Frameworks for identifying priority plants and ecosystems most impacted by major fires

Keywords: disease, drought, fire frequency, fire history, fire planning, fire regime, fire response, fire severity, fire spatial extent, herbivory, life history traits, recovery actions, threats, weeds.

Australian vegetation has been shaped by fire over evolutionary timescales (Kemp 1981; Lamont et al. 2019) because fire is a major factor affecting the life histories and ongoing persistence of plants, animals and ecosystems. Increasing fire activity in the landscape since the mid- to late-Tertiary (30–60 million years ago) is supported by evidence of increasing deposits of charcoal in the paleorecord, notwithstanding periodic fluctuations (Lynch et al. 2007). Similarly, the rising importance of fire on Australian vegetation through time is evidenced by the co-incident proliferation of traits associated with recovery from fire, such as epicormic resprouting, serotiny, fire-stimulated seed germination and the expansion of fire-prone biomes across the continent (Clarke et al. 2015). Fire continues to maintain contemporary vegetation structure across Australia, with vast areas (∼70%) of the continent’s vegetation being dependant on fire as the major disturbance promoting successional processes (Russell-Smith et al. 2007). Further, the non-linear patterns of fire behaviour define and maintain some vegetation boundaries, most notably between rainforest and sclerophyll communities (Bowman 2000).

Fire regimes vary across the landscape because fire-regime components (frequency, intensity/severity, season and type; Box 1) exhibit biogeographic patterns related to climatic factors at regional, continental and subglobal scales (Bradstock et al. 2005; Murphy et al. 2013; Young et al. 2017). However, fire regimes also vary markedly across the landscape owing to environmental factors such as topography (Miller and Murphy 2017), local weather and vegetation feedbacks (Zylstra 2018), along with anthropogenic factors (Bird et al. 2016). Ecosystems occurring in different climatic zones thus comprise species that evolved under different fire regimes and vary in their sensitivity to changes in fire-regime components, as well as the magnitude and direction of those changes. Just because a species or ecosystem is burnt in a fire does not mean there has been an adverse impact on that entity. Understanding the likely impacts of fires on plant species and ecosystems requires knowledge of how fire regimes affect species and their life histories in the context of recent and historical fire regimes, and the varying sensitivities of ecosystems across the landscape to different components of the fire regime. Persistence of species depends on interactions between population processes and fire regimes (Keith 1996, 2012). Declines may occur when one or more components of the fire regime moves outside a species’ tolerable range. For example, population declines are likely to occur in obligate-seeding or non-resprouting plants (sensu Pausas et al. 2004) if fires recur before these species can replenish their seed banks.

The survival of populations of plants and animal species is strongly influenced by the fire regime (Box 1; Gill et al. 1981; Whelan 1995; Keith 1996; Bradstock et al. 2002, 2012; Bowman et al. 2019), population vital rates (survival, stage transitions, recruitment) in combination with environmental factors such as climatic fluctuations (e.g. drought), anthropogenic impacts (clearing and fragmentation that affect fire behaviour and spread), and herbivory and competition with other species (native or exotic). The impacts of a fire event therefore depend on the interactive effects of environmental factors, fire regime components and species’ life histories. Further, as changing climates reduce plant survival and reproductive rates, the range of fire regimes that facilitate species persistence are predicted to become more restricted (the concept of interval squeeze, Enright et al. 2015), potentially leaving many species and ecosystems vulnerable to decline. More extreme fire weather is predicted under a changing climate, likely increasing the frequency and severity of fires (and potentially reducing fire patchiness and fire refugia), along with the extent and severity of droughts and storms (Bradstock 2010; Cary et al. 2012; Miller and Murphy 2017; Abram et al. 2021), all of which affect ecological responses to fires (Miller and Murphy 2017). The widespread fires that occurred in south-eastern Australia in the summer of 2019/2020 are consistent with predicted outcomes of global warming (Nolan et al. 2020a), as are recent large wildfires in California (Keeley and Syphard 2021) and Mediterranean Europe (Ruffault et al. 2020).

Fire management for biodiversity conservation relies on an ability to accurately identify and implement fire regimes that facilitate species persistence and maintain diverse plant communities and ecosystems. In essence, this means identifying and avoiding those fire regimes that are likely to be detrimental to the persistence of species and ecosystems. Species and associated vegetation evolved under lightning-ignited fires driven by climatic, topographic and edaphic drivers until the arrival of Indigenous Australians (at least 50–60 000 years ago) who used fire and initiated additional human ignitions. In the past 230 years, fires initiated by European settlers include prescribed ignitions for management purposes (e.g. pastoral management and more recently, hazard reduction), accidental ignitions and arson. Ignitions have therefore not only increased in frequency with human intervention, but also in their spatial and temporal distribution across landscapes, seasons and years. Such ignitions have led to fire regimes diverging in different directions from natural patterns through different eras of human activity, as well as because of past and current climate changes. Fire regimes in contemporary Australia are a product of fires that originate from both natural (primarily lightning) and human ignitions (Indigenous burning and planned fires for pastoralism, fuel reduction or conservation, along with arson or accidental fire escapes; Bowman et al. 2020). Understanding of how alterations to fire regimes affect plants and other biota has only recently begun to develop (Parson and Gosper 2011; Keith 2012; Kelly et al. 2020).

Fires of human origin add to the natural complexity of fire regimes in the landscape; however, conversely, European land management practices may also supress the outcomes of natural ignitions through active fire suppression and landscape fragmentation (Parsons and Gosper 2011). Many current applications of fire by humans aim to reduce fuel loads, either over broad-scale landscapes, or in and around areas that contain dwellings and other assets. Fire is also used to promote regeneration in logged forests. Unplanned fires (wildfires resulting from lightning or unintentional human ignitions) combine with this prescribed fire matrix to create complex patterns of fire regimes in different parts of the landscape (Bradstock et al. 2005). The consequences are expressed in the interval, timing and spatial dimensions of fire regimes (Bond and van Wilgen 1996). For example, some patches are frequently burnt or have short intervals between some successive fires, whereas others are rarely burnt or remain entirely unburnt, either because fragmentation has isolated them from fire pathways, or because feedbacks have allowed them to develop low-flammability properties. Event characteristics, including fire severity, seasonality and extent, all vary spatially and temporally across different landscape and pyroclimate types, (e.g. see Russell-Smith et al. (2007) for variation in fire seasonality across Australia). In summary, landscapes have both complex spatial patterns of fire regimes, as well as complex requirements for the persistence of the full diversity of plant species, vegetation communities and ecosystems under recurring fires.

The fires that occurred in south-eastern Australia in the 2019/2020 fire season were among the most extensive of southern Australia’s European era (Davey and Sarre 2020; Nolan et al. 2020a). The fires burnt ∼7 Mha across south-eastern Australia between September 2019 and March 2020 and burnt a greater proportion of Australia’s temperate broadleaf and mixed forest biome than any other global forest biome in the past 20 years (Boer et al. 2020; Collins et al. 2021). Much of the area was estimated to have burnt at high severity (Collins et al. 2021) and the fires affected a substantial proportion of the known ranges of large numbers of vertebrate fauna (Ward et al. 2020; Legge et al. 2022), vascular plant species (Auld et al. 2020; Gallagher et al. 2021), terrestrial (Lindenmayer and Taylor 2020; Keith et al. 2022b) and aquatic (Silva et al. 2020) ecosystems. This created huge challenges for land managers dealing with the conservation of biodiversity and demonstrated the need for frameworks to guide what species and ecosystems are most likely to have been adversely affected as a result of the 2019/2020 Australian wildfires.

The scale and severity (sensu Box 1) of the 2019/2020 fires was clearly unprecedented and influenced by the trend for increased area of forest burning driven by climate change (Canadell et al. 2021). Diagnosing the potential impacts of these fires is not straightforward because fire is a recurring event, even if only rarely, in virtually all of the ecosystems affected (Keith and Tozer 2017; Miller and Murphy 2017). To forecast potential declines, a variety of factors need to be evaluated, including the complex patterns of antecedent fire history, regional variability in pre- and post-fire weather, diverse land uses within the fire footprint and an array of threats posed by alien predators, herbivores, competitors and pathogens that pose considerable risks of adverse impacts to plant species and ecosystem recovery. Timely assessments across the full range of biota and their national or global distributions are essential to inform effective management responses.

Here we present novel predictive frameworks developed to identify, first, plant species and, second, ecosystems that are expected to have been most affected following a major wildfire event, so that those most in need of management actions can be identified. These frameworks go beyond simple reporting of how much habitat was burnt in a fire. The Frameworks assume that risks of decline from fire-related impacts are related to the proportion of the species/ecosystem range affected by a given mechanism of decline, in a comparable way to assessing risk ranking on the basis of geographic distribution in IUCN Red List Criteria (Keith et al. 2018). They build on the understanding that fire regime impacts (Box 1) are important in the response of biodiversity to individual fire events (Bradstock et al. 2012; Keith 2012; Miller and Murphy 2017) as are both biotic and abiotic factors, including threats from, in particular, weeds, pests and pathogens and human impacts. We integrated these components into a scheme that permits comparisons of relative exposure to impacts across either species or ecosystems. The frameworks are decision support tools to guide identification of species and ecosystems that are likely to need active management during post-fire recovery, including what factors may need to be addressed in such recovery. We provide worked examples of applications of these frameworks from the emergency response to the 2019/2020 Australian bushfire season, noting that they are equally applicable to other major wildfire events within Australia and globally. The frameworks

1. provide transparent, logical pathways for decision-making that supports well reasoned strategic policy, management and resourcing responses in an emotionally charged post-fire social environment when fires occur;

2. identify the key issues that need to be addressed to ameliorate impacts of megafires on plants and ecosystems. This includes consideration of the fire regime and its components (Box 1), species’ life histories, ecosystem processes, environmental conditions and other biotic and abiotic threats;

3. use available data to prioritise species and ecosystems for post-fire conservation management, including field-impact assessments and any necessary recovery actions (both immediate and medium- to long-term); and

4. allow for ongoing re-evaluation of which species and ecosystems are most likely to be at risk from future fires and landscape-scale threats.

Here we describe related individual frameworks that were designed for species and ecosystems respectively. The frameworks predicted the likelihood of poor post-fire recovery via three components (mechanisms, sensitivity and exposure):

1. Mechanisms of decline and their interactions, e.g. combinations of life-history traits and threats that make species prone to population declines or local extinctions if they are affected within the spatial extent of a fire;

2. Sensitivity of species and ecosystems to the identified mechanisms of decline (e.g. sensitivity to high fire frequency or to fire-promoted pathogens (such as Phytophthora or myrtle rust)); and

3. Exposure in the landscape where these mechanisms are most likely to be expressed (e.g. the overlap/intersection of species’ or ecosystem distributions with the spatial extent of the fire event being investigated (excluding unburnt patches and refugia that do not burn) AND the particular threat of concern).

Each framework includes 11 criteria or mechanisms of decline (across four main themes, see below) related to the fire regime, environmental conditions, life history of species and concurrent threats (Fig. 1, Tables 1, 2). These mechanisms are as follows:

Fig. 1.  Summary of major elements for post-fire assessment criteria in frameworks for identifying priority plant species and ecosystems most affected by major fires.

Table 1.  Prioritisation framework for assessment of plant-species impacts following major fire events such as the Australian 2019/2020 bushfires. Risk is high (red), medium (orange), low (yellow) and no risk (no colour). Framework criteria are in dark grey, while main subcriteria are in light grey.

Table 2.  Prioritisation framework for assessment of ecosystem impacts following major fire events such as the Australian 2019/2020 bushfires. Risk is High (red), medium (orange), low (yellow) and no risk (no colour). Framework criteria are in dark grey, while main subcriteria are in light grey.

• Components of the fire regime (see Box 1 for explanation terms and potential impacts on biodiversity) that are most likely to have adverse impacts on species or ecosystem recovery through disruption of life-history processes (I, fire sensitivity; II, short fire intervals; III, high fire severity; IV, recruitment failure)

• Fire-environment abiotic interactions, including prevailing environmental conditions (for species: V, drought; VI, erosion; VII, elevated temperatures; or for ecosystems: V, drought; VI, erosion, disturbance or pollution; and VII, altered hydrology)

• Fire-biotic interactions, including sensitivity of species or ecosystems to a fire and biotic threats to post-fire recovery (VIII, herbivore impacts; IX, disease; X, weed invasions)

• Fire–human interactions, including XI, localised anthropogenic disturbances such as disturbances from vehicles or foot traffic, rubbish dumping, clearing of habitat and logging (among others).

The two frameworks (Fig. 1, Tables 1, 2) were developed and implemented nationally in Australia, for species and ecosystems respectively, following the 2019/2020 fire season. The criteria enable each species or ecosystem assessed to be assigned to an ordinal category (‘high’, medium’, ‘low’, ‘no impact’ or ‘data deficient’) on the basis of spatial thresholds drawn from the IUCN risk-assessment protocols (Bland et al. 2017). Categories are indicative of the risk of recovery failure and decline, as follows:

• HIGH: high degree of exposure to the risk of decline. An urgent assessment of initial fire impacts and threats to recovery is required and post-fire monitoring of recovery where impacts are significant.

• MEDIUM: medium degree of exposure to the risk of decline. Assessment of initial fire impacts and threats to recovery is required and post-fire monitoring is recommended.

• LOW: low degree of exposure to the risk of decline. Post-fire monitoring may be conducted opportunistically during site visits.

• NO IMPACT: negligible exposure to the risk of decline or not burnt in the fires.

• DATA DEFICIENT: insufficient data to enable an assessment.

Species or ecosystems should be assessed against as many criteria as possible, depending on available data. Outcomes of each criterion are evaluated concurrently, with the highest category of concern across all criteria being used to allocate a species or ecosystem to an overall category, recognising that factors driving decline may operate independently or interact in complex ways, and hence the criteria are best not combined in an additive or multiplicative manner (Burgman et al. 1999; Keith 2009). This emulates the approach used in established frameworks such as the IUCN Red List Criteria for Threatened Species (IUCN 2022) and IUCN Red List Criteria for Ecosystems (Bland et al. 2017). Separate reporting on each criterion also allows an informed comparison of the factors driving the highest likelihood of decline for groups of interest, such as groups of species or higher taxa (e.g. families), life-form groups, fire-response groups, or regional floras.

Species or ecosystems for which the highest ranking is ‘NO IMPACT’ on the basis of assessments of at least three criteria are assigned an overall rank of NO IMPACT, but otherwise must be assigned to the ‘DATA DEFICIENT’ category until at least three criteria are assessed. This is an important distinction and highlights species or ecosystems with important knowledge gaps that require addressing for their effective conservation management.

While the frameworks require assessment of each criterion individually, interactions among criteria may also arise. In such cases, it is useful to recognise that individual criterion outcomes may be affected by other factors. For example, a pre-fire drought may exacerbate the impact of high fire severity. This may occur where pre-fire drought reduces the capacity for species to resprout post-fire, although prior depletion of carbohydrate reserves and the impact of this effect vary with fire severity. In many such cases, the measurement of impacts (post-fire recovery) will necessarily include outcomes of the interaction among factors because each factor cannot be clearly separated when assessing (or sampling) post-fire impacts. In other interactions among criteria, there may be temporal differences in the expected timing of impact; for example, the interaction between high fire frequency and post-fire herbivory is a two-step process, with initial impacts of high fire frequency and subsequent impacts of herbivory on recovery. When examining the strength of these effects through one-off sampling of post-fire recovery, often all that can be measured is their combined interaction. Hence, in most cases, major interaction effects are included by recognising the combined effects on species or ecosystems under each driving criterion in the frameworks. When the level of impact of individual factors within interactive effects can be identified, one option is for the rank to be increased by one level when there is a synergistic (or additive) interaction of threats, and decreased by one level where there is a compensatory interaction (in the same fashion as application of the IUCN Red List criteria at regional or national scales; IUCN 2012).

Assessing priorities for post-fire conservation action requires data on the spatial distribution and fire-response traits of species and ecosystems, as well as the distribution, extent and severity of threats and other relevant environmental factors. Assessing impacts after a major fire event specifically requires data on fire spatial extent and fire-severity mapping for the fire event being investigated, as well as reliable fire-history mapping. Data availability will vary among assessment areas, taxa and ecosystems, and it is important to aim for standardised national fire-related datasets (Bowman et al. 2020). Here, we provide several examples to illustrate the types of data that could be used in the assessments and the need to address standardisation, licensing, transparency and efficient workflows, drawing on the assessments of potential impacts for plants (Auld et al. 2020; Gallagher 2020; Gallagher et al. 2021, 2022) and ecosystems (Keith et al. 2020a, 2022b) conducted in response to the Australian 2019/2020 fire season.

An appropriate target list of species or ecosystems forms the foundation of any assessment of impacts or management needs after a fire event. Ideally, all species co-occurring within the spatial extent of the fire event should be targeted for assessment. This requires species-occurrence information (see below) and validation of current species nomenclature against a taxonomic authority, including identifying taxonomic synonyms. For instance, the Australian Plant Census provides up-to-date information on the status of taxonomic names of Australian plants, and similar resources exist in many countries (e.g. USA: https://www.itis.gov/) and globally (e.g. The Plant list, http://www.theplantlist.org/ and Plants of the World Online http://www.plantsoftheworldonline.org/; Catalogue of Life: https://www.catalogueoflife.org/). Checking of taxonomy can be automated through workflows in R using packages such as taxise (Chamberlain and Szöcs 2013), or via online portals such as the Taxonomic Resolution Naming Service (TNRS; https://tnrs.biendata.org/).

For ecosystems, the analogous data required are a typology that identify all the ecosystem types within a study area at a suitable level of thematic resolution. The units of the typology should represent the critical processes and dependencies that sustain ecosystem components and functions. IUCN’s Global Ecosystem Typology (GET, Keith et al. 2020d) provides a suitable framework for this purpose. The GET facilitates the integration of national or jurisdictional-level typologies based, for example, on the classification of vegetation communities under a hierarchy of types, with similar functional responses to fire regimes in combination with the posited threats.

National and international databases can be used to access georeferenced information on occurrences of species. For instance, international resources such as the Global Biodiversity Information Facility (GBIF; https://www.gbif.org/) or national resources such as the Australasian Virtual Herbarium (AVH; https://avh.chah.org.au/) provide open access to vouchered records of species occurrence. Herbarium specimens have several advantages over unvouchered records of species-occurrence information, including verification of taxonomic identity by experts, the ability to trace information back to a vouchered collection and to update analyses following taxonomic revisions (Heberling and Isaac 2017). However, bias in spatial patterns of sampling in herbarium specimens towards roads and access tracks may limit the accuracy of species ranges inferred from them (Haque et al. 2017). To address this issue, secondary sources of information on species range may be used in conjunction with herbarium occurrence data, such as records with lower standards of verification (e.g. research-grade observations in iNaturalist, https://www.inaturalist.org/), data from systematic vegetation surveys (e.g. BioNet systematic flora surveys, https://www.environment.nsw.gov.au/atlaspublicapp/UI_Modules/YETI_/FloraSearch.aspx), expert-derived range maps, or the spatial projections from species distribution models. Ideally, multiple lines of evidence about species occurrence should be used in parallel when applying the frameworks and the highest risk (most precautionary) category across the different sources used to rank species.

Steps should be taken to minimise spatial inaccuracies of species-occurrence records. For example, the full, original, as-held data for any unique threatened species records should always be obtained directly from the original custodian to ensure accurate locality descriptions and georeferences (because sensitive information is often denatured prior to incorporation into third-party repositories). Checking georeferences for consistency against the location description allows detection of erroneously georeferenced points that may otherwise bias predictive modelling of species distributions or population-level assessments of fire impact/escape. This highlights the need for resourcing to maintain and update databases of reliable species occurrence records so that georeferences are an accurate representation of the location information.

For ecosystems, maps of proxy units such as vegetation types can be obtained from government repositories or scientific literature, or generated via a wide range of remote-sensing and modelling approaches (Bredenkamp et al. 1998). Similar verification and error-correction procedures should be applied as those recommended above for species.

The framework criteria use a suite of traits to identify species or ecosystems that are susceptible to different fire-related causes of decline. Traits are the measurable characteristics of organisms that shape their ecological performance (Westoby et al. 2002; Gallagher et al. 2021) and in the context of post-fire assessment frameworks, specifically include factors such as habit, woodiness, fire-response strategy (capacity to resprout versus only recruit via post-fire seedings, sensu Pausas et al. 2004; Clarke et al. 2015; Prior and Bowman 2020), seed-bank type (canopy- vs soil-stored), seed-dormancy types (physical, physiological and morphophysiological, among others) as well as various germination requirements and drought-avoidance mechanisms. It is now possible to source data on plant species traits directly from large, aggregated databases such as TRY (Kattge et al. 2020), LEDA (Kleyer et al. 2008), BIEN (Maitner et al. 2018) and AusTraits (Falster et al. 2021), as well as from more bespoke datasets for particular plant groups (e.g. bryophytes; Bernhardt-Römermann et al. 2018; or plant parts; e.g. roots; Iversen et al. 2017). Fire-response traits may also be sourced from environmental management agencies that use ecological information in planning for prescribed burning.

For some species, multiple observations of the same trait may lead to conflicts in assigning species to categories (e.g. because of natural variation, variation owing to particular factors such as fire severity, observer error etc.), and strategies are required for choosing the appropriate trait value. For instance, consensus across datasets may be used to assign a species as woody or non-woody. Similarly, continuous or ordinal trait values may need to be partitioned or aggregated to form appropriate categories. This may include grouping information on growth forms (e.g. tree, shrub, grass, herb) and standing plant longevity (e.g. annual, biennial, perennial) to inform woody and non-woody categories. As per species-occurrence records, prior to use, taxonomic names associated with trait observations need to be assessed for consistency and taxonomic synonyms to ensure alignment with the species list of interest.

The requirement for large, comprehensive datasets on plant traits and plant responses to components of the fire regime (including variation in responses), in addition to comprehensive and spatially accurate databases of species-occurrence records, highlights the need for forward planning and investment. These databases need to be established and maintained at the national scale in advance of future megafire events to enable a rapid and effective responses.

The 11 criteria representing key mechanisms of post-fire decline identified in each framework (Tables 1, 2) and the potential methods for assessing each criterion are presented below.

Some plant species, including some functionally important species within ecosystems, have no means of in situ persistence through fire events, because they lack regenerative organs, and on-site propagule storage (e.g. R-P- of Pausas et al. 2004) or long-distance dispersal traits to facilitate recolonisation. Some species or ecosystems may currently persist in disequilibrium states and may be incapable of re-establishment under present-day conditions. Examples include certain rainforest species (Boxes 2, 3) and peatlands, forests or heaths dominated by paleo-endemic species (e.g. Kirkpatrick et al. 2010; Bliss et al. 2021). A single fire event may eliminate these entities or substantially diminish their role in the community, an effect that persists until they slowly disperse and re-establish from unburnt patches.

There is no globally comprehensive database on fire sensitivity of plant species, although geographically scoped fire-response databases and classification schemes (e.g. Gill and Bradstock 1992) can inform fire sensitivity, as may data held in global or regional databases such as TRY (Kattge et al. 2020) and AusTraits (Falster et al. 2021). Pausas et al. (2004) compiled data on a small number of species that lacked regenerative organs and seed banks to support post-fire recovery by sprouting or seeding respectively (R-P-), and this was also flagged by Prior and Bowman (2020). Information on the susceptibility of species to single fire events can also be based on available scientific literature (e.g. Athrotaxis, Bliss et al. 2021) and expert opinion.

Species: long-lived trees that are prone to collapse from prolonged basal charring during fires are candidates for Criterion I (Table 1, Boxes 2, 3). For example, Gallagher (2020) used trait and spatial information to identify 463 rainforest-tree taxa greater than 30 m in maximum height (taking height as a proxy for longevity) using the AusTraits database. Long-lived tree species sensitive to fire may also occur in low-productivity environments and require substantial time (>30–50 years) to regenerate and set seed post-fire.

Ecosystems: Keith et al. (2020a) identified ecosystems that rarely experience fire and were sensitive to a single fire event, because species with key structural or functional roles lack regenerative organs and seed banks that enable autogenic recovery. The main ecosystem types that include such species are rainforests and samphire shrublands and herbfields, sphagnum bogs, as well as peatland ecosystems, because peat is combustible should substrate fires occur, and may take many decades or centuries to re-accumulate to similar depths. Most rainforest trees have thin bark (Lawes et al. 2013) and, even though a surprising number has basal regenerative buds, top-kill resulting from fires with scorch heights as low as 2 m may result in structural transformation of large stands that may take many decades to re-establish their mesic micro-climate, arboreal substrates for tree-dependent flora and fauna and structural complexity.

Exposure of fire-sensitive species and ecosystems to any fire event can be estimated by intersection of their distributions with the fire extent (e.g. for the Australian 2019/2020 fires – the National Indicative Aggregated Fire Extent Dataset; Department of Agriculture, Water and the Environment 2020).

Ecosystems: Keith et al. (2020a) identified ecosystems that rarely experience fire and were sensitive to a single fire event, because species with key structural or functional roles lack regenerative organs and seed banks that enable autogenic recovery. The main ecosystem types that include such species are rainforests and samphire shrublands and herbfields, sphagnum bogs, as well as peatland ecosystems, because peat is combustible should substrate fires occur, and may take many decades or centuries to re-accumulate to similar depths. Most rainforest trees have thin bark (Lawes et al. 2013) and, even though a surprising number has basal regenerative buds, top-kill resulting from fires with scorch heights as low as 2 m may result in structural transformation of large stands that may take many decades to re-establish their mesic micro-climate, arboreal substrates for tree-dependent flora and fauna and structural complexity.

Exposure of fire-sensitive species and ecosystems to any fire event can be estimated by intersection of their distributions with the fire extent (e.g. for the Australian 2019/2020 fires – the National Indicative Aggregated Fire Extent Dataset; Department of Agriculture, Water and the Environment 2020).

The framework for species has been applied to plants after the Australian 2019/2020 fires with both a national focus (Gallagher 2020; Gallagher et al. 2021, 2022) and a regional one (the state of NSW, Auld et al. 2020). These assessments applied the framework to assess fire impacts across a species’ national distribution so that priority outcomes were not biased by local/regional impacts. Similarly, Keith et al. (2020a, 2022b) applied the ecosystem framework at the national scale to ecosystems recognised in national and state jurisdictions (the latter particularly for ecosystems largely endemic to a particular state).

The frameworks can be applied to any area of management interest, but it is also important to consider the total distribution of species or ecosystems to provide relevant context. This may help inform management priorities at the scale of reserve networks or bioregions, with species or ecosystems exposed to high impacts throughout their distribution being assigned a higher priority than those exposed to high impacts in the assessment area, but not throughout their range. The benefits of applying the criteria at global or national scales for megafires include a more comprehensive understanding of relative risks to different functional groupings of plants or types of ecosystems, and the factors driving reduced post-fire recovery, along with targeting of subsequent management actions to where they are most needed. Improvements in national datasets on plant species responses to fire and ecological traits will enhance the capacity of such comparisons. Gallagher et al. (2021) and Godfree et al. (2021) both estimated the percentage of Australian native plant species affected by the 2019/2020 fires, which helped convey the magnitude of potential impacts to a wide audience. However, the additional application of the risk framework for plant species allowed Gallagher et al. (2021) to identify key mechanisms driving potential decline, which is the first critical step towards development and implementation of strategies for risk reduction and impact mitigation.

The frameworks can be applied to threatened species and ecosystems to identify those most in need of management action. For example, Gallagher (2020) and Gallagher et al. (2021) found that some 67 nationally threatened plants in Australia were likely to decline as a result of the 2019/2020 fires (i.e. medium or high risk) owing to too frequent fire. The framework has also been used to identify priority species for possible statutory listing, and a number of conservation status assessments are currently underway (see https://www.dcceew.gov.au/environment/biodiversity/threatened/seap). This fills an important gap, given that the current statutory listings markedly underestimate the number of plant species (and ecosystems) that are likely to be threatened (Alfonzetti et al. 2020). Auld et al. (2020) found that over 100 nationally threatened plants that occurred in NSW were likely to decline as a result of the 2019/2020 fires, with an additional 60 NSW endemic species listed as threatened either under the IUCN Red List of Threatened Species or the NSW Biodiversity Conservation Act also being likely to decline. These latter 60 species potentially need to be added to the national Environment Protection and Biodiversity Conservation Act threatened species list. Finally, some 230 NSW endemics that are not currently listed as threatened under any legislation were identified by Auld et al. (2020) as potentially declining because of the megafires. These species are candidates for possible threatened species listings, should the predicted impacts and threats to natural recovery result in population declines in response to the 2019/2020 fires.

As well as individual species and ecosystems, the framework criteria could be applied to a range of other priority conservation assets affected by megafires, including keystone species, refugia and key biodiversity areas (sensu IUCN 2016). For example, Gallagher et al. (2021) examined the spread of impacts for the 2019/2020 Australian megafires across different plant family groupings, highlighting high levels of impacts in three major families (Proteaceae, Fabaceae and Myrtaceae).

Despite uncertainties in assessments, in many cases the impacts of a number of key threats are sufficiently understood to recommend specific conservation actions for affected species and ecosystems (Table 4). Some actions are needed in the short term (0–2 years after fire), whereas others may require longer-term implementation (>5 years post-fire, Fig. 2). Key recovery actions should be identified for species or ecosystems that are most likely to decline or fail to recover after fire. Prioritisation and implementation of actions should recognise that fire plays a key role in the life history of many plant and animal species, and that the abundance of certain groups such as obligate seeders may fluctuate in response to fire as a consequence of their natural population dynamics (i.e. sustained population declines should be distinguished from population fluctuations, sensu IUCN 2022). Other species may recover vegetatively after a fire, but their detectability may be highest during post-fire flowering (e.g. pyrogenic species of terrestrial orchids), with many plants persisting below ground in the intervening period. In these cases, population trends can be difficult to detect and require monitoring designs that are cognisant of life-history dynamics.

Table 4.  Priority actions likely to be needed for species or ecosystems at high risk, or otherwise a priority, following major fires such as Australian 2019/2020 bushfires. All criteria require an immediate priority action of field inspections to assess the impacts of threats to recovery.

Fig. 2.  Context of framework criteria for prioritisation of recovery actions post-fire.

For species and ecosystems ranked with the potential for medium to high declines (Table 4), the primary recommendation is to undertake post-fire field surveys. These are essential to assess realised impacts and the degree of recovery, verify causes of decline inferred in the assessment, identify any emerging, previously unidentified threats to recovery, and to devise actions to mitigate the threats. This is particularly critical for species that have sites/populations affected by high fire frequency (Criterion II) and species sensitive to fire (Criterion I). For long-lived rainforest trees prone to top-kill and/or collapse from basal charring (Box 5), there is an urgent requirement to conduct field inspections within the first year post-fire so as to assess the scale of tree loss or damage and the rehabilitation actions required.

Development of risk-reduction strategies is essential if an understanding of impacts is to be translated into positive conservation outcomes for both species and ecosystems. Field surveys should inform the relative conservation priorities for recovery actions and the nature and timing of the actions needed for recovery (Fig. 2, Table 4, Box 5) as well as identifying any new or emerging threats. This is essential to modify and update proposed management actions as circumstances change, and to adjust priorities where required, so as to ensure the best possible investment of the limited resources available for post-fire recovery.

Depending on the particular threats identified, actions required to assist recovery include the following:

• Development of a fire management plan to reduce the likelihood of future fires burning over recovering sites (Criteria I–III). For example, avoiding additional fires (including hazard reduction burns) in all recently burnt habitat is necessary to allow time for recovery of species and ecosystems.

• Protecting unburnt parts of a species’ range, including fire refugia (i.e. no burning or clearing or logging in that habitat) to provide insurance and avoid having the entire species’ range at risk from a future fire recurrence at the one time (i.e. avoid risks from Criterion IV for species) and to protect any refugia.

• Avoid works that may affect surface or subsurface drainage or increase erosion risks (Criteria V and VI (species) or V, VI and VII (ecosystems)). In emergency situations, consider options for supplementary watering if post-fire drought conditions are evident (Criterion V).

• Exclusion or control of invasive herbivores or predators (Criterion VIII).

• Maintain suitable disease hygiene and mitigation (Criterion IX).

• Post-fire weed control (Criterion X).

• Minimise any localised site disturbances (Criterion XI).

• Monitoring species’ recovery to determine the time required

• 1. – to replenish seed banks (especially in obligate seeders);

  2. – for juveniles resprouting plants to become fire resistant; and

  3. – for top-killed trees to recover their stature, structure and reproductive capacity.

Species or ecosystems that show limited post-fire recovery may require further interventions. These may include collections of seed or vegetative material for various conservation purposes, including restoration and translocation, reintroductions and establishing seed orchards (e.g. Bassett et al. 2015; Commander et al. 2018; Martyn Yenson et al. 2021, Box 15). For some species, emergency germplasm collection may be needed (Martyn Yenson et al. 2021), for example because of post-fire impacts of myrtle rust.

Species and ecosystems that have a long evolutionary history of persisting under recurring fires have developed various mechanisms and strategies for recovering their populations or structure, function and composition. Their continued ability to do so is dependent on their tolerance to changing fire regimes, in combination with how other threatening processes compromise resistance, resilience or regenerative responses to the fire event. Here we have highlighted a novel mechanistic approach to rapidly predict the impacts of megafires on plant and ecosystem biodiversity. Key advances on unstructured methods of assessment include the incorporation of (1) fire-regime data (see Box 1; e.g. fire frequency, severity, season, type and fire patchiness), rather than fire extent alone, (2) species life histories, biology and ecosystem properties to define potential mechanisms of decline, and (3) interactions between fire events with biotic and abiotic threats to survival and recovery. The frameworks explicitly recognise that knowledge of the antecedent fire regime in combination with that of coincident biotic and abiotic threats in the landscape are required to predict and understand the impacts of megafire events on species and ecosystems. The frameworks help integrate and contextualise the interpretation of single fire events into fire-regime impacts.

The two assessment frameworks were applied to guide conservation responses to the Australian 2019/2020 mega-fires across southern Australia (Gallagher et al. 2021; Keith et al. 2020a, 2022b). They enabled rapid assessment of potential impacts of large fires on plant species and ecosystems, establishing immediate priorities for post-fire surveys, which are now guiding management and monitoring strategies, as well as new statutory listings to protect threatened species and ecosystems.

The post-fire assessment frameworks support rapid decision-making, in part because of their flexibility to utilise a combination of readily available remotely sensed data, detailed species-trait data, available literature and expert opinion. They allow for addition of further criteria should novel threats emerge that affect species and ecosystem recovery. A structured approach that leverages diverse sources of existing knowledge provides a way of informing prioritisation for resourcing of urgent and on-going field inspections and recovery actions.

Although the frameworks were developed as an immediate response to the unprecedented Australian 2019/2020 wildfires, the frameworks can be applied to any fire and any landscape, provided the combination of fire variables, life-history traits, threats to recovery and environmental variables are known or can be estimated or inferred. We highlight that post-fire surveys of prioritised species and ecosystems are essential to ensure that management priorities are robust and well supported, and to identify the most effective management responses. Finally, the frameworks highlight the need for development and maintenance of national and global plant databases to inform plant and ecosystem responses to fire and other threats.

Data used in this study are available on request from the senior author (tony.auld@environment.nsw.gov.au).

Mark Ooi is one of the current EiCs and Richard Williams is the previous EiC of the Australian Journal of Botany, but they were blinded from the peer-review process for this paper. The authors declare that they have no other conflicts of interest.

This research did not receive any specific funding.