Year: 2010
Type: dissertation
Abstract: The management of biodiversity within natural landscapes is an imperfect science due to uncertainty resulting from inherent ecosystem complexity. In disturbance mediated systems, reservation alone does not guarantee the preservation of biodiversity. Information is typically scarce; however methods which can improve the yield of ecological understanding from available data can aid conservation by allowing more precise intervention and reducing adverse or ineffective management actions. This dissertation attempts to improve methodology in vegetation ecology by using case studies to assess novel analysis methods which provide quantitative estimates of pattern. While quantification is common in production systems, in natural ecosystems the focus has typically been on individual species distributions and community change. Quantitative approaches were developed, implemented and assessed in four aspects of vegetation ecology; the community, individual species pattern, species responses to fire and variation in fuels.\n \n The continuum concept is dominant in ecology; however this is not evident in the typical representation of communities as discrete classes. The potential for aligning community representations with the continuum concept was assessed by considering the use of fuzzy community groupings. Fuzzy groups derived from floristic data were found to account for more compositional variation than both floristically derived discrete groups and physiognomic classifications. Models were developed to associate fuzzy values with environmental properties and these were used to create maps of both fuzzy and traditional communities.\n \n When considering variation of individual plant species, it is common to predict distribution but not abundance. The potential for modelling abundance was assessed by using two related species (Xanthorrhoea australis and X. caespitosa) as exemplars. These species were modelled using zero inflated models to predict variation in response to environmental pattern. Substantial variation in each species could be explained and models were used to successfully generate abundance maps. A relationship between occupancy and abundance was present for X. caespitosa, but not for X. australis.\n \n An investigation of variation in species abundances in response to fire found that while an ordination indicated fire had minimal influence on composition, when considered individually, the majority of species exhibited some form of response to fire. A wide variety of responses was observed, with different species being favoured by various aspects of the fire regime. There was no evidence of fire affecting species richness; however species diversity was observed to peak at about 15 years after fire. There was evidence that the effects of burning at invariant intervals may be additive for several species.\n \n An exercise of modelling fuel within the landscape found that fuel was highly dependent on species composition, with certain species disproportionally affecting different fuel elements. Fuel levels could also be robustly related to environmental conditions, enabling the recognition of ecological drivers and prediction of fuel across the landscape.\n \n Vegetation has long been recognised to vary in a continuous manner, yet many of the methods used to assess vegetation change still focus on discrete classes. Using quantitative assessment methods, models were found to be better representations of reality, providing deeper ecological insight and a more accurate understanding of vegetation condition.
Author: Thomas J. Duff
Language: English
Subfield: Nature and Landscape Conservation
Field: Environmental Science
Domain: Physical Sciences
SDG: Life on land
Open Access status: green
OpenAlex ID: W1490876803