Faunal fire refuges

Published 04/03/16 | by Dr Steve Leonard

Large bushfires, such as those that occurred in Victoria in February 2009, are a feature of Australian forests. As well as threatening human life and property, they have profound impacts on native animals and plants. Yet even within these large fires, unburnt or less severely burnt patches remain. These patches have the potential to act as refuges for biota, acting as sites in which animals and plants may be able to survive fire, persist post-fire and ultimately re-colonise the landscape as it recovers. 

Dr Steve Leonard reports on the recently completed Faunal Refuges Project, a collaboration between researchers from La Trobe and Deakin Universities examining the occurrence and function of refuges within the area burnt during the Kilmore East-Murrindindi fires of February 2009. A major focus of the study was to shed light on the interactions of planned burning and bushfire. With bushfires predicted to become more frequent and severe as climate change takes effect, planned burning is increasingly seen as a means of reducing the negative impacts of bushfire on both human assets and ecological values. However, the ecology of planned burning remains poorly understood. 

The study was carried out in foothills (messmate-peppermint) forest within and adjacent to the bushfire-affected area. Field surveys were undertaken in 2010-12 (i.e. 1.5-3.5 years post-fire). 

What is a refuge? 

Refuges arise from natural processes, or by human manipulation of the environment. Natural refuges include patches of unburnt vegetation that missed being burnt by chance, or because they are less flammable (e.g. moist gullies). Other features, such as rock outcrops or unburned logs, may also serve as natural refuges for some species of plants and animals. Refuges may also be created by reducing fire fuel loads. Planned burning can reduce fuel such that a particular site is less likely to burn in a subsequent bushfire or it can be used strategically to prevent a bushfire spreading into adjacent unburnt habitat. 

Refuges can serve three main functions: a) assist organisms to survive during and immediately after a fire; b) facilitate the persistence of organisms and populations within the fire boundary; c) assist the re-establishment of populations within the burnt area as it recovers. 

Where were the refuges? 

Unburnt forest patches within the fire boundary were rare. Aerial photographs taken after the fire revealed only 85 unburnt patches greater than 1 ha in size (patch size ranged from 1 – 306 ha), together making up less than 1% of the overall fire area. These mostly consisted of rainforest or wet eucalypt forest along gullies. Unburnt patches were most likely to arise due to moister fuels in sheltered locations. 

However, a small number of unburnt patches occurred in dry eucalypt forest. These patches were more common where burning had occurred less than three years prior to the 2009 bushfires. 

Fire severity and fauna 

Variation in fire intensity created a mosaic of patches burnt to varying degrees. In most of the area burnt during extreme  conditions on 7th February, the tree canopy was incinerated or scorched. As conditions ameliorated later that day, the fire mainly burnt the forest understorey (ground burn). 

Fire severity refers to the degree of damage to vegetation caused by fire. Fire severity had a strong effect on the vegetation structure present 2-3 years post-fire. Crown-burn and crown-scorch sites had a dense layer of eucalypt seedlings and saplings that had germinated post-fire. Ground-burnt sites tended to have fewer shrubs than sites that remained unburnt. The response of the fauna to fire severity reflects a response to vegetation structure, rather than to fire severity per se. 

Birds. A total of 79 bird species was recorded during surveys. The number of species and the abundance of birds was lower at sites that were burnt more severely. The composition of bird communities also was influenced by fire severity. Some species, such as the Flame Robin that favours open habitat, were more common in severely burnt sites; while other species such as Eastern Spinebill, Eastern Yellow Robin and Silvereye, were more common in unburnt sites. 

Mammals. Six species of arboreal mammals were observed during surveys, all in low density. The overall abundance of arboreal mammals, and of the Greater Glider, were lower in severely burnt sites (crown scorch or crown burn), but did not differ between unburnt sites and ground-burnt sites. In severely burned forest, isolation was important. Arboreal mammals were more common with increasing amounts of unburnt or ground-burnt forest nearby (within 1 km radius). Terrestrial mammals overall did not show strong responses to fire severity at 2-3 years post-fire. However, the Bush Rat was more common in severely burnt sites than in unburnt sites, while the Agile Antechinus was more common in unburnt sites than ground-burnt sites.

Gullies as refuges for fauna and habitat 

Topography is a major source of variation in the landscape, with slope, aspect and landscape position influencing soils, vegetation and fire behaviour. Gullies are important habitat features: they often are more productive, and contain more complex vegetation and a higher abundance of hollow logs and trees than adjacent slopes. 

Severe fire is sometimes thought to have an homogenising effect on landscapes, reducing all sites to a common baseline. We hypothesised that with increasing fire severity, the differences in habitat attributes, and hence bird communities, between gullies and slopes would be diminished. This was not the case. Across all fire severity classes, gullies had higher species richness and abundance of birds, higher abundance of arboreal mammals and more large logs. These findings indicate that gullies retain their habitat value through a major fire and remain important sites for fauna despite being burnt. 

Planned burning and refuges 

Most of the 85 unburnt forest patches identified within the fire boundary were in gullies vegetated by wet eucalypt forest or rainforest. In general, time since fire (a surrogate for fuel load) had limited influence on whether patches burnt or not. However, in dry eucalypt forest, while gullies were still less likely to burn, recent planned burning decreased the chances of a site burning in the bushfire. This difference appears to be due to variation in fuel accumulation rates amongst forest types. In wet eucalypt forest and rainforest, fuels build up rapidly after fire, such that there is no window during which low fuel loads prevent fire. In dry eucalypt forest, fuel loads are low enough to inhibit fire spread for around four years post-fire. It is important to note that no unburnt patches remained in areas burnt under extreme weather conditions on 7th February 2009, irrespective of whether they had been recently burnt or not. 

Planned burning is likely to contribute to the retention of unburnt patches during bushfire in dry eucalypt forest only under moderate fire weather conditions. While fuel reduction can reduce the chance of a patch burning during bushfire, it also simplifies vegetation structure. This will diminish the quality of patches as refuges for species that depend on complex habitat structure. 

Patches that remain unburnt due to prior burning have simple vegetation structure. 

Both ‘young’ (burnt ≤ 3 years prior to 2009) and ‘old’ (burnt > 20 years prior to 2009) unburnt sites acted as refuges for birds. Old unburnt sites had the highest numbers of birds within the fire boundary; they had a greater abundance of birds than young unburnt sites, as well as all sites burnt by bushfire. 

Old unburnt sites also had a distinct assemblage of bird species that differed from that at burnt sites. 


  • Large, severe bushfires can have a profound effect on wildlife and ecosystems, but that effect is not uniform. Variability in fire severity contributes to landscape heterogeneity (patchiness) after fire. 
  • Unburnt patches of forest arose mainly due to the effects of topography and vegetation type on fuel moisture and were more common in gullies with rainforest and wet forest, than in drier foothill forests. In dry forests, recent (< 4 yr) prior burning contributed to the formation of unburnt patches under low intensity bushfire, but across the region these were fewer than those arising ‘naturally’. In extreme fire conditions, no unburnt patches remained. 
  • Surveys of the flora and fauna 2-3 years post-fire revealed rapid recovery, with most species present within the burned landscape, albeit frequently in lower abundance. 
  • Fire severity is a key influence on the post-fire status of plants and animals. Greatest impacts generally occurred at sites that were severely burnt (crown scorch, crown burn). 
  • Unburnt forest patches have an important role as refuges for fauna. They often supported a greater richness and abundance, and distinct composition, for faunal groups, compared with burnt sites. Old unburnt patches, in particular, may harbour a greater abundance of wildlife following severe bushfire. 
  • Increased levels of planned burning could result in more unburnt patches in dry forest exposed to bushfire, under moderate weather conditions. These have a simplified vegetation structure compared with long unburnt patches and so their value to wildlife may be limited. Planned burning potentially could be used strategically to protect long-unburnt stands, enhancing and preserving these valuable components of landscape heterogeneity. Given limited resources for fire management, planned ecological burning could target areas of relative topographic uniformity where ‘natural’ occurrence of unburnt patches is less likely to occur. 

Further information

Contact s.leonard@latrobe.edu.au or go to: www.latrobe.edu.au/ecology-environment-evolution/research/ specialisations/fire-ecology/habitat-refuges 

About the Author

Steve Leonard is a research fellow at LaTrobe University’s Department of Ecology, Environment and Evolution. Steve is intrigued by the complex relationships between disturbances such as fire and herbivory and ecosystem composition and function, and the application of this knowledge to conservation management. His research interests in the ecology of disturbance is not only in how disturbance influences ecosystems, but how vegetation and fauna may influence disturbance regimes.