Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessArticle

Peer-Review Record

Shifting States, Altered Fates: Divergent Fuel Moisture Responses after High Frequency Wildfire in an Obligate Seeder Eucalypt Forest

Forests 2019, 10(5), 436; https://doi.org/10.3390/f10050436

by Jamie Burton^1,*

, Jane Cawson²

, Philip Noske¹ and Gary Sheridan¹

Reviewer 1: Anonymous

Reviewer 2: Anonymous

Reviewer 3:

David B. Wester

Forests 2019, 10(5), 436; https://doi.org/10.3390/f10050436

Submission received: 25 April 2019 / Revised: 8 May 2019 / Accepted: 17 May 2019 / Published: 20 May 2019

(This article belongs to the Section Forest Ecology and Management)

Round 1

Reviewer 1 Report

This paper is looking great, well done. I’ve only got minor comments, otherwise, I think it’s a very good contribution to the discussion.

Line 31 and other areas referring to feedbacks (852-853, 857) – as per my previous comments, you need to specify that you’re talking about feedbacks following ecosystem collapse. This is a much less studied area, so given that the main body of literature examines feedbacks leading to collapse, most readers will probably assume that’s what you’re saying.

Line 158 – “increases in fire frequency under climate change [have] been acknowledged”

Line 160 “…boreal forests and lodgepole pine Pinus contorta var. latifolia (Engelm. ex S. Watson.) [, and] in coniferous…”

Line 270 – probably list the source of classification (Specht?)

Line 474 – Do you have a source to show that 7% moisture is linked to erratic fire behaviour? I am aware of a published value of 9.9%¹, but if you want to use 7% perhaps be explicit about the reliability of the number. Fire literature suffers from too many magic numbers that come from nowhere, so if this is such a number that gets relied on but no one knows where it comes from, just make that clear.

Line 713-714 – “However, the opposite pattern was identified for 2009-Non-eucalypt, surface fuels [which] were wetter…”

Line 729 – The other cited studies establish a correlation, but it’s worth also referencing the mechanistic argument² to demonstrate causation.

Lines 743-744 – again, just qualify the 7% value unless you have evidence.

Line 878 – perhaps reference the model you’re referring to³

References

1. Nolan, R. H., Boer, M. M., Resco de Dios, V., Caccamo, G. & Bradstock, R. A. Large scale, dynamic transformations in fuel moisture drive wildfire activity across south-eastern Australia. Geophys. Res. Lett. 43, 4229–4238 (2016).

2. Matthews, S. A process-based model of fine fuel moisture. Int. J. Wildl. Fire 15, 155–168 (2006).

3. Zylstra, P. J. et al. Biophysical mechanistic modelling quantifies the effects of plant traits on fire severity: species, not surface fuel loads determine flame dimensions in eucalypt forests. PLoS One 11, e0160715 (2016).

Author Response

This paper is looking great, well done. I’ve only got minor comments, otherwise, I think it’s a very good contribution to the discussion.

· Sentences have been changed to highlight focus is on feedbacks after ecosystem collapse or state transition.

· Line 30 ‘This indicates there is potential for both positive and negative flammability feedbacks following state transition depending on the composition of the non-eucalypt state’

· Line 378 ‘Fire severity ….. may mediate the direction of flammability feedbacks following state transition’.

· Line 425 ‘This indicates there is potential for both positive and negative feedbacks following state transition which has implications for future fire behaviour and forest conversion under climate change’

Line 158 – “increases in fire frequency under climate change [have] been acknowledged”

· Changed.

Line 160 “…boreal forests and lodgepole pine Pinus contorta var. latifolia (Engelm. ex S. Watson.) [, and] in coniferous…”

· Changed to ‘In the Northern hemisphere, increases in fire frequency under climate change have been acknowledged as a key mechanism of decline for obligate seeder species such as black spruce Picea mariana ((Mill.) Britton.) and lodgepole pine Pinus contorta var. latifolia (Engelm. ex S. Watson.)’

Line 270 – probably list the source of classification (Specht?)

· Listed (Specht 1981) as reference for vegetation classification.

· This is a good point, we have adjusted the language to highlight that there is possibility for more severe fire behaviour as fuel moisture reaches this level. This threshold was established by (McArthur 1967) and was also discussed by (Tolhurst and Cheney 1999) but it is true that falling below this threshold does not guarantee erratic fire behaviour, as it is dependent on other drivers such as fire weather and ignitions.

Line 713-714 – “However, the opposite pattern was identified for 2009-Non-eucalypt, surface fuels [which] were wetter…”

· Changed.

Line 729 – The other cited studies establish a correlation, but it’s worth also referencing the mechanistic argument² to demonstrate causation.

· I believe this is referring to the paragraph in the discussion referring to mechanisms which affect fine fuel moisture and the role of the forest canopy. Matthews (2006) reference in following sentence: ‘Canopy interception of rainfall and wind speed at the forest floor are additional mechanisms which affect microclimate and fuel moisture (Van Wagner 1969; Viney 1991; Matthews 2006)’.

Lines 743-744 – again, just qualify the 7% value unless you have evidence.

· Changed to reflect possibility of more severe fire behaviour given other conditions (weather, wind speed).

Line 878 – perhaps reference the model you’re referring to³

· Reference added.

Reviewer references

2. Matthews, S. A process-based model of fine fuel moisture. Int. J. Wildl. Fire 15, 155–168 (2006).

References

Matthews, S (2006) A process-based model of fine fuel moisture. International Journal of Wildland Fire 15, 155.

McArthur, AG (1967) Fire behaviour in eucalypt forests. Department of National Development, Forestry and Timber Bureau, Canberra, ACT, Australia.

Specht, RL (1981) Foliage projective cover and standing biomass. In 'Vegetation classification in Australia.' (Eds AN Gillson, DJ Anderson.) pp. 10-21. (CSIRO, Australian Capital Territory, Australia: Canberra)

Tolhurst, K, Cheney, N (1999) Synopsis of the knowledge used in prescribed burning.

Van Wagner, CE, 1969. Combined effect of sun and wind on surface temperature of litter. Chalk River, Ontario. 1-8.

Viney, NR (1991) A review of fine fuel moisture modelling. International Journal of Wildland Fire 4, 215-234.

Reviewer 2 Report

The revised manuscript comprehensively addresses reviewer comments and I recommend it be accepted for publication. Congratulations to the authors.

Author Response

Thank you for your comments.

Reviewer 3 Report

I appreciate the authors’ thorough response to comments offered on the original version I reviewed. I believe that the revised ms provides valuable information that will add to the literature. I appreciate the opportunity to review this paper.

Author Response

Thankyou for your comments.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Well done to the authors on a very interesting study of an important issue. They have asked a critical question regarding a major form of disturbance, and their results have important implications.

Unfortunately, I think the authors have misunderstood a fundamental issue that undermines the value of their interpretations and conclusions. Their central conclusion is that their findings contradict earlier work, finding a negative feedback due to frequent fire when others had proposed a positive one. They argue this because some of the shrublands resulting from frequent fire were less available to burn than were the forests. This is indeed an important finding, but it does not address the issue that they claim.

The paper title refers to “changes in fuel moisture in an obligate seeder forest”, but these shrublands are no longer part of the obligate seeder forest. They are an alternative stable state (Scheffer et al. 2001) resulting from ecosystem collapse (Keith et al. 2013). With the exception of a speculative comment in one paper (Fairman et al. 2016), the previous studies did not claim that the ASS would be more flammable than the forest, rather their concern was that the forest itself would be more flammable in its recently burnt state than in a mature state. Their focus was about what happens prior to ecosystem collapse, because positive feedbacks are so heavily implicated in ASS theory (Wilson & Agnew 1992). Certainly, positive or negative feedbacks can also exist in the shift between stable states, but this was not the focus of those studies and therefore the findings of this current study do not contradict them. On the contrary, this in fact study supports the earlier work because it shows that old forests are less frequently available to burn than regrowing forests – exactly consistent with their findings.

A second important issue is that - as the authors mention in discussion, fuel moisture is only one aspect of flammability. It does work as a switch and if a site was always too wet to burn it would be safe to say that it is less flammable. However, the authors point out that their study was conducted in a wet period. The 1939 shrubland may indeed have dried out in other conditions, in which case other aspects of flammability may have become more influential. For this reason, measuring fuel moisture alone cannot be used to test the third objective: “Is there evidence of a positive feedback between high-frequency fire and flammability in Mountain Ash forest?”.

This touches on a larger issue regarding the weight of evidence. Positive feedbacks have been empirically proven in Ash forests (Taylor et al. 2014; Lindenmayer et al. 2011; Zylstra 2018). Analysis of a mechanism does not change that, but what it can do is help to explain it. This paper partially achieves that by showing that moisture dynamics as a forest ages are consistent with the empirical observations of flammability dynamics. This is even more interesting when considered along with similar studies that appear inconsistent. This paper for example found that 9yo forests were drier than old forest, but (Cawson et al. 2017) found that 7yo forests were wetter. Consider that Taylor et al found that Ash forests were least flammable for 7 years post-fire, but then rapidly increased in flammability – could the current paper have identified an explanatory mechanism for that by measuring either side of the division?

The findings re shrubland moisture dynamics are fascinating. Of the three groups, the young high intensity burnt Acacia shrubland dried faster than forest or other shrublands, so that transition to this form would increase the area of landscape available to burn and may be part of the wider positive feedback. This was not the case with the young mixed shrubland however, and the old shrubland was even wetter. Fire severity may be the deciding factor in this instance, and that will have very significant implications because factors leading to more intense fire will also lead to more of the dry Acacia shrubland and less of the wet mixed form. A second observation however is that – as touched on in the discussion – the old shrubland was composed of rainforest species and may be more accurately described simply as rainforest. Whatever the case – shrublands became wetter with age just as forests did. The authors may have therefore compiled evidence that:

1. Provides a mechanism consistent with the empirical observation of positive feedbacks

2. Identified that ASS may take multiple forms with entirely different moisture dynamics and probably dependent on fire severity, and

3. Identified a mechanism driving possible positive feedback in the ASS

Such findings would be very significant.

The only other comments I will add are that the study depends heavily on the calibration of the moisture stick measurements. More information on the accuracy of this would be very valuable – if estimates are skewed for a community for example, they could make all the difference in the findings. I suggest adding in mean error and MAE for each community. The equations on Fig. 3 also don’t look right – you may need to incorporate the logarithm back into them.

I apologise that my review is not more positive, but these are critical issues. I have withheld comments on minor issues as I believe the paper needs major changes to address what I have raised; when these are done, I will be happy to review any resubmission.

My best wishes to the authors, and I hope to see the next version of this paper before too long.

References

Cawson, J.G. et al., 2017. Fuel moisture in Mountain Ash forests with contrasting fire histories. Forest Ecology and Management, 400, pp.568–577.

Fairman, T.A., Nitschke, C.R. & Bennett, L.T., 2016. Too much, too soon? A review of the impacts of increasing wildfire frequency on tree demography and structure in temperate forests. International Journal of Wildland Fire, 25(8), pp.831–848. Available at: http://www.publish.csiro.au/WF/WF15010.

Keith, D.A. et al., 2013. Scientific foundations for an IUCN Red List of ecosystems. PLoS one, 8(5), p.e62111.

Lindenmayer, D.B. et al., 2011. Newly discovered landscape traps produce regime shifts in wet forests. Proceedings of the National Academy of Sciences, 108(38), pp.15887–15891.

Scheffer, M. et al., 2001. Catastrophic shifts in ecosystems. Nature, 413(6856), pp.591–6. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11595939.

Taylor, C., McCarthy, M.A. & Lindenmayer, D.B., 2014. Nonlinear effects of stand age on fire severity. Conservation Letters, 7(4), pp.355–370. Available at: http://doi.wiley.com/10.1111/conl.12122 [Accessed October 30, 2014].

Wilson, J.B. & Agnew, A.D.Q., 1992. Positive-feedback switches in plant communities. Advances in Ecological Research, 23, pp.263–336. Available at: https://www.sciencedirect.com/science/article/pii/S006525040860149X [Accessed March 6, 2019].

Zylstra, P., 2018. Flammability dynamics in the Australian Alps. Austral Ecology, 43(5), pp.578–591. Available at: https://onlinelibrary.wiley.com/doi/abs/10.1111/aec.12594.

Author Response

REVIEWER 1

Comments and Suggestions for Authors

Well done to the authors on a very interesting study of an important issue. They have asked a critical question regarding a major form of disturbance, and their results have important implications.

· We appreciate the positive feedback from the reviewer.

· Yes, we were not clear in what we meant by positive and negative feedbacks, in our study we were interested in the feedback arising from state changes at a stand-scale and whether change to non-eucalypt forest results in a vegetation state that is more available to burn. We have adjusted the introduction to better articulate the current understanding of state changes and where our study fits in. PG3 L108-114 ‘The interactive effects of multiple disturbances (wildfire, timber harvesting) on flammability have been the focus of recent research in Ash eucalypt forests (Lindenmayer et al. 2011; Attiwill et al. 2014; Taylor et al. 2014). Despite this, few studies have specifically considered the fuel moisture dynamics of alternative stable states, limiting our understanding of feedback mechanisms in wet sclerophyll forests. Our study aimed to better understand the fuel moisture dynamics of alternative stable states resulting from high-frequency stand-replacing wildfire’.

· Yes that is a valuable point and we have altered the title and research questions to better suit the concepts the paper addresses.

· Yes it is true that the results from our study support the hypothesis that longer unburnt ash forests are less available to burn than regenerating ash forests, however this was not an objective of our study as we were interested in the implications to flammability after state change.

· Yes, previous studies were not concerned with the flammability dynamics after ecosystem collapse and perhaps we did not address this point clearly in the manuscript, we have changed the introduction to better convey the uncertainty around the implications of state-changes for flammability (see previous comment).

· Yes we acknowledge that fuel moisture is one aspect of flammability and collecting data from one fire season limits the interpretation of the results. We have reconsidered the third objective and instead asked ‘what are the implications for flammability’. We now discuss the implications for flammability and fire-vegetation feedbacks more broadly whilst acknowledging that the results in this study alone do not provide the whole picture.

· That is an interesting point, however whilst we do compare to other studies, we are cautious to extrapolate our results beyond our study sites. The study by Cawson et al. (2017) mentioned above had a similar study design to our study but measured different stands across different fire seasons. Thus we do not think this provides solid evidence of an age-flammability mechanism.

· Yes as we mention in the discussion, the two forms of regenerating non-eucalypt forest may be a function of fire intensity, however as we do not know how fire behaviour differed between sites we wouldn’t want to draw strong conclusions about this. We have included a line in the discussion exploring this idea. PG12 L375-379 ‘The two recently burnt non-eucalypt stands were qualitatively assessed to represent the spectrum of post-fire regeneration that was observed at the study area and were also assumed to differ in the severity of the last fire. Fire severity is an important factor that influences plant population survival and recruitment (Ashton and Martin 1996). Thus, fire severity may mediate the direction of flammability feedbacks (Barker and Price 2018)’.

1. Provides a mechanism consistent with the empirical observation of positive feedbacks

· Yes our findings support positive feedback in ash-type eucalypt forest, however this was not the main focus of our study.

2. Identified that ASS may take multiple forms with entirely different moisture dynamics and probably dependent on fire severity, and

· Yes this is a well-expressed point and we agree there may be a variety of alternative stable states depending on fire severity and other attributes of the fire regime such as fire interval.

3. Identified a mechanism driving possible positive feedback in the ASS

· Yes this is true and we did not articulate this clearly, we have included a discussion on the implications for flammability PG12 L371-374 ‘Our results suggest that non-eucalypt forest can take divergent forms with different fuel availability. This has implications for future fire behaviour and forest conversion as depending on the structure and composition of non-eucalypt forest, it may be more or less available to burn than the eucalypt forest it replaces’.

Such findings would be very significant.

· Yes this is a valid point, we have included the RMSE values for the log-log regression in a Table (Table 2) on PG9 L273. We have also incorporated the log-log axis into the regression figure on PG8. Pre-analysis of the data did not support the use of site-specific regressions, we performed an ANCOVA which suggested that the categorical variable (site) was not significant (p=0.05) for any fuel types. We have included the results of this analysis in the appendix in Table A1 (PG13 L445).

· Thank you for your constructive comments, we have endeavoured to address the critical issues you raise and believe this improves the quality of the manuscript.

My best wishes to the authors, and I hope to see the next version of this paper before too long.

· Thank you for your helpful and encouraging advice.

References

Ashton, DH, Martin, DG (1996) Regeneration in a pole-stage forest of Eucalyptus regnans subjected to different fire intensities in 1982. Australian Journal of Botany 44, 393-410.

Attiwill, PM, Ryan, MF, Burrows, N, Cheney, NP, Mccaw, L, Neyland, M, Read, S (2014) Timber harvesting does not increase fire risk and severity in wet eucalypt forests of southern Australia. Conservation Letters 7, 341-354.

Barker, JW, Price, OF (2018) Positive severity feedback between consecutive fires in dry eucalypt forests of southern Australia. Ecosphere 9, e02110.

Lindenmayer, DB, Hobbs, RJ, Likens, GE, Krebs, CJ, Banks, SC (2011) Newly discovered landscape traps produce regime shifts in wet forests. Proceedings of the National Academy of Sciences 108, 15887-15891.a

Taylor, C, Mccarthy, MA, Lindenmayer, DB (2014) Nonlinear effects of stand age on fire severity. Conservation Letters 7, 355-370.

Reviewer 2 Report

The authors studied differences in fuel moisture and fire-relevant microclimatic variables between long unburnt and recently burnt sites, and between forest and shrubland sites within these categories. They found that shrublands were no dryer, and presumably no more flammable, than forests of the same age. They conclude that transitions from forest to shrubland do not necessarily imply a positive fire feedback. The paper is clearly presented, well written and generally sound. Despite the considerable effort put into fieldwork, my concern is that the sample sizes are too small to allow robust conclusions to be drawn. While the authors openly acknowledge these caveats in the text, they could be incorporated more clearly into the high level summaries such as the abstract, conclusions and title. Some of the measures used could have tighter links with measures of fire risk or fire activity, to better explain how material the observed differences are.

Detailed comments

- The difference between high and low frequency fires is just the difference between one fire or two. It would be more accurate to talk about one vs two stand-replacing fires in the last 80 years, and even that does not account for time since fire

- Reduced rainfall is not a high confidence climate projection, especially compared to temperature.

- As the paper is not about prescribed burning, a fuel moisture threshold linked to wildfire risk or activity may be more appropriate

- the high fire days definition of days over 25 degC and days under 25% RH doesn’t appear to have any link with wildfire risk or activity, nor does it appear to represent the uppermost extremes of the distribution of local fire weather conditions

- P10 L330-331 there is no evidence for claim that fires are likely to occur on these days, I suggest rephrasing this

- When discussing severity could refer to Barker & Price 2018

Author Response

REVIEWER 2

Comments and Suggestions for Authors

· We agree that sample size is a limitation and should be acknowledged into high-level summaries. We have made the following changes to incorporate the caveats more clearly.

· Abstract PG1 L21-25, ‘A vegetation mosaic in the Central Highlands, Victoria created a unique opportunity to measure fuel moisture in adjacent forest stands that differed in overstorey species composition and time since fire. Specifically, we measured fuel moisture and microclimate at 2 eucalypt sites (9 and 79 years old) and 3 non-eucalypt sites (two 9 year old and one 79 year old).’.

· Introduction PG3 L114-116 ‘We measured fuel moisture in five forest stands that were within close proximity (<1km apart) but differed in overstorey species composition and age due to varying fire history’.

Detailed comments

· This is an interesting point and we agree it may be better to refer to the sites in terms of structure (the outcome of differences in fire frequency) and time since fire.

- Reduced rainfall is not a high confidence climate projection, especially compared to temperature.

· Reduced rainfall is not a just a projection for SE Australia, it is a multi-decadal observation. It is important to acknowledge this as since the turn of the 21^st century in Victoria, cool season (April to October) rainfall has averaged less than 15% less than the last century (Hope et al. 2017). This trend is predicted to persist and intensify in the longer term.

- As the paper is not about prescribed burning, a fuel moisture threshold linked to wildfire risk or activity may be more appropriate

· This is a great suggestion, we have included a fuel moisture threshold of 7% that is linked to erratic fire behaviour in eucalypt forests. Tolhurst and Cheney (1999) report a rapid increase in rate of spread due to more efficient pre-heating of fuels and McArthur (1967) noted an increase in possibility of crown fire development when fuel moisture content is less than 7%. We have included the fuel availability statistics for number of days fine fuel moisture content was less than 7% in Table 3 (PG10 L299).

· This is a good point and we have reformed the analysis to include a fuel availability statistic that is better associated with erratic fire behaviour and increased fire risk as above.

- P10 L330-331 there is no evidence for claim that fires are likely to occur on these days, I suggest rephrasing this

· We have rephrased the sentence from ‘On days when fire was more likely to occur surface fuel moisture was similar between recently burnt Acacia forest and recently burnt E.regnans whereas the non-eucalypt sites with a mixed overstorey were consistently wetter (>10% moisture content) than E. regnans forest at both recently burnt and long-unburnt sites’ to Severe heatwave conditions such as consecutive days above 35℃ in addition to long-term low rainfall patterns can contribute to lower dead fuel moisture and increase the spatial connectivity of fuels across the landscape, drying out the entire litter bed and increasing the potential for large fires (Cruz et al. 2012; Nolan et al. 2016; Nyman et al. 2018). In this respect, the effect of different vegetation states on fuel moisture becomes less important as fuels fall below threshold moisture levels. It may be useful to extend the study over more than one fire season to capture temporal variability in landscape dryness’.

- When discussing severity could refer to Barker & Price 2018

· Thank you for suggesting this reference, we have considered the findings of their work and included when discussing the effect of varied fire severity of regeneration response. See PG12 L378-379, ‘Fire severity is an important factor that influences plant population survival and recruitment (Ashton and Martin 1996). Thus, fire severity may mediate the direction of flammability feedbacks (Barker and Price 2018)’.

References

Ashton, DH, Martin, DG (1996) Regeneration in a pole-stage forest of Eucalyptus regnans subjected to different fire intensities in 1982. Australian Journal of Botany 44, 393-410.

Barker, JW, Price, OF (2018) Positive severity feedback between consecutive fires in dry eucalypt forests of southern Australia. Ecosphere 9, e02110.

Cruz, MG, Sullivan, AL, Gould, JS, Sims, NC, Bannister, AJ, Hollis, JJ, Hurley, RJ (2012) Anatomy of a catastrophic wildfire: the Black Saturday Kilmore East fire in Victoria, Australia. Forest Ecology and Management 284, 269-285.

Hope, P, Timbal, B, Hendon, H, Ekström, M, Potter, N (2017) A synthesis of findings from the Victorian Climate Initiative (VicCI). Bureau of Meteorology, Australia.

McArthur, A (1967) Fire behaviour in eucalypt forests. Commonwealth of Australia Forestry and Timber Bureau, Canberra, ACT, Australia.

Nolan, RH, Boer, MM, Resco De Dios, V, Caccamo, G, Bradstock, RA (2016) Large-scale, dynamic transformations in fuel moisture drive wildfire activity across southeastern Australia. Geophysical Research Letters 43, 4229-4238.

Nyman, P, Baillie, CC, Duff, TJ, Sheridan, GJ (2018) Eco-hydrological controls on microclimate and surface fuel evaporation in complex terrain. Agricultural and Forest Meteorology 252, 49-61.

Tolhurst, K, Cheney, N (1999) Synopsis of the knowledge used in prescribed burning.

Reviewer 3 Report

Review comments on Manuscript ID: forests-46706

General comment: This is a well-written paper that presents some useful data. I appreciate the authors’ attention to detail in the preparation of this paper. The Introduction does a good job of explaining the “big picture” ideas that motivated this research, with appropriate citations, and the Discussion stays within the limitations of the data set.

Specific comments:

The word “data” is plural and thus corresponding verbs should be plural (e.g., lines 174, 218, 220).

Line 24: change “indicate flammability” to “estimate flammability”.

Throughout the ms, it seems a space is needed in “E.regnans” as in “E. regnans”. Likewise for “A.dealbata” and other latin binomial abbreviations. Also for “2009Acacia forest” and other site designations—for these, I would suggest “2009-Acacia forest”.

Lines 130-130: change “It was estimated approximately 15,394 hectares…was burnt…” to “An estimated 15,394 hectares…were burnt…”

Lines 323, 325: change “86 MJ/m2” to “86 MJ/m²” or “86 MJ m^-2”

A minor point, but I think that using the word because in a sentence such as line 387 is much stronger than the word as. You write: “…does not support the hypothesis…as the non-eucalypt sites were...”. II would like to suggest: “…does not support the hypothesis…because the non-eucalypt sites were...”.

Line 395: change “that have report” to “that have reported”.

Literature citation format: use of capitalization is not consistent. E.g., for most citations, the first word (only) is capitalized, but not for citation nos. 28, 42, 54, 56. Also, italicize latin binomials in citations 38 and 61.

Regarding experimental design:

The authors have done a good job in explaining the limitations of their data set regarding inferences beyond their study sites (lines 408ff)—this caveat is necessary and I am pleased to see it acknowledged.

I would suggest, however, that this point be brought up earlier in the ms (in the Methods section)—and in a way that makes it more clear just how data were collected. In particular, my understanding is that there are 5 “treatments” = “site types” (time since fire and forest type) being compared in this study, with one stand per site type (line 408); it is also clear that each site type is represented by multiple data points (e.g., Fig. 3, the CIs in Table 2 and the standard deviations in Table 3): how many of these “samples” (for lack of a better term) per site type needs to be provided, as well has how they were selected and what exactly they represent. From the perspective of Hurlbert’s classic 1984 Ecological Monograph, this study suffers from pseudoreplication, and my experience is that many reviewers would reject the paper outright because of this. I am not one of those reviewers. At the same time, however, I would like to see the authors place this limitation in a more direct (and positive) light and (if I may) in a light that can be justified (see below). In short, I would like the authors to explain their experimental design more completely (along the lines that I’ve tried to explain in the comment—just how many “samples” per site type, and how were they selected?) and then state something like: “Because of lack of true replication of site types [or whatever term best describes the “time-since-fire-fuel-type” conditions you are comparing], our statistical comparisons are justified but our inferences are limited to our particular study sites (Wester 1992).” (Wester, D.B. 1992. Viewpoint: Replication, randomization and statistics in range research. J. Range Manage. 45:285-290).

Lines 209ff: regarding data analysis;

About the log-transformation of the data—I think this is fine but your description is not complete: were the data analyzed on a linear-log scale? A log-linear scale? Or a log-log scale? That is, depending on the data, one might log either one or both axes, and it is not clear what was done in this ms.

Fig 3: this comment is based solely on a visual inspection of these graphs, but I wonder if perhaps the regressions should have been done for each site separately. E.g., for Fig. 3c, if appears that a linear-linear scale might be appropriate for 2009-Acacia forests and 2009-non-eucalypt forests, whereas for the other sites, a log-transformation would be helpful. In short, it appears that the relationship between fine fuel moisture content and fuel stick moisture content may depend on site type—I would strongly encourage you to look at these data more carefully (I realize that this will reduce the sample size for the regressions).

Fig. 4: I would like to suggest that your inferences and conclusions can be strengthened by testing the null hypothesis that the slope in these regressions equals “1”—this is certainly how you are interpreting these results, and I think it is very appropriate to do so; I would just like you to formalize the process with a test of hypothesis. Thus, in Fig. 4a, if Ho: β₁ = 1 is not rejected in favor of Ho: β₁ ≠ 1, you can formally conclude a 1:1 relationship (e.g., one site is neither wetter not drier (P = whatever as in 0.282) than the other); and for Figs. 4, b and c, instead of writing “…with most points falling below the 1:1 line” (e.g., line 261), you can provide an alpha-supported conclusion if the slope actually turns out to be less than “1” (one site type is (P = whatever as in 0.003) wetter than the other). For these regressions, I would also strongly suggest looking into (probable?) heteroscedasticity—all I have to go on is a visual assessment of your graph, but it certainly looks as if variability around the line increases at higher moisture contents—so, perhaps the regression as well as the tests of hypothesis that I suggest should be based on heteroscedastic-consistent standard errors (e.g., Long, J.S. and L.H. Ervin. 2000. Using heteroscedasticity consistent standard errors in the linear regression model. The American Statistician 54(3):217-224. White, H. 1980. A heteroskedastic-consistent covariance matrix estimator and a direct test of heteroscedasticity. Econometrica 48:817-838). I think that this would considerably strengthen your conclusions.

Likewise for Tables 2 and 3: with the caveat of the pseudoreplication but with a justification that your statistical comparisons are valid but limited to your sites, why not do a formal F test (or, given that the data are discrete rather than continuous, a corresponding nonparametric test) comparing fuel availability (Table 3)—I realize that the data seem straightforward and that such a comparison likely will not change your conclusions, still, it will support your conclusions in a firmer way—so that, when you write (e.g., “…relative humidity was similar across sites” (line 315), you can back this up with a P-value. Similarly with other statements in the Discussion.

I appreciate the opportunity to review this paper. Signed: David Wester

Comments for author File: Comments.pdf

Author Response

REVIEWER 3

Comments and Suggestions for Authors

Specific comments:

The word “data” is plural and thus corresponding verbs should be plural (e.g., lines 174, 218, 220).

· The following changes have been made: PG6 L177 ‘Data were collected over a four month period (6^th December 2017 to 31^st March 2018) which incorporated one fire season’. PG7 L228 The microclimate data were summarised for the entire study period by calculating the mean daily maximum temperature and mean daily minimum relative humidity’. PG7 L231 ‘The fuel moisture stick data were summarised by first calculating the mean hourly stick moisture content (n=3) then computing the daily minimum stick moisture content’.

Line 24: change “indicate flammability” to “estimate flammability”.

· PG1 L26 has been changed from “indicate” to “estimate”.

Throughout the ms, it seems a space is needed in “E.regnans” as in “E. regnans”. Likewise for “A.dealbata” and other latin binomial abbreviations. Also for “2009Acaciaforest” and other site designations—for these, I would suggest “2009-Acacia forest”.

· We have changed the Latin binomial as suggested throughout the manuscript. We have altered the site terminology to 2009-Acacia forest, 2009-Non-eucalypt forest, 1939-Non-eucalypt forest, 2009-Eucalypt forest and 1939-Eucalypt forest.

Lines 130-130: change “It was estimated approximately 15,394 hectares…was burnt…” to “An estimated 15,394 hectares…were burnt…”

· PG4 L132 has been changed to ‘An estimated 15,394 hectares in the Neerim forest district (which includes the Bunyip River catchment) were burnt two to three times within 13 years and did not regenerate with Eucalypts owing to the lack of mature seed-bearing trees’

Lines 323, 325: change “86 MJ/m2” to “86 MJ/m²” or “86 MJ m^-2”

· This has been changed in Table 4 on PG10 and removed from discussion.

· The discussion no longer contains this sentence.

Line 395: change “that have report” to “that have reported”.

· The sentence on P12 L391 has been changed to ‘Indeed, this is consistent with other studies that have reported decreasing flammability with time since disturbance’.

· The citation format has been revised and made consistent.

Regarding experimental design:

· Further caveats have been added in response to reviewer 2.

· This is a significant point, there were five sites, we compared eucalypt and non-eucalypt sites of the same age (or time since fire). We collected fuel moisture samples on five (2009 sites) and six (1939 sites) occasions. This is represented in regressions in Figure 3, for each site there are multiple data points which denote the different sampling campaigns. In Table 4 (microclimate), the average was calculated from 116 days of microclimatic values at a site.

· We have altered the methods section to better articulate experimental design, data collection and what ‘replicates’ represent. PG7 L228 ‘The microclimate data were first summarised by calculating daily maximum temperature and daily minimum relative humidity at each site, we then calculated the mean for these microclimatic parameters for the study duration at each site (n=116 days)’. In the caption of Figure 3 PG8, we have included the following explanation ‘Points are from different sampling campaigns of fine fuel collection (n=5 for 2009 sites) and n=6 for 1939 sites).’

· We have included the following sentence PG8 L261 ‘Furthermore, because of the lack of true replication of sites, our statistical comparisons are justified but our inferences are limited to our particular study sites (Wester 1992). We interpret the results in this context’.

Lines 209ff: regarding data analysis;

· We have made the following changes to the data processing and analysis section to make it clearer that data were analysed on log-log scale. PG7 L217 ‘Prior to the analysis of the relationship between fuel moisture stick and gravimetric fuel moisture content, data were log₁₀ transformed to meet normality assumptions and decrease the influence of high moisture values (FMC>150%), which were highly variable, and less relevant to the FMC value range that is significant for fire’.

· This is a valid point, also raised by reviewer 1. Our pre-analysis of the data did not support the use of site-specific regressions, we performed an ANCOVA which suggested that the categorical variable (site) was not significant (p=0.05) for any fuel types. We have included the results of this analysis in the appendix in Table A1 (PG13 L445) to support our decision.

· We greatly value this suggestion and have incorporated a statistical test into the analysis. PG8 L242 ‘We used the slope.test function in the R ‘smatr’ package to test the null hypothesis that the slope of the regression line equals 1, indicating that surface fuel moisture between sites are similar. Regression lines were fitted using robust linear regression at the full range of moisture values and at fuel moisture contents less than 25% for 2009 site pairs’.

· As per your suggestion, we used robust linear regression to account for heteroscedasticity and displayed the results (slope value and p-value) in Figure 4 on PG9.

· We have modified the results section to include this analysis. PG9 L275 ‘Figure 4 compares the daily minimum surface fine fuel moisture content of different site pairs, 2009-Acacia forest – 2009-Eucalypt forest (Figure 4a), 2009-Non-eucalypt forest – 2009-Eucalypt forest (Figure 4b) and 1939-Non-eucalypt forest– 1939-Eucalypt forest (Figure 4c). 2009-Acacia forest was drier (p<0.001) than 2009-Eucalypt forest for the full range of moisture values but not for moisture contents below 25% (p=0.42) (Figure 4a). 2009-Non-eucalypt forest was wetter than 2009-Eucalypt for the full data range (p<0.001) and for moisture contents below 25% (p=0.003) (Figure 4b). 1939-Non-eucalypt was wetter (p<0.001) than 1939-Eucalypt site, on a given day surface fuel moisture content was 1.5 to 2 times higher in the 1939-Non-eucalypt forest (Figure 4c)’.

· We do not believe this analysis is suitable as this would mean using time as a replicate and would be pseudoreplication. A similar study (Cawson et al. 2017) did not include formal test of microclimatic or fuel availability.

I appreciate the opportunity to review this paper. Signed: David Wester

· Thank you for your comments, we appreciate the advice on improving the analysis of the results and language formatting.

References

Cawson, JG, Duff, TJ, Tolhurst, KG, Baillie, CC, Penman, TD (2017) Fuel moisture in mountain ash forests with contrasting fire histories. Forest Ecology and Management 400, 568-577.

Wester, DB (1992) Viewpoint: Replication, Randomization, and Statistics in Range Research. Journal of Range Management 45, 285-290.

Article Menu

Shifting States, Altered Fates: Divergent Fuel Moisture Responses after High Frequency Wildfire in an Obligate Seeder Eucalypt Forest

Further Information

Guidelines

MDPI Initiatives

Follow MDPI