Fire Impacts and Dynamics of Seasonally Dry Tropical Forest of East Java, Indonesia
Round 1
Reviewer 1 Report
Seasonally Dry Tropical Forests of East Java This is the first thorough paper aware to me which rigorously describes an example of the lowland tropical deciduous forest which occurs from East Java east to the Lesser Sunda islands and its dynamic relationship with putative savanna. (The authors report 'relatively high tree cover, but this is generally regarded as a factor which distinguishes forest from savanna).The text is well written, lucid, and overall well referenced. The description and discussion of the dynamic relationships between the stands sampled with differing burn histories are exemplary. It therefore serves as a template for others, and merits some further elaboration. That said, I have the following comments. Although there is a comprehensive discussion of the climatic and anthropogenic impacts, particularly use of fire, on the dynamic relationships of deciduous tropical forests in Africa and the Neotropics, there is no review of terminology for tropical deciduous forest types, nor definition for the term savanna in the context here. Further, if the several references to papers on the invasive exotic Acacia nilotica are included, very little is mentioned for this forest formation in Asia (e.g. 18 and 19 for Lesser Sunda, 26, 51, 52 elsewhere. The forest under study is therefore inadequately put in regional context. Champion's classification for South Asia ('British India'), (A preliminary Survey of the Forest Types of India. (Indian Forest Records, New Series, 1, 1936, updated and condensed in Ashton, On The Forests of Tropical Asia: lest the memory fade, R.B.G. Kew & Arnold Arboretum, 2014, revised and updated in Ashton & Lee, Trees & Forests of Tropical Asia: Exploring Tapovan, 2022, U. Chicago Press) is still the most widely accepted east to Burma and Thailand. In it, the forest in question is termed dry deciduous forest, currently modified to short deciduous forest, and characterizes single wet monsoon climates with at least eight dry months as the authors report. (A fuller description of climate, perhaps summarized in a klimadiagram, would be in order). In important contrast to the more completely referenced Neotropical forests of their type, anthropogenic influence of fire in tropical Asia goes back 1.8 mya (V. Ling to R. Rangham in Ashton 2014), while Java Man would doubtless used such forest for hunting during his time in the region in the Pleistocene. These forests therefore bear the longterm stamp of human influence. The plant geography of the flora therefore has some relevance. The tree flora appears to mostly be that of similar forests in India and Indo-China, notwithstanding the likelihood that equatorial Indonesia perhaps always offered a dispersal barrier. Exceptions include, besides invasives, Flacourtia rukam and Pterocarpus indicus, both of which are evergreen and widespread in seasonal evergreen forests in the region though apparently deciduous here, but neither is indigenous to the equatorial dipterocarp forests, so may have been introduced in classical times (like the several Indian species illustrated on the Borobudor bas reliefs.
Domestic cattle are a major factor influencing tropical deciduous forest composition and dynamics through grazing and browsing in South Asia, the presence of catle is only red briefly referred, nor any spatial relationships between fire-influenced stands and habitations.
Author Response
COMMENT: Although there is a comprehensive discussion of the climatic and anthropogenic impacts, particularly use of fire, on the dynamic relationships of deciduous tropical forests in Africa and the Neotropics, there is no review of terminology for tropical deciduous forest types, nor definition for the term savanna in the context here.
RESPONSE: In the first paragraph we add some extra detail on the terminology and defining characteristics of SDTF and why we have used this term. Overall we have changed the main focus of the paper to concentrate on SDTF – its characteristic species and its response to and recovery following fire.
COMMENT: Further, if the several references to papers on the invasive exotic Acacia nilotica are included, very little is mentioned for this forest formation in Asia (e.g. 18 and 19 for Lesser Sunda, 26, 51, 52 elsewhere. The forest under study is therefore inadequately put in regional context. Champion's classification for South Asia ('British India'), (A preliminary Survey of the Forest Types of India. (Indian Forest Records, New Series, 1, 1936, updated and condensed in Ashton, On The Forests of Tropical Asia: lest the memory fade, R.B.G. Kew & Arnold Arboretum, 2014, revised and updated in Ashton & Lee, Trees & Forests of Tropical Asia: Exploring Tapovan, 2022, U. Chicago Press) is still the most widely accepted east to Burma and Thailand. In it, the forest in question is termed dry deciduous forest, currently modified to short deciduous forest, and characterizes single wet monsoon climates with at least eight dry months as the authors report. (A fuller description of climate, perhaps summarized in a klimadiagram, would be in order). In important contrast to the more completely referenced Neotropical forests of their type, anthropogenic influence of fire in tropical Asia goes back 1.8 mya (V. Ling to R. Rangham in Ashton 2014), while Java Man would doubtless used such forest for hunting during his time in the region in the Pleistocene. These forests therefore bear the longterm stamp of human influence. The plant geography of the flora therefore has some relevance. The tree flora appears to mostly be that of similar forests in India and Indo-China, notwithstanding the likelihood that equatorial Indonesia perhaps always offered a dispersal barrier. Exceptions include, besides invasives, Flacourtia rukam and Pterocarpus indicus, both of which are evergreen and widespread in seasonal evergreen forests in the region though apparently deciduous here, but neither is indigenous to the equatorial dipterocarp forests, so may have been introduced in classical times (like the several Indian species illustrated on the Borobudor bas reliefs.
RESONSE We have also included more detail and context on SDTF in Asia using the recommended references (thank you!) and others we have found. Overall we have tried to improve regional context for the study in the first section of the Discussion.
COMMENT: Domestic cattle are a major factor influencing tropical deciduous forest composition and dynamics through grazing and browsing in South Asia, the presence of catle is only red briefly referred, nor any spatial relationships between fire-influenced stands and habitations.
RESPONSE: we have recognised the importance of grazing by cattle and other herbivores and added a recommendation (in the discussion) to study fire-herbivore interactions.
Reviewer 2 Report
The authors sought to assess a seasonally dry tropical forest in East Java, Indonesia, and its potential transition to a savanna due to burning. It is a simple but possibly useful study. Unfortunately, I do not think that it shows that the areas sampled are transitioning to savannas. If there were repeated fires over an extended period, and their plots were structurally and floristically different then I would say maybe they could show a transition. I think that this part of the study should be jettisoned (transition to a savanna). What the authors’ study does show, based on the structure of the areas sampled after different periods following burning, is that the SDFs are recovering. What the study is highlighting is as follows:
1) That the areas sampled were recovering and the 10-year plots were structurally, very similar to the long-burnt plots. The authors should try to determine if there are significant structural difference between the dhb classes for the different groups using Generalized linear model with a passion distribution. If they decide to use this, the method was outlined in and can easily be done in R: Lévesque, M., McLaren, K.P. and McDonald, M.A., 2011. Recovery and dynamics of a primary tropical dry forest in Jamaica, 10 years after human disturbance. Forest Ecology and Management, 262(5), pp.817-826.
2) However, although the two burn classes were similar structurally, floristically they were not similar, and this may be due to my point below.
3) As has been highlighted in previous literature, coppicing/resprouting is an important mechanism for recovery of SDTFs. It would be useful to indicate what proportion of the trees had resprouted versus those present as seedlings/saplings in each burn class if this is possible, if not at least for the most recently burnt class. Also, try to compare them using a statistical analysis. Additionally, a comparison or figures of the IVI for the species found in each fire age class will help to determine which species are favoured by burning. Although coppicing is advantageous, that is, species that can readily coppice can replace themselves in-situ more quickly, species that are susceptible to burning will be lost locally, and can only re-establish themselves after recruiting by seeds, which can be perilous (due to the harsh dry season which reduces the chances of successful recruitment). As a result, diversity takes a much longer time to rebound that structure, and this was illustrated by this study (10 years after burning is similar in diversity to the recently burnt than the long burnt).
4) Species that are potentially different among the burn classes. These species may not coppice/resprout readily (the ones found in the long burnt and not the other classes) or better able to handle burning by sprouting or producing seedlings readily. The results from the NDMS based on Bray-Curtis similarity and the SIMPER show this, and should be used to highlight these species.
5) Burning may promote the spread of invasives, as one invasive species was able to recover right after burning.
I think the focus of the paper should be changed. I do not think the authors have highlighted a transition from SDF to savannas. Their own assessments show that some plots from the different burn classes are mixed in with other classes when using the NDMS. This might indicate the effects of different burn severity, rate of recovery etc. If they can delve a bit deeper into SDTF recovery/regeneration/resilience literature and use this paper to support or elucidate some of the long-standing research needs for SDTFs, it would be suitable for publication. Most studies use successional forests at different stages of recovery to try to elucidate the changes in structure and diversity that occurs from conversion from pasture to SDTFs, following cutting and following cutting and burning, etc. This paper is very similar and does have some interesting findings (and the authors have already done most of the analyses, but focused on the wrong thing, i.e., transition to a savanna) and as such, the authors should use those papers as a guide when restructuring their papers. This is focus I feel the authors should pursue.
Author Response
COMMENT: The authors sought to assess a seasonally dry tropical forest in East Java, Indonesia, and its potential transition to a savanna due to burning. It is a simple but possibly useful study. Unfortunately, I do not think that it shows that the areas sampled are transitioning to savannas. If there were repeated fires over an extended period, and their plots were structurally and floristically different then I would say maybe they could show a transition. I think that this part of the study should be jettisoned (transition to a savanna). What the authors’ study does show, based on the structure of the areas sampled after different periods following burning, is that the SDFs are recovering.
RESPONSE: We agree and have changed our conclusions and overall lessened the coverage of savanna-forest transitions (although not jettisoned as outlined in our last point below).
What the study is highlighting is as follows:
COMMENT: That the areas sampled were recovering and the 10-year plots were structurally, very similar to the long-burnt plots. The authors should try to determine if there are significant structural difference between the dhb classes for the different groups using Generalized linear model with a passion distribution. If they decide to use this, the method was outlined in and can easily be done in R: Lévesque, M., McLaren, K.P. and McDonald, M.A., 2011. Recovery and dynamics of a primary tropical dry forest in Jamaica, 10 years after human disturbance. Forest Ecology and Management, 262(5), pp.817-826.
RESPONSE: We followed this advice and compared the structural (size class) distribution of the three fire-age groups. However, we did this using chi-square test of independence as the distribution was quite different for each fire-age group with none of them really showing a Poisson distribution with a logarithmic relationship with size class (as per Levesque et al. 2011). This test, the results of which have been added to the manuscript, showed that the overall distribution across the fire-age groups were similar (not significantly different from each other), although it is clear the recently burnt plots had far fewer seedlings and saplings.
COMMENT However, although the two burn classes were similar structurally, floristically they were not similar, and this may be due to my point below.
RESPONSE: We agree and have made this point more explicitly in the discussion and conclusions.
COMMENT: As has been highlighted in previous literature, coppicing/resprouting is an important mechanism for recovery of SDTFs. It would be useful to indicate what proportion of the trees had resprouted versus those present as seedlings/saplings in each burn class if this is possible, if not at least for the most recently burnt class. Also, try to compare them using a statistical analysis. Additionally, a comparison or figures of the IVI for the species found in each fire age class will help to determine which species are favoured by burning. Although coppicing is advantageous, that is, species that can readily coppice can replace themselves in-situ more quickly, species that are susceptible to burning will be lost locally, and can only re-establish themselves after recruiting by seeds, which can be perilous (due to the harsh dry season which reduces the chances of successful recruitment). As a result, diversity takes a much longer time to rebound that structure, and this was illustrated by this study (10 years after burning is similar in diversity to the recently burnt than the long burnt).
RESPONSE: We could not adequately assess resprouting in the 10yo and long unburnt plots as the fire was too long ago and post-fire recovery/regeneration difficult to assess and differentiate from inter-fire recruitment. For the most recently burnt class, we have added (to the text) the overall % of trees which resprouted as well as common species which were not observed to resprout. For species showing resprouting response, the % of resprouting trees for each species is shown in Fig. 7. Overall this demonstrates high rates of recovery via resprouting and this has been further emphasised and explored in the discussion.
COMMENT: Species that are potentially different among the burn classes. These species may not coppice/resprout readily (the ones found in the long burnt and not the other classes) or better able to handle burning by sprouting or producing seedlings readily. The results from the NDMS based on Bray-Curtis similarity and the SIMPER show this, and should be used to highlight these species.
RESPONSE: The main species driving the differences between fire age groups are listed in Table 2 and have been described in the results (section 3.2).
COMMENT: Burning may promote the spread of invasives, as one invasive species was able to recover right after burning.
RESPONSE: Yes we have expanded on the potential effect of fire to promote A. nilotica regeneration.
COMMENT: I think the focus of the paper should be changed. I do not think the authors have highlighted a transition from SDF to savannas. Their own assessments show that some plots from the different burn classes are mixed in with other classes when using the NDMS. This might indicate the effects of different burn severity, rate of recovery etc. If they can delve a bit deeper into SDTF recovery/regeneration/resilience literature and use this paper to support or elucidate some of the long-standing research needs for SDTFs, it would be suitable for publication. Most studies use successional forests at different stages of recovery to try to elucidate the changes in structure and diversity that occurs from conversion from pasture to SDTFs, following cutting and following cutting and burning, etc. This paper is very similar and does have some interesting findings (and the authors have already done most of the analyses, but focused on the wrong thing, i.e., transition to a savanna) and as such, the authors should use those papers as a guide when restructuring their papers. This is focus I feel the authors should pursue.
RESPONSE: rather than jettisoning the parts of the paper exploring the potential transitions to savanna, we have downgraded these in terms of the importance in the paper (the main focus now being on fire impacts and recovery of SDTF) and in terms of conclusions (we now explicitly accept the evidence for transition is not strong, but include a small section on how future studies can contribute more evidence to test such ideas). Overall the coverage of forest-savanna transitions is much reduced.
Round 2
Reviewer 2 Report
The authors made the suggested changes, however, the abstract needs to better reflect the conclusions. Also, the method used to sample the savanna should be included and I have a few questions that I would like the authors to think about. The abstract needs to be corrected before publication and the method modified but this change is very minor. I can therefore recommend their manuscript for publication.
This is what was stated in the conclusion:
"Overall results show Baluran SDTFs are likely to be resilient to single low-intensity burning as studied here. Although SDTF and savanna have potential to be alternative ecosystem states, this was not conclusively shown here. Further studies are needed to obtain more evidence of such ecosystem transitions, as well as the controlling factors."
But this is what was stated in the abstract:
"We evaluate some evidence that fire absence in SDTF may facilitate a transition to savanna as some long unburnt savanna had floristic similarities to dry forest (particularly in terms of characteristic 25 tree species), but recognize the need for more study to determine the degree and mechanisms of 26 forest-savanna transitions, with a future research agenda outlined."
Was this an error and did the authors mean to say that fire absence may facilitate a transition from a savanna? Also, there is no mention of the key findings that they found no evidence of of the SDTF and savanna being alternative states. Also, they should mention the role of the invasive A. nilotica in seemingly compounding/obfuscating a transition, resulting in a false indication of a transition, thus highlighting the importance of ground truthing. The ground truthing bit does not need to be added, but it would have been a criticism I would have had with the mapping bit. Thanks for including your observations.
Furthermore, consider including the method used to assess the savannas as I had to hunt down the dissertation to determine what was done. Just a succinct sentence or two on plot size, number and sample protocol.
Additionally, I note that the assessment of the savanna included only stems >= 10 cm. Is this correct? This might explain the lack of similarity between the SDTF and savannas, particularly the long unburnt ones. Food for thought and maybe something the authors should add as a draw back or quickly do this analysis: restrict the SDTF plots to only trees >=10 cm and see how different they are from the savannas..If they aren't then this should be added to the paper.
Finally, in the abstract use rainforests and not rainforest, or consider dropping it, because they didn't actually compare their plots to rainforest plots.
Author Response
RESPONSE TO REVIEWER 2 – 2nd ROUND OF COMMENTS Dec 2022
COMMENT 1: The authors made the suggested changes, however, the abstract needs to better reflect the conclusions. Also, the method used to sample the savanna should be included and I have a few questions that I would like the authors to think about. The abstract needs to be corrected before publication and the method modified but this change is very minor. I can therefore recommend their manuscript for publication.
RESPONSE 1: Thanks for your time and helpful suggestions to improve the ms. Yes, we agree the abstract needed further updating to reflect changes in the discussion and results. Other changes are outlined below.
COMMENT 2: This is what was stated in the conclusion:
"Overall results show Baluran SDTFs are likely to be resilient to single low-intensity burning as studied here. Although SDTF and savanna have potential to be alternative ecosystem states, this was not conclusively shown here. Further studies are needed to obtain more evidence of such ecosystem transitions, as well as the controlling factors."
But this is what was stated in the abstract:
"We evaluate some evidence that fire absence in SDTF may facilitate a transition to savanna as some long unburnt savanna had floristic similarities to dry forest (particularly in terms of characteristic 25 tree species), but recognize the need for more study to determine the degree and mechanisms of 26 forest-savanna transitions, with a future research agenda outlined."
Was this an error and did the authors mean to say that fire absence may facilitate a transition from a savanna? Also, there is no mention of the key findings that they found no evidence of of the SDTF and savanna being alternative states. Also, they should mention the role of the invasive A. nilotica in seemingly compounding/obfuscating a transition, resulting in a false indication of a transition, thus highlighting the importance of ground truthing. The ground truthing bit does not need to be added, but it would have been a criticism I would have had with the mapping bit. Thanks for including your observations.
REPONSE 2: Yes this was a wording error and we have changed these sentences to: “We did not find consistent evidence of ecosystem transitions between SDTF and savanna despite a small number of long unburnt savanna sites having floristic similarities to dry forest (particularly in terms of characteristic tree species), and we identify the need for more study to determine the degree and mechanisms of forest-savanna transitions in the region, with a future research agenda outlined. Relatively large areas of savanna-dry forest transitions demonstrated from remote sensing analyses were primarily attributed to spread of Acacia nilotica (an alien invasive small tree or shrub) into long unburnt savanna, and its decline in areas where the species is being successfully controlled via burning and cutting.”
COMMENT 3: Furthermore, consider including the method used to assess the savannas as I had to hunt down the dissertation to determine what was done. Just a succinct sentence or two on plot size, number and sample protocol.
RESPONSE: This information is under section 2.3 data analysis (lines 227-234): “To evaluate evidence of transitions between savanna and SDTF, we combined our SDTF plot data with other data collected in nearby savanna and other patches of SDTF as part of a larger study in BNP [42]. Altogether, we had data from 11 sites of relatively long unburned savanna (>ten years of fire absence), 11 sites of relatively unburned SDTF (also >ten years of fire absence), 12 recently burned SDTF (fire in last 1-2 years) and 10 of recently burned savanna which were burned at least twice in the last 6 years. All sites were consistently sampled with species cover and abundance being measured in plots of 50 x 50 m in size.”
COMMENT 4: Additionally, I note that the assessment of the savanna included only stems >= 10 cm. Is this correct? This might explain the lack of similarity between the SDTF and savannas, particularly the long unburnt ones. Food for thought and maybe something the authors should add as a draw back or quickly do this analysis: restrict the SDTF plots to only trees >=10 cm and see how different they are from the savannas..If they aren't then this should be added to the paper.
RESPONSE 4: It is true that only trees >10cm in diameter were measured in terms of stem diameter, but we did count the number of tree seedlings and any saplings <10cm in diameter and overall we estimated the cover of each species (combining cover of seedlings, saplings and trees in the plot). Therefore, there was consistency between the savanna and SDTF plots in terms of how cover was estimated (and it was cover which was used to do the PRIMER multivariate analyses (ordination and clustering)).
COMMENT 5: Finally, in the abstract use rainforests and not rainforest, or consider dropping it, because they didn't actually compare their plots to rainforest plots.
RESPONSE 5: We have used rainforests here. We have kept the reference to rainforests as it is important comparison in terms of research and conservation agenda globally.
