School of Agriculture, Food and Ecosystem Sciences - Theses
Permanent URI for this collection
Search Results
1 - 3 of 3
-
ItemThe influence of fire mosaics on insectivorous bats: From genes to communities( 2023-11)Australia is one of the most fire-prone countries in the world, with many ecosystems adapted to historical fire regimes. However, fire regimes are shifting, and significant changes to historical patterns of fire are recognised as a considerable threat to biodiversity. Inappropriate fire regimes have been linked to population declines of mammals across Australia. Despite this there are significant knowledge gaps in understanding how different components of the fire regime effect biodiversity, at different levels of biological organisation. This research examined the distribution and ecology of echolocating bat species (commonly referred to as microbats) in a fire-prone region of south-eastern Australia, to answer key questions of wider relevance to other fire-prone ecosystems. I explored the influence of the spatial pattern of different fire histories in the landscape – fire mosaics – on different levels of biological organisation, to draw inference on how fire-generated mosaics might affect the distribution of bat communities, individual species, populations, and genetic diversity. The research aims of this thesis were explored using a landscape scale field study, in the temperate forests and woodlands of south-western Victoria, Australia, which collected two types of microbat data. The first, an acoustic survey by means of passive acoustic detectors of 136 sites, spanning a range of fire histories, conducted over two seasons. The second a genetics study of 146 sites, also with a range of fire histories, over two seasons, which captured 647 individual bats. Passive acoustic detectors are often used to monitor echolocating bat species. However, identification of calls collected at large scales is hindered by substantial variation within and between species, and the considerable time investment needed to manually identify acoustic data. A pilot study designed to confirm the range of species present in the study area, along with the level of sampling needed to address the research aims, confirmed that the acoustic signature for many species overlapped. This meant that existing approaches for call identification would not be suitable for either automated or semi-automated identification of species from acoustic data. Therefore, as a first step, I developed a new method that allowed for much improved identification of species with overlapping call structures. This improvement then allowed for a semi-automated call identification approach, important for analysing large volumes of acoustic data collected across landscape scales. I built a machine learning classifier (Random Forest) from field collected data, to better differentiate acoustically overlapping species. The method improved overall classification success, including a 60% improvement for bats that navigate in open spaces (Chapter 2). The classifier was used to identify to species, a dataset containing 687,377 recordings over 1,632 detector nights collected at 136 sites. The results were used to model responses of species relative occurrence and specie richness to fire mosaic variables (Chapter 3). In Chapter 3, I determined how the amount, diversity, and configuration of fire influenced species richness and the relative occurrence of 13 individual species. Sites were primarily stratified to represent a range of fire age-classes: recently burnt (0-3 years post fire), early successional (4-10 years), mid successional (11-34 years) and late successional (>34 years). I built regression models for 13 species and species richness against different measures of the fire mosaic across six spatial scales (500 m – 5 km). Bat species richness responded positively to the diversity of fire age-class at all spatial scales, as did 9 out of 12 species, at one or more scales. Models of individual species responses to fire diversity indicated that the scale at which fire is measured can be important. The amount of a fire age-class was also an important determinant of relative occurrence for five species over a range of scales. Contrasting responses to fire configuration, displayed by closely related species, revealed that even subtle differences in wing morphology can contribute to differing responses of otherwise similar species. Next, I determined the influence of the fire mosaic on the genetic structure, connectivity, and diversity of two ecologically different species (Chapter 4). High resolution genetic data was derived from tissue samples collected from 280 individuals across 71 sites. This research employed emerging techniques in landscape genetics, to identify genetic variations among individuals within populations of two ecologically distinct species. I used a multi-step analytical approach. Firstly, for both species, a spatial admixture model was employed to establish likely ancestral populations. Secondly, these ancestral populations were then incorporated as random factors in Generalized Linear Mixed Models (GLMMs) to explore the correlation between genetic distance (indicative of genetic connectivity) and a range of fire metrics representing gradients of spatial patterns in fire history, availability of woodland habitat, and spanning six scales. Finally, GLMMs were used to determine the connection between individual genetic diversity and the same fire metrics across these six scales. High resolution genetic research yielded insights on the interplay between species' genetic patterns and the fire mosaic. The phylogenetic structure of Chalinolobus morio uncovered in this study supported the presence of male and female philopatry, a notable finding considering that such philopatry is rare among mammals. Conversely, the phylogenetic structure of Vespadelus vulturnus, coupled with decreased genetic connectivity and diversity, indicated that external constraints related to the fire mosaic likely disrupt the connectivity of habitat for this species. In summary, this research developed a new approach to surveying and monitoring echolocating bats species at large spatial scales. The approach is flexible and applicable to acoustic surveys more broadly. I quantified the influence of the spatial pattern of fire on insectivorous bats to define fire mosaics that benefit bat conservation. Areas of long unburnt vegetation were found to be important for several species. Importantly, the configuration and diversity of fire age-classes were also important drivers, in addition to the amount of habitat. Specific to this landscape patchy burns at the current scale of management will promote bat diversity. Specific to individual species conservation, this study found that maintaining ‘enough’ of the extent of an age-class to support roosting and foraging habitat varied among species. Genetic signatures uncovered differences between individual habitat selection and muti generational gene flow, important for the persistence of populations in the face of changing climates. Moreover, variation in the spatial scale or buffer zone that produced the strongest models highlighted the variety of ways that bat species perceive and interact with the landscape. This thesis highlights how this knowledge can be used to improve fire managed landscapes for conservation. This research frames the response of occurrence and genetic data to ecological understanding of species responses, and by doing so is applicable more generally to other fire managed ecosystems.
-
ItemThe effects of fire and landscape structure on animal communities, species, and connectivity( 2023-04)The loss and fragmentation of habitat associated with land use change is the primary driver of global biodiversity declines. Changes to fire regimes that alter habitat suitability also threaten a range of animal taxa. Fire has been increasingly recognised as an important ecological process and is now used to manage fire-prone landscapes around the world, but important questions remain about the effects of fire regimes on animals, especially in fragmented landscapes. The aim of this thesis is to determine the influence of landscape structure (the composition and configuration of landscape elements) on animals in heathy woodlands in southern Australia in terms of fire, fragmentation, and interactions between them. First, I explored post-fire growth stage and land use together as components of landscape structure and assessed their relative and interacting effects on mammal communities. I used camera traps to collect mammal presence-absence data in 2019-20 and analysed it using ordination and linear modelling. I found that land cover composition was the primary influence on community composition. The composition of the fire mosaic had a secondary, weaker effect and one that may change depending on land cover composition. Second, I explored habitat structure as a mechanism by which fire regimes may affect mammal species, using a species activity index derived from the same camera trap data. Post-fire growth stage (a categorical representation of time since fire) was not a direct predictor of any species’ activity levels, but some habitat structure attributes were linked to certain growth stages and were therefore a mediating influence on animals. Finally, I assessed how the growth stages influence functional connectivity for a litter-dwelling skink. I used genetic data, landscape resistance modelling, and circuit theory-based mapping to find the relative connectivity of land use types and growth stages. Functional connectivity varied little with growth stage, with the primary influence on connectivity being the matrix of pasture and forestry plantation. Overall, I did not find direct effects of growth stage on animal communities, species, or connectivity. However, less obvious effects such as the composition of the fire mosaic beyond the site-scale and indirect effects through habitat structure are important to consider in future fire management. The extent of native heathy woodland was also vital for native mammal communities and functional connectivity; remaining native vegetation must be protected and expanded for the best outcomes for native diversity and species persistence.
-
ItemLandscape-scale disturbance history and the dynamics of the conifer taxa of Bidoup–Nuiba National Park, Vietnam( 2023-03)Tropical forests are complex and poorly understood ecosystems. In recent decades large-scale, long-term forest dynamics plots have been providing important insights into basic demographic processes such as recruitment, growth, and mortality. However, this information offers little insight into long-term forest dynamics and the role of infrequent disturbances that may occur at the scale of decades to centuries. Understanding how species-rich tropical forests will respond to global change requires understanding these long-term dynamics. Dendroecology, the study of tree rings, has provided the foundation for understanding forest dynamics in temperate forests in the northern hemisphere and parts of the southern hemisphere. It has been little applied in the tropics where most tree species do not have annual growth rings. The overarching aim of this study was to better understand the ecology and dynamics of Vietnam’s Central Highlands, and, in particular, the rich conifer taxa that are found there. To address this, my thesis involves the application of dendroecological approaches to: (1) reconstruct historical fire regimes in the Central Highlands of Vietnam; (2) understand the historical dynamics of Pinus kesiya forests and (3) the spatial and temporal patterns of recruitment and growth of the endemic conifer taxa in mixed conifer-angiosperm forests, and (4) identify angiosperm species with significant potential for future tree-ring based studies. A detailed reconstruction of how disturbances have varied across the Central Highlands landscape over recent centuries will provide important insights into the response of these forests to past disturbances and climate variability, as well as potential future trajectories of development under a changing climate. My thesis presents one of the most comprehensive reconstructions of fire in tropical forests using tree rings of Pinus kesiya in the Bidoup Nuiba National Park (BNNP) in the Central Highlands of Vietnam. A tree ring-based fire reconstruction showed that fire occurrence increased after 1900 and was highly correlated with climate indices (ENSO and PDSI) during the period 1900 – 1960. However, after 1964 the relationship between climate and fire disappeared due to the overwhelming pressure of human ignition sources, which have effectively eliminated climate as a factor limiting fire in these landscapes. While climate change is seen as a potential driver of changing fire regimes in many parts of the world, the role of local human populations may be equally or more important in shaping them. A detailed history of fire in these forests provides a baseline against which recent and future changes can be assessed. Dendrochronological analysis also indicated that extreme droughts, fire, and canopy disturbance have been associated with tree regeneration, growth, and canopy recruitment for centuries. The age distributions of Pinus dalatensis, Keteleeria evelyniana and Fokienia hodginsii show a distinct pulse of recruitment during the 1750s and during the period 1800-1850 that consistently coincided with extreme regional droughts. Periods of sudden and sustained growth release amongst these endemic conifers were associated with regional mega-droughts in the late 18th and 19th centuries. Results from superposed epoch analysis between Palmer Drought Severity Index and discrete tree release events indicated that extreme droughts and abrupt changes in moisture availability may have contributed to increased canopy disturbance rates and tree-level mortality. The effects of drought on tree growth were typically recorded in the tree rings 1-5 years after the drought event. This indicated that extreme drought conditions may drive large-scale canopy disturbances in BNNP. My results also confirmed that growth releases occurred within ~5 years of recorded fire events. This relationship between growth releases and fire events is consistent over 250 years of tree-ring records (1770-2020). Finally, my thesis demonstrated that a number of the angiosperm tree species occur in BNNP have significant potential for tree-ring studies. This is the first study to systematically examine the potential of a large number of angiosperm species for dendrochronological study in Vietnam. Demographic information derived from tree-rings of angiosperms could provide well-dated, long-term data that complements the tree-ring record from the conifers at BNNP. This would further advance our understanding of forest dynamics and the impacts of climate change in these species-rich tropical landscapes.