An examination of fire history and fire management in the context of fauna conservation in south-eastern Australia

Fire can influence biodiversity by changing the spatial patterns of vegetation and the suitability of habitat for fauna species. Understanding the spatial patterns of fire in the landscape is critical to the conservation of biodiversity, and this is reliant on sound knowledge of fire history.

In this thesis I examined fire history and fire management in the context of fauna conservation in New South Wales (NSW), south-eastern Australia. Initially, I used digital fire records for NSW for years 1902 to 2018 to identify sources of error in the database, then examined how data quality may affect interpretation of fire history for biodiversity conservation. I then examined variation among, and predictors of, distributions of time since the most recent fire (TSMRF) for extant native vegetation in NSW. I tested whether distributions of time since fire for different vegetation types follow predicted trends at varying scales and identified the degree to which these distributions are influenced by landscape management actions. Finally, I used a field-based study to investigate the impacts of fire history, vegetation type, and environmental factors on habitat attributes, and the extent to which they vary among vegetation types.

Thirty-two years and 76% of extant native vegetation in NSW (38% in areas managed for biodiversity conservation) had no fire records, and significantly more in the drier, less populated western parts of the state. However, my results indicate this is an incomplete representation of fire history. Repeated records accounted for >50% of the records in the database, however only 8% were identified by an automated clean using software tools. Without a manual-clean the number of fires and area burnt per year would be exaggerated by >300% in some years. The number of repeated records and records with missing data reduced dramatically from the early 2000s.

After cleaning, change point analyses indicate two periods of change in the patterns of recorded area burnt, the mid-1930s and late-1960s, although it remains unclear whether this is a change in fire regime or improved record keeping. Ninety-one percent of records were missing attributes important for biodiversity conservation (month of fire occurrence and intensity), severely limiting the usefulness of the database for biodiversity management, planning, and research.

In 2018, TSMRF varied from 0 to 92 years. TSMRF was skewed to shorter intervals in 2018, 2008 and 1998. In more coastal eastern areas, distributions were skewed to shorter times since fire but were similar among wet and dry vegetation types. In arid and semi-arid areas, distributions were greatly variable but skewed to longer TSF than more coastal areas and were often heavily influenced by one or two large fire events.

At a local scale, distributions of TSMRF varied within vegetation types that spanned large geographic areas. At a finer scale still, distributions were similar in locations in close proximity, regardless of vegetation type. Within areas managed for biodiversity conservation, distributions of TSMRF were skewed to shorter times since fire than in general for the vegetation type. Previous fire history, mean annual temperature and rainfall, and land-use were stronger predictors of the distribution of time since fire than proximity to roads and population density.

Although similar in structure and physiognomy, it was the floristic and fine-scale structural differences that were the greatest drivers of variation in habitat in the three vegetation types I examined. Characteristics of the most recent fire and the interval to the next fire did not have as strong an effect on habitat as vegetation type. The fire type (wild versus prescribed) was one of the best predictors for variation in the landscape for some habitat types. There was a weaker, but significant effect on fauna habitat of time unburnt prior to a fire and whether the fire is wild or prescribed. However, these effects of fire varied among vegetation types.

My findings demonstrate that key habitat types important for the persistence of fauna did not have a uniform response among vegetation types at a local scale. Fire management practices designed for managing plant species survival are based on predictable plant responses to fire at a formation-scale, however these results demonstrate that the response of key fauna habitat attributes is not predictable at this scale and fire management may need to be directed at a finer-scale than that at which they are currently managed.