Little, Jeremy Keenan (2015) A song of fire and water: will climate change interacting with fire affect the distribution of vegetation in the Australian Wet Tropics? PhD thesis, James Cook University.

Preview

PDF (Thesis) Download (12MB) | Preview

Abstract

Vegetation, fire and climate are intrinsically interrelated phenomena. Changes in one of these elements will affect the others. Climate change poses an immediate threat to ecosystems, affecting both fire regimes and vegetation around the world. Climate change interacting with other stressors pose the greatest threat to species and ecosystems. Hence, fire as an environmental stressor can exacerbate climate change impacts. The synergistic effects of direct climate change impacts and climate-induced shifts in fire regimes have substantial implications for vegetation and the distribution of vegetation types. For example, fire sensitive vegetation globally already only occupies half its potential distribution due to fire. Changes in fire regime have the potential to increase this pressure and induce fire-driven tipping points where vegetation shifts into an alternative stable state.

The Australian Wet Tropics, a world heritage region of international significance, hosts a range of vegetation types, narrow ecotones, fire regimes and climatic variability in close proximity along steep elevation gradients. Broad vegetation types in the region are rain forest, tall eucalypt forest and savanna, which occur along this environmental gradient. Rain forest vegetation dominates the high rainfall coastal ranges and savanna woodlands dominate the drier inland areas, with a narrow band of tall eucalypt forest in between. Fire is a regular occurrence in Australian savanna and accounts for 80% of the fire activity in Australia. Rain forests on the other hand, are unlikely to burn on a regular basis and are characterised by fire sensitive species. Tall eucalypt forests are also sensitive to frequent fire, but infrequent fire can aid the recruitment of its shade-intolerant canopy eucalypt species. Thus, these vegetation types represent a spectrum of fire regimes and fire tolerance along an environmental gradient.

The aim of this thesis was to address the question 'will climate change interacting with fire affect the distribution of vegetation types'? Fine-resolution vegetation distribution models were developed using topographic, edaphic and climatic variables for current conditions. Complex interactions between vegetation, modelled macroclimate, topography and edaphic factors were detected. These required further exploration to account for the influence of climate relative to other factors, such as fire, competition and vegetation feedbacks. The relative influence of climate and topography on the distributional patterns of vegetation was determined (Chapters 2 and 3) and the potential presence of alternative vegetation states was quantified (Chapter 2). Vegetation feedbacks on microclimate and fire danger were detected (Chapter 3 and 4). Accounting for these complexities in vegetation models is currently a major barrier to making effective predictions of future distributions. The potential influences of model inaccuracies, alternative vegetation states, spatially interpolated climate models, and inability to account for fire indicated modelling would be unlikely to result in accurate outputs. These complexities were explored to ascertain how they affect current distributions and how climate change might impact them, rather than relying on simple model predictions of distribution under future climates.

A spatial analysis of in situ below-canopy micrometeorological and fire danger conditions was assessed from a regional network of monitoring sites within different vegetation types. These were compared with standard meteorological stations and used in an assessment of historic climate and fire danger trends within the region. How spatial climate models related to in situ topoclimate conditions was tested. A review of historic trends and potential future trajectories was made and how data may better explain future projections of future climate, fire and vegetation distribution, but no hard evidence could be presented.

Three vegetation types (rain forest, tall eucalypt forest and savanna) were modelled at a fine-resolution with a geospatial residuals autocovariate technique to assess model capacity to accurately predict current vegetation distribution (Chapter 2). Models generally performed well, but were not near perfect despite use of high-resolution data, robust spatial techniques, full data set and known distributions. This result suggested that there were other important variables influencing vegetation distribution. Comparisons of observed and potential vegetation distributions provided insight into landscape patterns and suggested competition and feedbacks between vegetation types within overlapping environmental niches. Alternative stables states of vegetation and stochastic disturbances by fire are mechanisms also likely to be contributing to vegetation distribution and thus affecting model performance. The relative performance of models between vegetation types and occupancy of potential distributions by other vegetation types indicated that savanna was the most stable vegetation type and tall eucalypt forests the least. Tall eucalypt forests, a threatened ecological community, occupied a narrow environmental space between rain forest and savanna. They are exposed to both long-term shade intolerance from rain forest and short-term frequent fire intolerance from encroaching savanna fires. Other vegetation types occupied large areas of the modelled core environmental niche for tall eucalypt forest, which instead occupied sub-optimal environmental conditions with very low probability of occurrence. This suggested that tall eucalypt forests are closer to the edge of their environmental niche than the other communities and are likely to be less resilient to additional threats, such as from climate change or increased fire risk. Combined these threats suggest that tall eucalypt forests could be at risk of ecosystem collapse.

Predictions of species or vegetation distribution under current or future climate scenarios are generally based upon spatial interpolated climate data. However, spatial climate data are based on relatively simple interpolation algorithms, which do not accurately capture the idiosyncrasies of montane meteorology. Spatially interpolated climate has seldom been assessed against actual in situ conditions, particularly in complex terrain. The reliability of spatially interpolated climate data in reflecting in situ topoclimate conditions relative to vegetation types was tested.

A network of 32 micrometeorological sites along eight transects encompassing rain forest, tall eucalypt forest and savanna was established throughout the Wet Tropics region. Three years of microclimate measurements were made at each site and were compared with parallel data from a nearby official meteorological station (Mareeba). They were also compared with spatially interpolated climate data extracted for each site (Chapter 3). Microclimate showed significant differences between vegetation types along the environmental gradient and with Mareeba. Temperature, for example, decreased along the environmental gradient from savanna at lower elevations, to rain forest at higher elevations. However vegetation was a better predictor of microclimate than topography (up to 99% of overall model performance), suggesting the potential effect of vegetation feedbacks on microclimate conditions. This was consistent with case studies of alternative stable state theory for the region. Interpolated climate variables did not relate well and were generally poor predictors of in situ microclimate. Again, vegetation was a better predictor of micrometeorological conditions than spatially interpolated climate, contributing up to 90% of overall model performance. Biota respond to topoclimate conditions, suggesting that spatially interpolated climate data alone is unlikely to reliably predict vegetation distributions under any climate scenarios. Incorporating vegetation type, topographic, edaphic and meteorological data in distribution or bioclimatic modelling will result in more meaningful and realistic models.

Climate and fire interact and can strongly affect vegetation distribution, particularly fire sensitive vegetation. Fire danger is a metric that assesses fire risk as a function of climate. The McArthur's Forest Fire Danger Index (FFDI) was calculated from microclimate data for each of 32 sites within the three vegetation types (Chapter 4). These were compared with parallel FFDI calculated for a key official meteorological station at Mareeba. There was a strong association of the Mareeba FFDI values with those from the three vegetation types, albeit they were substantially lower. FFDI values were significantly different between each vegetation type. Values decreased from more open vegetation (savanna), through to closed vegetation (rain forest), a pattern that was consistent across each transect. Only very rarely would rain forest be flammable, despite being adjacent to highly flammable savannas. These results demonstrated a stronger effect of vegetation type on fire danger (as well as microclimate), compared to topography, consistent with a fire – vegetation feedback, which is associated with alternative stable state theory. However, fire restricts rain forest to half their potential distribution around the globe, suggesting that fire is a stronger influence on vegetation distribution than any microclimatic feedbacks that might suppress fire and prevent it from encroaching into fire-sensitive vegetation.

Distribution models were concluded to be too inaccurate to predict how climate change might influence vegetation distribution at a scale relevant to existing distributions of vegetation and biota because: spatially interpolated climate data does not capture extreme events or accurate represent topoclimate conditions; inability to account for vegetation feedbacks and alternative stable states, and vegetation distribution models using current climate were not perfect. All of these issues complicate the climate - vegetation relationship, making a simple modelling strategy questionable without factoring in these complexities and how climate change might impact them. Other methods were used to assess implications of climate change and fire on vegetation distribution. This was done by assessing recent meteorological trends and determining likely trajectories of change in climate and fire danger at a fine-scale within the region. Variability, extremes and trends in climate and fire danger were identified for two key sites and compared with projected future climate trajectories.

Observed daily meteorological data at Cairns (1890-2010) and Mareeba (1957-2010) were analysed for trends in climate and fire danger, including variability and extremes (Chapter 5). Known relationships between Mareeba climate and fire danger, with those for the three vegetation types (Chapter 3 and 4) were used to extrapolate historic climate and fire danger conditions for those vegetation types. Cairns and Mareeba displayed consistent trends for some variables, but opposing trends for others. There were significant increases in all fire danger trends, including average and extreme FFDI at Cairns between 1890 and 2010, however, few fire danger trends were significant at Cairns between 1957 and 2010. Mareeba had no significant trends(1957-2010), but some noteworthy trends were near-significance. These near-significance trends indicated a possible increase in extreme fire danger, but also a possible decrease in average fire danger conditions. Climatic variables underlying FFDI calculation contributed in varying ways to these results. Temperatures increased at both sites, however, rainfall showed no trend at Cairns, but an increasing trend at Mareeba. Climatic trends for each of the vegetation types were expectedly consistent with trends at Mareeba, but with different values.

Historic climate and fire danger trends were broadly consistent with future climate projections. Intra-regional trend variation may help explain some of the uncertainty and weak climate projections made by coarse projections for the Wet Tropics region. With site-specific intra-regional trend information regional variability can be assessed. Observed climate patterns and trends were not consistent throughout the region and by 'averaging' these conditions across the region, as is done with global climate models, the detail, accuracy and certainty of trends is weakened by masking variability. Historic trends provide evidence of some consistent and some divergent climate change trajectories within the region. If increasing extreme FFDI is real and is matched with increased fire occurrence, then this may pose a threat to fire sensitive tall eucalypt forests and rain forests, with implications for fire management practices in these ecosystems.

The potential presence of alternative stable vegetation states and vegetation feedbacks on microclimate and fire danger were quantified for rain forest, tall eucalypt forest and savanna (Chapter 2). The presence of vegetation types existing in disequilibrium with climate and in alternative states suggests that climate change alone is unlikely to drive vegetation change or shift in stable states. However, a potential increase in extreme fire danger conditions could disrupt existing alternative stable states in vegetation to drive vegetation change. This would impact fire sensitive tall eucalypt forests and rain forests. This threat is exacerbated by a legacy of European disturbance and increased vulnerability of forest types. The interaction of multiple stressors, including climate change and fire has the capacity to destabilise vulnerable vegetation, promoting shifts into alternative stable states. Tall eucalypt forests, an already threatened vegetation type, are the most at risk of ecosystem collapse in this region.

Will climate change interacting with fire affect the distribution of vegetation types? Evidence including the relative vulnerability of vegetation types and potential changes in fire danger, suggest that vegetation distributions could change. The tendency is for an expansion of savanna, at the expense of a contraction of tall eucalypt forest and rain forest edges, presumably mediated by fire. But an increase in fire danger alone is not enough. Ignitions are required for changes in fire potential to be realised. Ignition sources, including their frequency, timing and location may be a key stressor to exacerbating climate and fire risk to vegetation. Appropriate fire and ignition management could be used to mitigate this risk. How we manage ignitions and fire in the landscape in consideration of ecosystem risks could affect future vegetation distributions. Appropriate fire regimes in the present may be of greater concern to the resilience, distribution and persistence of vulnerable vegetation in the future, than directly from climate change.

Fire frequency is perhaps more strongly linked to ignitions than to climate directly. The overwhelming majority of ignition sources causing fire frequency are anthropogenic. Arson and perverse fire management actions have the potential to exacerbate potential climate change impacts, to tip the balance and drive change in the distribution of vegetation types, including shifts into alternative vegetation states. The main issue may not be about how climate change and a change in fire danger could affect vegetation, but how fire is managed into the future. Fire managers and agencies must adapt to increased fire danger conditions if they are to prevent potential vegetation change and distributional shifts. Consistent with other national and regional reports, it is suggested that reducing landscape pressures, such as fire, are the best option for mitigating climate change impacts on ecosystems. The message is, never mind the warming, watch out for the fire!

Item ID:	46828
Item Type:	Thesis (PhD)
Keywords:	adaptation management, alternative stable states, alternative state stable theory, bioclimatic models, climate, climate change, environmental gradient, feedback, fire danger rating, fire ecology, fire weather danger rating, fire, forest boundaries, heat stress, historic, interpolated climate, micrometeorology, montane biodiversity, refugia, rain forest, residuals autocovariate regression, savanna, spatial autocorrelation, spatial interpolation, tall eucalypt forest, temperature buffering, topography, vegetation distribution model, Wet Tropics, Wet Tropics of Queensland World Heritage Area
Related URLs:	https://researchonline.jcu.edu.au/16280/ https://researchonline.jcu.edu.au/24309/
Additional Information:	Publications arising from this thesis are available from the Related URLs field. The publications are: Chapter 3 & Appendix 3.1: Shoo, Luke P., Storlie, Collin, VanDerWal, Jeremy, Little, Jeremy, and Williams, Stephen E. (2011) Targeted protection and restoration to conserve tropical biodiversity in a warming world. Global Change Biology, 17 (1). pp. 186-193. Chapter 4 & Appendix 4.1: Little, Jeremy K., Prior, Lynda D., Williamson, Grant J., Williams, Stephen, and Bowman, David (2012) Fire weather risk differs across rain forest-savanna boundaries in the humid tropics of north-eastern Australia. Austral Ecology, 37 (8). pp. 915-925.
Date Deposited:	13 Feb 2017 01:46
FoR Codes:	05 ENVIRONMENTAL SCIENCES > 0502 Environmental Science and Management > 050202 Conservation and Biodiversity @ 50% 05 ENVIRONMENTAL SCIENCES > 0501 Ecological Applications > 050101 Ecological Impacts of Climate Change @ 50%
SEO Codes:	96 ENVIRONMENT > 9603 Climate and Climate Change > 960307 Effects of Climate Change and Variability on Australia (excl. Social Impacts) @ 100%
Downloads:	Total: 225 Last 12 Months: 10
	More Statistics