nearly three-quarters of the original forest cover. I assessed historical patterns of deforestation and forest fragmentation in relation to all major topographical, climatic and anthropogenic variables that may drive forest loss. Much of the deforestation occurred in regions with drier climates, reflecting the fact that human population density has always been highest in areas with moderately dry climates and that dry forests burned much more readily and extensively. The large remaining tracts of forest are mainly restricted to high elevations, while the lowland forests have been fragmented into small, isolated remnants. Fragmentation of the surviving forests increases their susceptibility to edge effects and invasion by adventive species, indelibly altering the ecological communities they support. Although a large proportion of the remaining forest is owned or managed by the Department of Conservation, the distribution of that protection is greatly skewed towards areas of low economic value and is not representative of the relative conservation value of landscapes that differ in their environments and degree of forest cover. Forest cover in the majority of New Zealand landscapes has been reduced below the level of an expected extinction threshold of 30 % forest cover in the landscape, and ongoing deforestation threatens to force more landscapes below the critical threshold. Deforestation is still occurring across the country, and it is concerning that current deforestation rates in some areas are far greater than those observed in tropical, developing nations. I showed that the remaining forest fragments in New Zealand have complex, irregular shapes, and find ubiquitous evidence that core habitats within individual fragments are spatially discontinuous, comprising multiple, disjunct cores of small average area. Because population density of forest-interior species typically decreases with decreasing habitat area, multiple, disjunct cores support a lower total population size than a single, discrete core of the same total area. I found in a spatially explicit, landscape-level analysis of habitat fragmentation in New Zealand that simple core-area models consistently overestimate the carrying capacity of habitat fragments. Habitat fragmentation and habitat destruction are widely recognised as two of the leading threats to the continued maintenance of global biodiversity. The effects of habitat fragmentation on biodiversity fall into five categories that describe the spatial and landscape attributes of fragmented ecosystems; (1) fragment area, (2) edge effects, (3) fragment shape, (4) fragment isolation, and (5) matrix structure. Each attribute affects species individually according to their particular biological requirements and life history strategies, leading to complex, and often conflicting, sets of results in the empirical literature. Furthermore, it is now apparent that the effects of fragmentation can take many decades to become apparent and that the spatial arrangement of habitat fragments can interact with other ecological processes to magnify the detrimental impacts of fragmentation on species. I synthesised the published effects of habitat fragmentation on the morphology, distribution and abundance of invertebrate populations, species and communities, and present examples of time lags and synergies from the fragmentation literature. I explicitly considered the underlying mechanisms determining the responses ofindividuals to fragmentation and discussed the role of species traits in determining species vulnerability to changes in the spatial attributes of fragmented landscapes. I sampled 35,461 beetles from a fragmented forest and matrix system in New Zealand over very large gradients of fragment area (10-2 to 106 ha) and edge distances (up to 1,024 m from the forest edge into both the forest and the adjacent matrix interiors). The beetle fauna was very diverse, with 893 species identified in 65 families, representing nearly 20 % of the known species in New Zealand. Beetle communities were strongly structured by forest fragmentation, but in species-specific ways. Distance to edge was consistently shown to have the largest effect on community composition, but, surprisingly, an interaction between area and distance to edge had a stronger impact on community structure than fragment area alone. I developed a new method to partition the variance in community composition that was explained by putative area and edge effects. The method uses backwards stepwise regression to determine significant predictors of gradients in beetle species composition that were identified by canonical ordination. I found that edge effects were driven partially by small-scale alterations to microhabitat and microclimate and partially by changes in landscape composition that varied with distance to edge. In contrast, fragment area effects were driven primarily by edge effects, the strength of which varied significantly with fragment area. I took a novel approach to characterising the responses of 185 common species to habitat edges by modelling species abundances across edges with a general logistic model that described sigmoid trends in abundance for forest specialist and matrix specialist species, as well as unimodal trends in abundance for edge specialist species. I used the second derivatives of the logistic and unimodal models to statistically determine the width of species response zones to edge effects. Beetle species responses to forest edges occurred over far greater scales than previously suspected, with edge response zones for some species extending for more than 1 km. Average edge response zones were 194 m wide and, for many species, began in the forest but extended into the adjacent matrix. Species were categorised according to their responses to fragment area and distance to edge. Closely related species were expected to be placed in similar response categories because they are predicted to share suites of traits that determine their susceptibility or resilience to fragmentation by virtue of common ancestry. Despite many species exhibiting responses that could be grouped into categories, individual species responses to fragmentation were largely idiosyncratic with even closely related species exhibiting strongly contrasting responses to fragmentation.