Carl R. Gosper

Contemporary fire regimes in a fragmented and an unfragmented landscape: implications for vegetation structure and persistence of the fire-sensitive malleefowl

Habitat fragmentation alters fire regimes by changing the spatial and temporal context in which fire operates, potentially altering ecosystem state and threatening taxa. In the fragmented wheatbelt of Western Australia, spatial patterns of contemporary fire and their effects on biodiversity conservation are poorly understood. We addressed this by: (1) determining if fire regimes differed between vegetation remnants of differing sizes and uncleared vegetation, using analysis of satellite imagery; (2) determining vegetation structural responses to time since fire in three habitats: mallee- shrub, Acacia shrublands and mallee-heath; and (3) exploring the consequences of these differences, using the fire- sensitive malleefowl (Leipoa ocellata) as a case study. Fire was infrequent in small remnants, more frequent in large remnants, and most frequent in uncleared areas. Key vegetation structural attributes for malleefowl, such as canopy and litter cover, increased beyond 45 years post-fire in mallee-shrub, reached a plateau in mallee-heath after 30-40 years, and declined in Acacia shrublands after 25-40 years. Senescence in long-unburnt vegetation, combined with rare contemporary fires, suggest progressive decline in habitat quality of Acacia shrublands for malleefowl in the wheatbelt. In the adjacent, continuously vegetated landscapes, more frequent (and extensive) fires in structurally developing mallee- shrub communities are of concern for malleefowl conservation.

A synthesis of postfire recovery traits of woody plants in Australian ecosystems

Postfire resprouting and recruitment from seed are key plant life-history traits that influence population dynamics, community composition and ecosystem function. Species can have one or both of these mechanisms. They confer resilience, which may determine community composition through differential species persistence after fire. To predict ecosystem level responses to changes in climate and fire conditions, we examined the proportions of these plant fire-adaptive traits among woody growth forms of 2880 taxa, in eight fire-prone ecosystems comprising ~87% of Australias land area. Shrubs comprised 64% of the taxa. More tree (>84%) than shrub (~50%) taxa resprouted. Basal, epicormic and apical resprouting occurred in 71%, 22% and 3% of the taxa, respectively. Most rainforest taxa (91%) were basal resprouters. Many trees (59%) in frequently-burnt eucalypt forest and savanna resprouted epicormically. Although crown fire killed many mallee (62%) and heathland (48%) taxa, fire-cued seeding was common in these systems. Postfire seeding was uncommon in rainforest and in arid Acacia communities that burnt infrequently at low intensity. Resprouting was positively associated with ecosystem productivity, but resprouting type (e.g. basal or epicormic) was associated with local scale fire activity, especially fire frequency. Although rainforest trees can resprout they cannot recruit after intense fires and may decline under future fires. Semi-arid Acacia communities would be susceptible to increasing fire frequencies because they contain few postfire seeders. Ecosystems dominated by obligate seeders (mallee, heath) are also susceptible because predicted shorter inter-fire intervals will prevent seed bank accumulation. Savanna may be resilient to future fires because of the adaptive advantage of epicormic resprouting among the eucalypts. The substantial non-resprouting shrub component of shrublands may decline, but resilient Eucalyptus spp. will continue to dominate under future fire regimes. These patterns of resprouting and postfire seeding provide new insights to ecosystem assembly, resilience and vulnerability to changing fire regimes on this fire-prone continent.

Repeated disturbance through chaining and burning differentially affects recruitment among plant functional types in fire-prone heathlands

Managing fire regimes is increasingly recognised as important for biodiversity conservation in fragmented agricultural landscapes in fire-prone regions. In the global biodiversity hotspot of south-west Western Australia, chaining and burning is a novel technique for facilitating fire management. Vegetation is first dislodged using a chain, then after a period of curing, burnt. The effects on plant communities are largely unstudied, despite the potential consequences of combining two disturbance events. We hypothesised that outcomes would vary depending on plant functional types defined by disturbance response. We compared plant community composition and recruitment and resprouting of plant functional types in mallee-heath subject to chaining and burning, burning only and neither of these. The effects of chaining and burning did not differ from only burning at the community level. Importantly, however, we recorded 90% fewer recruits of serotinous, obligate seeders in chained and burnt compared with only burnt plots, and a 44% decrease in their species richness. By contrast, recruits of obligate seeding shrubs and fire-ephemeral herbs with persistent soil-stored seed banks increased by 166% in chained and burnt plots. Sprouters showed little difference. We conclude that chaining and burning is likely to significantly alter vegetation composition, and potentially poses a significant threat to serotinous, obligate seeders. These impacts require consideration in fire management planning.

Estimating the time since fire of long-unburnt Eucalyptus salubris (Myrtaceae) stands in the Great Western Woodlands

Establishing the time since fire in infrequently burnt, yet fire-prone, communities is a significant challenge. Until this can be resolved for >50-year timeframes, our capacity to understand important ecological processes, such as the periods required for development of habitat features, will remain limited. We characterised the relationship between observable tree growth rings, plant age and plant size in Eucalyptus salubris F.Muell. in the globally significant Great Western Woodlands in south-western Australia. In the context of recent concerns regarding high woodland fire occurrence, we then used this approach to estimate the age of long-unburnt E. salubris stands, and the age-class distribution of Eucalyptus woodlands across the region. Time since fire was strongly predicted by trunk growth rings and plant size predicted growth rings with reasonable accuracy. The best model estimating growth rings contained parameters for trunk diameter, plant height and plot location, although simple models including either trunk diameter or plant height were nearly as good. Using growth ring–size relationships to date long-unburnt stands represents a significant advance over the current approach based on satellite imagery, which substantially truncates post-fire age. However, there was significant uncertainty over the best model form for estimating the time since fire of stands last burnt over 200 years ago. The management implications of predicted age-class distributions were highly dependent on both the choice of what, if any, transformation was applied to growth rings, and the theoretical age-class distribution to which the actual age-class distribution was compared.

Application and validation of visual fuel hazard assessments in dry Mediterranean-climate woodlands

Understanding fire behaviour and vegetation flammability is important for predicting the consequences of fires. Visual assessments of fuel, such as those developed in Project Vesta, have been widely applied to facilitate rapid data acquisition to support fire behaviour models. However, the accuracy and potential wider application to other plant communities of Vesta visual fuel assessments has received limited attention. We conducted visual fuel assessments and detailed quantitative structural measurements in Eucalyptus salubris (gimlet) woodlands in the world’s largest extant Mediterranean-climate woodland. With one exception, there was moderate to strong correlation between visual assessments of cover in vegetation layers and quantitative measurements, indicating that visual assessments adequately capture changes in fuels. This suggests that the Vesta visual fuel assessment methodology may have wide application in Australian eucalypt forests and woodlands and perhaps in similar communities around the world. However, several issues limiting the wider application of Vesta visual fuel assessments were identified, mainly associated with differences in community ecology between non-resprouter dominated E. salubris woodlands and the epicormic resprouter-dominated dry forests where the method was developed. Patterns of change in fuels suggest that flammability in E. salubris woodlands peaks at intermediate times since fire, potentially providing opportunities for fire management interventions.

Estimating fire interval bounds using vital attributes: implications of uncertainty and among‐population variability

Identifying the range of appropriate fire return intervals is crucial for ecosystem management in fire-prone environments. Plant vital attributes and changes in their associated trait values with time since fire are important indicators of suitable fire interval bounds to conserve biodiversity. However, using vital attributes to derive prescriptions for acceptable fire intervals remains challenging due to (1) uncertainty regarding how traits are best measured, (2) uncertainty in the acceptable ranges of trait values to avoid local extinctions, and (3) potential for variability among populations in the time taken postfire to reach trait threshold values. Using a time-since-fire gradient in contrasting mallee and mallee-heath vegetation types of southwestern Australia, we calculate, compare, and aggregate fire interval bound predictions from nine serotinous non-resprouters and seven serotinous resprouters across these three sources of uncertainty or variation. Relationships between time since fire and both trait measures reflecting minimum fire interval (mean number of closed fruit per plant or proportion of plants with closed fruit) were typically significant, had reasonable goodness of fit, and showed similar patterns of change with time since fire. Significant relationships reflecting maximum fire interval were less frequent but were more commonly detected using direct measures of mortality than using evidence for decline in reproductive potential. Of the two sources of uncertainty, trait value threshold selection caused more substantial differences in estimated interval bounds than trait measurement method. Variation between populations increased with greater estimated minimum interval length and, in some species, rendered interval estimates of limited practical value. On balance, we conclude that measures of vital attribute traits offer a transparent approach for estimating fire interval bounds at the plant community level, but selection of trait value thresholds is in need of stronger biological justification in their application. Further, variation between populations should be explicitly sampled if fire interval estimates are to be applied across the landscape.

Fire does not facilitate invasion by alien annual grasses in an infertile Australian agricultural landscape

Plant invasions are a significant threat to fragmented native plant communities in many agricultural regions. Fire potentially facilitates invasions, but in landscapes historically subject to recurrent fires, exclusion of fire is also likely to result in loss of biodiversity. We investigated the relationship between fire, fragmentation and alien plant invasion in mallee communities of the Western Australian wheatbelt. We hypothesized that invasion is limited by lack of propagules and the low soil nutrient levels of this old, infertile landscape, but that fire and/or fragmentation disrupt these limits. We tested the effects of three factors on establishment and abundance of alien annuals: ± fire, ± post-fire seeding with the locally invasive Avena barbata (propagule availability) and three landscape contexts. The three landscape contexts, exploring site limitations, were reserve interiors, perimeter edges adjacent to agricultural land and internal reserve roadside edges. Our first hypothesis was supported: Avena establishment was consistently greater in seeded plots, but away from perimeter edges, growth was poor. Our second hypothesis was supported only for perimeter edges: neither fire nor fragmentation by interior roads enhanced invasive plant establishment or biomass. At perimeter edges, invasive plant biomass was significantly greater. This was associated with higher propagule availability and elevated soil nutrient levels but was not enhanced by fire. We conclude that fire is unlikely to promote invasion by alien annuals in low-nutrient ecosystems such as mallee, hence is a viable disturbance strategy for biodiversity conservation away from nutrient-enriched edges.

Chaining and Burning Modifies Vegetation Structure, Fuel, and Post-Disturbance Sprouting Capacity

Abstract Prescribed fire and/or mechanical methods can be used to modify the quantity, continuity, and/or spatial arrangement of flammable fuel. Yet the consequences of fuel management, both in terms of ecological outcomes and in facilitating improved fire management, often are poorly documented. In the global biodiversity hotspot of southwest Western Australia, chaining and burning is a novel technique for manipulating fuels. Vegetation first is dislodged using a chain, then after a period of curing, burnt. We tested whether combining two disturbance events in this way results in different vegetation structure postfire than only burning, and whether the postfire sprouting capacity of community-dominant Eucalyptus spp. is compromised. Both chained and burnt and only burnt treatments had much less leaf litter and vegetation > 25 cm high than long-unburnt vegetation, indicating a fire management benefit of fuel modification. Chained and burnt strips had a threefold reduction in standing dead vegetation compared to only burnt samples. The stem number of Eucalyptus spp. was reduced by 20% in chained and burnt strips compared to only burnt vegetation, indicating that consecutive disturbances reduce resilience and might render sprouters vulnerable to subsequent disturbances. Balancing the fire management benefits of chaining and burning with the ecological consequences is a significant challenge facing land managers in this fire-prone landscape.

Long-term studies of post-fire reproduction in an Australian shrubland and woodland

Identifying appropriate fire-return times is critical for management of temperate southern Australia’s fire-prone shrublands and woodlands. The time to first flowering and peak flowering are useful attributes for understanding how species and vegetation will respond to different fire intervals. Using a plant fire-response trait framework, we investigate patterns of reproduction in south-western Australian kwongan and woodland after prescribed fires with a 30-year longitudinal study, with the aim of identifying appropriate fire intervals. Modelling of post-fire flowering responses of species and aggregating responses into plant functional types showed substantial differences in the time after fire to first and peak flowering. Tests of hypotheses of flowering patterns after fire with different fire-response traits showed that (1) resprouters flower sooner than recruits of non-resprouters, (2) serotinous non-resprouters reach first and peak flowering later than do non-resprouters with soil-stored seed, (3) species in taller strata reach first and peak flowering later than those in lower strata and (4) geophytes flower earlier than other resprouters that lose their aboveground biomass in fire. The most fire interval-sensitive non-resprouting serotinous species take 15–20 years in kwongan to reach peak flowering and in Allocasuarina woodland 25–30 years, providing a working hypothesis for minimum fire intervals in the plant communities under investigation. Our study illustrates the value of long-term ecological studies for informing biodiversity management.