Floristic response to landscape context in vascular plant communities in Eucalyptus obliqua and Eucalyptus regnans wet forest, southern Tasmania

Theories developed within the paradigm of landscape ecology propose that biodiversity within any given patch will be influenced by the surrounding landscape context (LC). Here LC is defined as the vegetation or land covers surrounding the site. This thesis used empirical vascular plant abundance data from the Huon forest district in southern Tasmania to test the hypothesis that LC influences the floristic composition and successional trajectory of patches within Eucalyptus obliqua and E. regnans wet forest. Secondary objectives included measuring the spatial and temporal variation in LC of the study area and its association with timber harvesting; describing differences in the response to LC between species and plant groups; comparing the effect size of plant responses to LC with other environmental predictors; finding the spatial and temporal scale at which plants respond most strongly to LC; and determining whether the effect of LC varied in response to disturbance regime differences. Most studies in landscape ecology have explored the effects of fragmentation on native vegetation in an agricultural matrix. In contrast, forest patches sampled in this study comprised native forest of various ages and successional stages within intact to variegated landscapes (sensu McIntyre and Hobbs 1999) in a frontier region dedicated substantially to timber production. Variation in LC was investigated for the study region using a new metric: the Landscape Context Index (LCI). This metric provided a relative scale of vegetation maturity in areas surrounding any given 50 x 50 m pixel, measured in 500 m, 1 km and 2 km radii. LCI was mapped for three years (1947, 1985 and 2009). The average LCI score was lower in 1985 than 1947, and was lower again in 2009. The greatest declines in LCI score were associated with timber harvesting, although in settled areas conversion to plantations was also an important contributor to LC score decline. The extreme reductions in LCI score observed in the period between 1947 and 1985 were not observed in the second period, possibly due in part to the mitigating effect of the Forest Practices Code, which resulted in changes to timber harvesting practices, including increased dispersal and reduced size of clearfelled patches. Distance to the nearest mature forest edge (DMFE), was used as a surrogate for mature forest influence, a component of LC. Between 15 to 200 m DMFE, there was an observed gradient in assemblage variation, species richness and diversity in all three age classes of silvicultural regrowth forest studied (4‚Äö-9, 22‚Äö-28, and 41‚Äö-45 years since regeneration). These trends were mainly driven by declines in richness and cover of mature forest affiliated species. Factors considered likely to contribute to differences in plant species response to LC are their sensitivity to micro-climatic variation (influenced by proximity to mature forest, and topography) and macroclimatic variation, their capacity to persist through disturbance, and their dispersal mode. Individual species abundance models for both mature forest affiliated species and pioneer species were stronger (e.g. 38 out of 56 common species using beta regression modelling) when they included both LC metrics and site environmental variables. Although LC was important in explaining variance in species abundance, it typically contributed less to model strength than other environmental predictors, such as soil, climate, topography and disturbance history. Abundance of most pioneer species was associated negatively with mature forest metrics, while the abundance of mature forest affiliated species was associated positively with these metrics. There was little evidence that plant species with bird or wind dispersed seed or that had seed able to persist through disturbance in the soil or in woody capsules were any less associated with LC metrics than species sensitive to disturbance with shorter dispersal capacity. It is therefore possible that such species were responding indirectly to LC through inter species competition, or because they were dependent on animal species sensitive to LC for their pollination or dispersal. LC effects on site micro-climate, soil and browsing pressure, which all vary with distance from edges of mature forest could also be influencing the distribution of species in ways that are associated with the distance from and proportion of mature forest in the landscape. Auto-correlation between LC metrics and site history are other factors that may be contributing to the observed responses of all plant species to LC metrics. There was only equivocal support for the hypothesis that floristic responses within regrowth forests are more strongly associated with the LC of the patch at the time that they were last disturbed rather than current LC. Interpretation of the response of vegetation to both the spatial and temporal scale of LC was hampered by the strong autocorrelation in LC scores between scales, and by the sampling methods chosen. The balance of evidence suggests that the LC in the years following disturbance and prior to canopy closure strongly influence the trajectory rate of succession, however colonization and extinction within sites may occur infrequently at any time so that later successional stages are still likely to be influenced by LC at all times. An interactive effect between proximity to mature forest and fire frequency was observed within secondary regrowth forests, such that patches burnt twice or more since 1898 exhibited greater assemblage differences with distance from the boundary compared with forest burnt only once. This was evidence that the response to LC is partly dependent on disturbance regimes at the site and that succession towards mature forest occurs more slowly in frequently disturbed patches. There was also a poor representation of the dominant rainforest trees and epiphytic ferns in regrowth forest that has colonized abandoned pastures (old fields), irrespective of the LC. Barriers to the colonization by rainforest tree species within old field regrowth may include the maintenance of more open canopy structure for longer periods than typical in uncleared regrowth forest disturbed by single fire events. An examination of all results in conjunction with those of other empirical studies suggest that successional trajectories in the species composition of wet eucalypt forests may be altered by changes in landscape configurations in response to silvicultural practices and changes in climate and associated fire regimes. Given the association between LC and succession in forest patches, the observed reductions in mean LCI scores across the study region may signal that recovery from the effects of wildfire, extreme climate events, and harvesting may be slower at the site level. If so, then current LC patterns reflect a reduction in resilience at the landscape scale. Predicted changes in climate and associated fire regimes may make local extinctions more likely in areas of low LCI score. In the context of landscape planning, although dispersal and pollination distances for most vascular plants is best measured in tens of metres rather than kilometres, landscape planning at 500 m to one kilometre may be sufficient to avoid local extinctions at finer scales and thereby prevent range contractions. Silvicultural methods and land management practices that may avert future losses in landscape resilience are discussed.