Jack Bradshaw, 13 October 2022

Earlier this year a paper was published in an international journal that argued that if left long enough the southern forest of Western Australia was ‘unlikely to burn’. Furthermore that ‘the predicted likelihood of fire in undisturbed forest was 1 in 208 to 1 in 1149 years.’

In my opinion, the methodology used to obtain the results was so badly flawed as to make the conclusions meaningless. A letter to that effect was sent to the Editor of the journal, Environmental Research Letters, with the request that it be published as a Comment.

It has now been four months since the journal acknowledged receipt of the comment but with several reminders and no further response it is apparent that it will not be published. The only available option is therefore the unusual step of publishing these views in a newspaper. 

It is reproduced below.

Comment on
Self-thinning forest understoreys reduce wildfire risk, even in a warming climate

P. J. Zylstra, S. D. Bradshaw and D. B. Lindenmayer, Environmental Research Letters 2022 Vol. 17 Issue 4 Pages 044022, DOI:10.1088/1748-9326/ac5c10

In this paper the authors argue that the southern forest of Western Australia is:

• unlikely to burn for 5-7 years after a fire (‘young period’),
• very likely to burn from then to 56 years after a fire (‘regrowth period’)
• unlikely to burn after 56 years (‘mature period’).

My comments on this paper do not consider the fire behaviour issues put forward by the authors but rather the methodology used to obtain the results.

The authors base their assertions on an analysis of mapped wildfire records held by the Western Australian Department of Biodiversity, Conservation and Attractions for the period 1964 to 2018 from a disjointed area of forest ranging from the wandoo (Eucalyptus wandoo) and jarrah (E. marginata) forest east of Perth, parts of the karri (E. diversicolor) forest to the tingle (E. jacksonii, E. guilfoylei) forest at Walpole which they collectively describe as the Southern Forest of Western Australia. The authors measured the proportion of each of the three ‘periods’ above that was burnt in a given year which they have called the Likelihood of Ignition at a Point.

The likelihood of an area being burnt begins with the probability of ignition in the first place i.e. if there is no ignition, there will be no area burnt. In this study the authors take no account of the differences in the number of ignitions that occurred in each of the three ‘periods’ relative to their area or whether the number of ignitions was proportional to the area of each ‘period’.

All of the wildfires in south-west forests that occurred in the period studied would have been subjected to some level of suppression activity that would have impacted on the area burnt. In most cases suppression action would have reduced the area burnt compared to the area that would have burnt if the fire was left to run without suppression activity. In some cases, however, the area would have been greater than its natural extent if the area of the fire was deliberately extended by burning out to a ‘secure’ boundary. In this study no account at all was taken of the impact of suppression activity on the area of forest burnt.

Without a consideration of these two factors, both of which have a major impact on the area burnt, none of the conclusions that follow can have any validity.

The authors also predict that the likelihood of fire occurring in the ‘mature’ ‘period’ is 1 in every 323 years – varying from 1 in 208 years to 1 in 1149 years depending on ‘synoptic conditions’ defined by them as the variability in the number of high and low pressure systems each season. If this prediction held true, it would be expected that it would be reflected in the history of the virgin forests. Since regeneration of karri is dependent on fire, the occurrence of fire is reflected (at least in part) in the age of various cohorts in the virgin forest. 

The authors claim that fire in karri forests prior to 1850 were extremely rare on the basis of a lack of fire scars in a study by Rayner (1992) of six virgin karri stands. However, they ignore the occurrence of the fire-induced regeneration that Raynor found to occur every 36 years on average in these stands in the period before 1850.  They also ignore the possibility that there were an unknown number of additional fires that did not result in regeneration, nor do they consider the occurrence of mild-intensity fires that do not cause datable fire scars.

The authors also ignore or are unaware of the fire-induced regeneration events that were mapped in the virgin karri forest as it existed in 1996 (Bradshaw and Rayner 1997a, 1997b). This mapping shows that:

• the age of the oldest significant cohort of regeneration (>25% crown cover) in the virgin forest ranged from 40 to 370 years.
• 99.4% of the virgin forest was burnt with a regeneration-inducing fire event before it was 320 years old. i.e. virtually all of it was burnt well before the predicted 323 years
•  Younger significant cohorts occurred in 55% of the virgin forest i.e. at least 55% of the forest had a regeneration-inducing fire more than once during its lifetime. Virtually all of these events occurred prior to the 1960s in the period when prescribed burning in the karri forest was minimal prior to the introduction of aerial ignition.
• Most karri (94%) only lives to 250 years of age. i.e. a minimum fire frequency of 1 in 250 years would be required to regenerate and maintain the karri forest.

The predicted likelihood of fire in undisturbed forest of 1 in 208 to 1 in 1149 years is not supported by the on-ground evidence of the present virgin karri forest.

The serious flaws in methodology used in this study and the misrepresentation and ignoring of relevant data invalidates the conclusions that have been reached about differences in fire behaviour or the predicted fire frequency. This is even more so when no distinction is made between the forests as dissimilar as jarrah, wandoo, karri and tingle forest. As a result, this paper does not provide any valid support for their theories concerning differences in the flammability of the southern forests of Western Australia.

References

Bradshaw FJ, Rayner ME. 1997a. Age structure of the karri forest: 1. Defining and mapping structural development stages. Australian Forestry. 60(3):178-187.
Bradshaw FJ, Rayner ME. 1997b. Age structure of the karri forest: 2. Projections of future forest structure and implications for management. Australian Forestry. 60(3):188-195.

F. J. Bradshaw
8 June 2022

Related stories:

The Zylstra theory: a final comment: Roger Underwood;
Philip Zylstra’s response #4 – self-thinning forest understoreys and wildfire debate;
Jack Bradshaw to Philip Zylstra #2 – self-thinning forest understoreys and wildfire debate;
Philip Zylstra’s response #3 – self-thinning forest understoreys and wildfire risk debate;
Self-thinning forest understoreys and wildfire risk debate – Roger Underwood responds;
Peter Rutherford to Philip Zylstra #2 – self-thinning forest understoreys and wildfire debate;
Philip Zylstra’s fire research: Adding value or creating risk? : Peter Rutherford;
Philip Zylstra continues the debate – self-thinning forest understoreys and wildfire risk;
Self-thinning forest understoreys and wildfire risk debate – Jack Bradshaw responds to Philip Zylstra;
‘Self thinning forest understoreys reduce wildfire risk, even in a warming climate’: Philip Zylstra responds to Jack Bradshaw.