By the year 2100, the world will quite possibly be warmer by 4 oC, with more frequent extreme fire weather, higher CO2, and new technologies relevant to ecosystems, such as gene drives and cell-culture meat. Ecosystems may change a lot. Ecological science needs to move beyond "conservation" and "restoration". Ecosystem change scenarios are being framed as states and transitions. A 2024 ESA workshop considered possible future states for four different Australian ecosystems: an alpine grassland, a subtropical rainforest, and two eucalypt woodlands. Four themes emerged repeatedly. They were often landscape ecology themes, involving spatial transfer of ecological forces or entities. (1) Increased fire risk is expected to shorten intervals between crown fires, making it difficult or impossible to sustain ecosystem states that develop only a long time after fire. (2) Do we or don't we want species to move polewards and uphill? Currently we favour corridors but worry about transplants. (3) Decline of some land uses such as livestock grazing may open opportunities for new or transplanted ecosystems. (4) There exist multiple criteria for ecosystem condition or nature positivity. Among the alternative criteria available, it is more important to keep as many species as possible in existence, than to keep them in the same locations and combinations as currently.

Global change is driving rapid shifts in vegetation, forcing many plant species to migrate to track suitable climates. Yet habitat fragmentation increasingly impedes such movement, isolating populations and threatening their long-term viability. While population genomic data provide valuable insights into connectivity and gene flow, they offer limited understanding of the mechanisms enabling or constraining movement and survival. We argue that combining genomic data with species-specific functional traits can help bridge this gap. Key traits, such as seed mass, dispersal mode, pollination syndrome and self-compatibility, shape a species capacity to maintain connectivity and reproduce in fragmented landscapes. These same traits may also influence species turnover and persistence, linking dispersal and reproductive ecology with population viability. Despite this, the relationship between traits and genetic diversity – the raw currency of adaptive capacity – remains largely untested. We propose a framework that integrates trait-based ecology with population genetics to better understand how connectivity arises at a landscape scale and what it means for species' persistence. This approach can reveal why some species thrive while others cannot maintain viable populations in habitat fragments. Ultimately, it may allow us to better predict which species will keep pace with climate change and which are at greatest risk of local extinction in fragmented landscapes.

Landscape-scale fluxes of carbon and water depend on plant cover. To understand how these fluxes might change under different conditions, plant anatomical traits e.g. growth form, chemical traits e.g. leaf N and physiological traits e.g. leaf turgor loss point (TLP) or maximum stomatal conductance (gsmax) are used to parameterise ecosystem models. But how much of the vegetation cover across the landscape do we have trait data for? Ranking species by their total cover at a continental scale could help coordinate research efforts of plant physiologists. To estimate the most abundant species in Eastern and Central Australia by total area covered, we used ~100,000 vegetation plots from Harmonised Australian Vegetation plot (HAVplot) database overlaid across the National Vegetation Information System. Just 113 plant species (< 1%) and 18 plant genera make up 50% of vegetation cover across this part of the continent. Ca 10% of species and genera (1130 and 140 respectively) make up 90%. The most important species are predominantly from drier lands that occupy large areas in northern and central Australia. Morpho-anatomical traits were available in the Austraits database for species that make up between 50 – ~100% of vegetation cover, e.g. plant growth form data for > 99% of vegetation cover, leaf N for 50% . Traits needing physiological measurement had distinctly lower coverage of important species, e.g. TLP or gsmax with data for < 5% of cover. Trait research targeted at the relatively few species and genera that contribute the most cover could be useful for ecosystem modelling at the continental scale, and more broadly for advancing Australian ecology.

Dynamic global vegetation models (DGVMs) are widely used to predict vegetation responses to climate change, but current DGVMs contain outdated representations of physiological responses to rising temperatures. Empirical evidence demonstrates that plants thermally acclimate their photosynthetic and respiratory rates in response to rising temperatures, but most DGVMs do not represent this process. Furthermore, default physiological parameters of DGVMs are generally based on data for northern hemisphere species, potentially leading to unrealistic values when applied in an Australian context. We measured photosynthetic and respiratory temperature responses of dominant rainforest and wet sclerophyll tree species along the Australian east coast latitudinal gradient. For a case study in Tasmania, we parameterised the DGVM LPJ-GUESS based on our field data and modified LPJ-GUESS’s photosynthetic equations to test whether these updates improve predictions of gross primary productivity (GPP) measured at a flux tower in Tasmania. Our field data demonstrates that the temperature optimum of photosynthesis increases with mean annual temperature, but that this response is stronger for rainforest than for wet sclerophyll tree species. For our Tasmanian case study, inclusion of photosynthetic temperature response parameters and revised photosynthetic equations in LPJ-GUESS improves predictions of GPP relative to the trunk version of the model. Our results demonstrate promising support for the ability of revised photosynthetic parameterisation and implementation to improve DGVM predictions, and suggest that rainforest and wet sclerophyll tree species may have different capacities and/or strategies to respond to rising temperatures. As global temperatures rise under climate change, it is crucial that we develop understanding of forest vulnerability to warming and how it might differ between forest types and along thermal gradients.

Eddy covariance (EC) systems provide state-of-the-art measurements of ecosystem-scale carbon fluxes, capturing net and gross exchange over footprints spanning metres to kilometres. While empirical models can interpolate between EC-towers to estimate regional heterogeneity, the observed period is often insufficient to capture long-term climate trends that might take 30 years or longer to emerge from interannual variability. The EC-method is also not well suited to observing the effects on ecosystem fluxes of large-scale disturbances such as bushfires. Process-based dynamic vegetation models, which simulate vegetation dynamics based on climate inputs and soil and vegetation parameters, may be calibrated against EC-observations. This provides a methodology to extrapolate from flux tower measurements to larger temporal and spatial scales. We demonstrate this approach for subalpine woodlands in South-eastern Australia, which have warmed by approximately 1.10 °C at 1700m ASL between 1910 and 2019 and where <1% of the area remains long unburnt. EC-observations from a recently undisturbed subalpine woodland at 1600m ASL (2024-2025) indicate a net carbon sink. To understand why this woodland is a sink for carbon, we simulated vegetation function and dynamics from 1960-2025 using the model LPJ-GUESS, prescribing fire in the 2003 burn footprint. We explore how climate change, vegetation demography, and soil biogeochemistry interact to produce a carbon sink in the model. This approach links short-term flux measurements to long-term ecological processes and disturbance, providing insight into the drivers of carbon balance in a changing subalpine landscape.

Although wildfires can facilitate the regeneration of new tree cohorts, the climate change-driven decrease in intervals between high-severity wildfires can exacerbate forest degradation, posing threats to the survival of fire-sensitive obligate seeders. In response to observed increases in fire occurrence in temperate forests, several studies have modelled future fire risk to identify the potential for extinction, but few studies have investigated recent fire histories to assess the potential for survival and persistence of these forests.

I used publicly available fire history records from government departments to summarise major fire events and fire years between 2003 and 2025. I then analysed the survival proportion of Australian mainland alpine ash (Eucalyptus delegatensis) forests in relation to varying fire frequencies and severities. Fire severity mapping was used to infer the likelihood of forest mortality (associated with high-severity fire classes) versus forest persistence (associated with low-severity fire classes).To evaluate regeneration survival under climate change, I applied an exponential survival function to fit the proportion of area remaining unburned since last fire. I used fire intervals to represent the complementary cumulative distribution function of regeneration lifetime.

I found that in VIC, mortality-inducing high-severity fire in areas burnt only once was approximately 65% greater than in areas burned multiple times. Exponential survival analysis further indicated that fire risk remains relatively low during the first five years following a fire but increases thereafter. These findings underscore the need for careful monitoring of regenerating stands that remain in productively immature stands. Striking a balance between inaction and overaction is essential for crafting adaptive and context-specific management practices for these stands.

This presentation will highlight the impacts of different fire frequencies and severities from 2003 to 2025 in alpine ash forests, and examine effective management strategies through the Resist–Accept–Direct (RAD) decision framework.

Alpine Ash (Eucalyptus delegatensis) is a dominant species in approximately 380,000 hectares of Victoria’s native forests. Alpine Ash are typically killed by bushfire and takes 15-20 years to re-establish canopy seed stores, and seed is not retained in soil seedbanks. In the last 20 years the species has been subject to widespread short-interval bushfires and until 2024 was commercially harvested for timber. As a result, a large extent of native Ash forest is currently considered below reproductive age. Annual flowering assessments inform landscape and seasonal seed availability which is important to guide recovery programs. For example, in the event a of a bushfire, knowledge of flowering and seed availability can assist when assessing a forest’s capacity to self-recover from natural canopy seed, or otherwise determine if intervention may be required.
Currently the spatial detection of flowering extent for Alpine Ash relies on high-risk fixed wing aerial assessment in conjunction with ground-based observations and monitoring. Collected aerial data is sporadic in nature and cannot account for the full Ash extent.
Here we present a new method in the landscape scale detection of flowering in Alpine Ash utilising freely available satellite data. Using historical flowering assessment data and high-resolution satellite data, we generated a machine learning model to predict locations of flowering activity across the spatial range of Alpine Ash throughout the flowering season. This data is more complete than aerially acquired data and will improve direction for seed collection and monitoring and can also be used to generate landscape scale maps showing mature Ash forests at risk where seed crops are predicted to be inadequate. Future use of landscape scale flowering data could be used to help investigate environmental factors which lead to flowering events in this at-risk forest community, of which the drivers of flowering are not fully understood.