Influence of Agricultural Expansion and Human Disturbance on the Encounter Rates of Nocturnal Mammals in Tropical Hill Forests in Bangladesh

Round 1

Reviewer 1 Report

Primary information on the status of tropical mammals rarely camera-trapped or live-trapped is really at a premium. Arboreal species are particularly likely to fit this description. This is a valuable body of information concerning what was seen when lamping in various Bangladesh localities. The authors deserves high credit for presenting the results as encounter rates, not as spurious population densities. Unfortunately the interpretation then abandons this caution and seems to assume (though this is not made explicit) that ERs are a reliable surrogate for abundance. They very rarely are, when habitats radically different in sighting opportunity are considered. Aggressive overhaul of what is being discussed, concluded and recommended is required to expunge the spurious and irresponsible. The remaining solid factual presentation of what was observed and where will be very useful. I have many specific comments on the annotated pdf file, here (I hope!) included.

Comments for author File: Comments.pdf

Author Response

Reviewer 1

We thank the reviewer for the many comments in the attached PDF. We integrated all of them so we hope there will be no more confusion with terms (especially abundance) in text. We add the following comments to the sections that we would like to expand.

Figure 1. If these scale bars are correct, none of these areas could in isolation support viable populations of most of the species bing surveyed. Essential context in thc MS is therefore (a) how long the current habitat layout has been broadly stable (if it is so, even yet) and what habitat is over the international border. And, for that matter in the Bangladesh part - I don't see any explanation of the drastic colour scheme of A and B, but hopefully/presumably it relates to habitat...

We used this map from a previous study, and we agree that the colour scheme is not relevant here and have replaced it with a similar map. One of the reasons we wanted to look for species in the more disturbed areas was to examine if these agricultural areas could supplement the more natural forest types as habitats for these species – in the hopes that working with local communities ultimately, they could produce viable populations. There is very limited natural forest left in these parts of Bangladesh, so we still feel it is important to show which species are persisting to help with our work in the future to restore aspects of habitat.

First sentence of the Conclusion. This is a very assertive statement and for SE Asia it is surely untrue; only a minority of species of "Small to medium sized nocturnal mammals" are grassland or otherwise non-forest. I doubt Bangladesh is much different. If this statement is retained, it needs to be given clear evidentiary backing. The effects of differing visibility in different habitats are profound and not acknowledged anywhere in this MS. They prevent any responsible inferences about abundance patterns based on ER patterns without careful consideration of effects on those ERs other than abundance. Lamping is generally more productive in encroached areas not because there are more animals there, but because the process of encroachment opens up visibility to see reflecting eyes.. Methods not dependent upon this 'chance' of detection, such as camera-trapping, do not typically show that for "Small to medium sized nocturnal mammals ... the forest types offered by protected areas frequently do not provide suitable habitat"; indeed quite the reverse. Among mainland SE Asian small carnivores this statement is applicable only to the various otters, to Small Asian Mongoose Urva javanica and to a lesser extent to Small Indian Civet Viverricula indica.

This differs from our experience, but we respect that each country is different in the remaining secondary forests that are parts of the protected area system and we have changed the tone of this sentence and hope it is less dramatic and more inclusive.

 

Reviewer 2 Report

Nocturnal mammals in tropical hill forests of Bangladesh. The paper is well written and covers much of the literature on the focal species and location. I have some suggestions that i think would clarify some points and provide more information to the readers. 

Line 14. Intensification of agriculture is not as much an issue as expansion of agriculture, as mentioned later in the introduction.

Figure 1. in panel B indicate the location of each of the 4 forests either with letters that match the subsequent panels or arrows.

Line 62. You are not estimating the abundance of nocturnal mammals, at best you are creating an index of relative abundance.

Lines 78-118. You provide details on each reserve. You do not provide a size for Rema-Kelenga. The measures for Adampur are in English units and should be in metrics as the others.  The fourth reserve – Adampur is listed in Figure 1 as Rajkandi.  Theses 4 reserves are relatively close to each others and seem to share the same temperature range and rainfall (maybe Adampur has less rainfall). This can all be consolidated. Also see no reason to report humidity.  You describe the natural vegetation in each reserve but not the vegetation differences between natural forest, gardens and plantations.

Section 2.2 – Survey methods. You do not report if you had any restrictions on weather – such as no surveys during rain.  It is not clear if you repeated transects or each was walked a single time.  You report the total km walked but not the distribution in each reserve.  If this was a camera trap survey you would need to track the vegetation density to account for detectability in the different forest types. Can you at least indicate the average visibility distance in each type? You never mention how you dealt with groups – were all observation of individuals or were groups counted as one observation?

Section 3 – results. I would like to see a comparison between the reserves. If the reserves are different in their mammal communities. you have measured several metrics – forest type along transects, human density, plantation crop. You test these metrics for the species, I would like to see if they differ between the reserves themselves.  One reserve is far larger than the others and may or may not retain additional species. Table 2 does not include site as a covariate, possibly due to sample size.

I would be interested in what range of HII was encountered in each reserve. It seems the values would be very low for the 4 reserves.  Maybe a short table that lists the range of metrics for each reserve. Including amount of each forest type along transects.

You drop jungle cat and leopard cat from the analysis due to small sample size. So they should be dropped from the table 1 and table 3.  It is enough to say they were encountered in x and y reserves, but at too low a number for consideration. Lines 240-242 are part of the same issue that you make conclusions about the 2 species based on little data.  They would have been detected more often with camera traps, as less sensitive to cameras than human presence.

Table 3. It should be pointed out that you are not aware of variance around these numbers so differences may not be significant.

Line 243. You mention camera traps and the vast majority of new surveys are using camera traps and not transects (diurnal or nocturnal). You advocate for more nocturnal transects (Lines 260-277). The real question is whether these species would have been detected with camera traps. Did you keep any records of the height of detection for your encountered. If the species were within 3 m of the ground they would have been detected with cameras. For species that are truly arboreal they would not have been detected without placing cameras in the canopy. The question is whether transect surveys (nocturnal or diurnal) are needed to supplement camera trap surveys or careful placement of camera traps would cover these species.  Two studies using canopy cameras are listed below.

Gregory, T., Carrasco Rueda, F., Deichmann, J., Kolowski, J. and Alonso, A., 2014. Arboreal camera trapping: taking a proven method to new heights. Methods in Ecology and Evolution, 5(5), pp.443-451.

Yihao, F., Guopeng, R., Ying, G., Shuxia, Z., Haohan, W., Yanpeng, L., Zhipang, H., Liangwei, C. and Wen, X., 2018. Impact of ground and canopy camera-trapping installation on wildlife monitoring. Biodiversity Science, 26(7), p.717.

 

Line 154-158. Maybe the 2 sentences should be joined with “, however” replacing “. Overall”

Table 1. add superscripts to indicate significant differences between the reserves for each species. 

Author Response

Editor

Please add some articles citation published in the past 3 years, and expand the introduction and conclusions sections when you make revisions?

We have added sections to the introduction and conclusions. We added new references in particular to the discussion but also updated several older references with newer ones on the same issues and topics.

Reviewer 1

We thank the reviewer for the many comments in the attached PDF. We integrated all of them so we hope there will be no more confusion with terms (especially abundance) in text. We add the following comments to the sections that we would like to expand.

Figure 1. If these scale bars are correct, none of these areas could in isolation support viable populations of most of the species bing surveyed. Essential context in thc MS is therefore (a) how long the current habitat layout has been broadly stable (if it is so, even yet) and what habitat is over the international border. And, for that matter in the Bangladesh part - I don't see any explanation of the drastic colour scheme of A and B, but hopefully/presumably it relates to habitat...

We used this map from a previous study, and we agree that the colour scheme is not relevant here and have replaced it with a similar map. One of the reasons we wanted to look for species in the more disturbed areas was to examine if these agricultural areas could supplement the more natural forest types as habitats for these species – in the hopes that working with local communities ultimately, they could produce viable populations. There is very limited natural forest left in these parts of Bangladesh, so we still feel it is important to show which species are persisting to help with our work in the future to restore aspects of habitat.

First sentence of the Conclusion. This is a very assertive statement and for SE Asia it is surely untrue; only a minority of species of "Small to medium sized nocturnal mammals" are grassland or otherwise non-forest. I doubt Bangladesh is much different. If this statement is retained, it needs to be given clear evidentiary backing. The effects of differing visibility in different habitats are profound and not acknowledged anywhere in this MS. They prevent any responsible inferences about abundance patterns based on ER patterns without careful consideration of effects on those ERs other than abundance. Lamping is generally more productive in encroached areas not because there are more animals there, but because the process of encroachment opens up visibility to see reflecting eyes.. Methods not dependent upon this 'chance' of detection, such as camera-trapping, do not typically show that for "Small to medium sized nocturnal mammals ... the forest types offered by protected areas frequently do not provide suitable habitat"; indeed quite the reverse. Among mainland SE Asian small carnivores this statement is applicable only to the various otters, to Small Asian Mongoose Urva javanica and to a lesser extent to Small Indian Civet Viverricula indica.

This differs from our experience, but we respect that each country is different in the remaining secondary forests that are parts of the protected area system and we have changed the tone of this sentence and hope it is less dramatic and more inclusive.

Reviewer 2

 

Nocturnal mammals in tropical hill forests of Bangladesh. The paper is well written and covers much of the literature on the focal species and location. I have some suggestions that i think would clarify some points and provide more information to the readers.

Thank you for your kind and helpful comments and we hope we have been able to address them.

 

Line 14. Intensification of agriculture is not as much an issue as expansion of agriculture, as mentioned later in the introduction.

We have amended this

 

Figure 1. in panel B indicate the location of each of the 4 forests either with letters that match the subsequent panels or arrows.

We have changed this

 

Line 62. You are not estimating the abundance of nocturnal mammals, at best you are creating an index of relative abundance.

We have changed this

 

Lines 78-118. You provide details on each reserve. You do not provide a size for Rema-Kelenga.

This is added

The measures for Adampur are in English units and should be in metrics as the others. 

This is amended

The fourth reserve – Adampur is listed in Figure 1 as Rajkandi. 

This is corrected

Theses 4 reserves are relatively close to each others and seem to share the same temperature range and rainfall (maybe Adampur has less rainfall). This can all be consolidated. Also see no reason to report humidity.  You describe the natural vegetation in each reserve but not the vegetation differences between natural forest, gardens and plantations.

We have consolidated and included the descriptions between the three vegetation types.

 

Section 2.2 – Survey methods. You do not report if you had any restrictions on weather – such as no surveys during rain. 

We include a line that all surveys were during dry periods.

It is not clear if you repeated transects or each was walked a single time.

We have included a statement that we only walked each transect once

 You report the total km walked but not the distribution in each reserve.

We already had included this information in the table in terms of number, but we add a line also to indicate the length.

  If this was a camera trap survey you would need to track the vegetation density to account for detectability in the different forest types. Can you at least indicate the average visibility distance in each type?

We agree with both reviewers that having a measure of probability of detection and visibility would have been very useful. Unfortunately, the team did not record that at the time and we now cover  this aspect in the discussion. Our main purpose was to understand presence of various species but in a quantifiable way that was accessible to the survey team.

You never mention how you dealt with groups – were all observation of individuals or were groups counted as one observation?

All animals spotted were single individuals and this is now added to the methods.

 

Section 3 – results. I would like to see a comparison between the reserves. If the reserves are different in their mammal communities. you have measured several metrics – forest type along transects, human density, plantation crop. You test these metrics for the species, I would like to see if they differ between the reserves themselves.  One reserve is far larger than the others and may or may not retain additional species. Table 2 does not include site as a covariate, possibly due to sample size.

We have added more information in the length of transects by habitat in each reserve, plus more details on each reserve in the methods.

I would be interested in what range of HII was encountered in each reserve. It seems the values would be very low for the 4 reserves.  Maybe a short table that lists the range of metrics for each reserve. Including amount of each forest type along transects.

We have now added HII for each reserve as well in Table 1.

 

You drop jungle cat and leopard cat from the analysis due to small sample size. So they should be dropped from the table 1 and table 3.  It is enough to say they were encountered in x and y reserves, but at too low a number for consideration. Lines 240-242 are part of the same issue that you make conclusions about the 2 species based on little data.  They would have been detected more often with camera traps, as less sensitive to cameras than human presence.

Since so few data exist on these taxa in Bangladesh, we really wanted to highlight their low abundance. Indeed, in foot surveys in other countries where we have worked, we regularly encounter them and even have been able to follow them to collect ecological data. So, we feel these low numbers have some validity – but indeed they are too low for any statistical analysis. We have altered the discussion too to point out why we feel including these data is important, partly too to guide locations of future studies.

Table 3. It should be pointed out that you are not aware of variance around these numbers so differences may not be significant.

We added this – thank you

 

Line 243. You mention camera traps and the vast majority of new surveys are using camera traps and not transects (diurnal or nocturnal). You advocate for more nocturnal transects (Lines 260-277). The real question is whether these species would have been detected with camera traps. Did you keep any records of the height of detection for your encountered. If the species were within 3 m of the ground they would have been detected with cameras. For species that are truly arboreal they would not have been detected without placing cameras in the canopy. The question is whether transect surveys (nocturnal or diurnal) are needed to supplement camera trap surveys or careful placement of camera traps would cover these species.  Two studies using canopy cameras are listed below.

Gregory, T., Carrasco Rueda, F., Deichmann, J., Kolowski, J. and Alonso, A., 2014. Arboreal camera trapping: taking a proven method to new heights. Methods in Ecology and Evolution, 5(5), pp.443-451.

Yihao, F., Guopeng, R., Ying, G., Shuxia, Z., Haohan, W., Yanpeng, L., Zhipang, H., Liangwei, C. and Wen, X., 2018. Impact of ground and canopy camera-trapping installation on wildlife monitoring. Biodiversity Science, 26(7), p.717.

We agree that camera traps would have been great. We have added a statement about the difficulties of camera trapping in Bangladesh. Furthermore, though we do not feel it is suitable to put in the paper, four years ago, when this study was undertaken, camera traps were almost wholly unavailable and if they could be imported, it was with a 400% import tax. These days arboreal traps and the methods to use them still are far beyond the reach of most researchers, including due to heavy import duties of such equipment. Indeed, to get any kind of climbing equipment, travel to Nepal is necessary. We added the two useful references to point out that indeed arboreal camera trapping is a new important method to consider for future studies.

 

Line 154-158. Maybe the 2 sentences should be joined with “, however” replacing “. Overall”

We have edited this section

 

Table 1. add superscripts to indicate significant differences between the reserves for each species.

Added

Round 2

Reviewer 1 Report

This MS retains fundamental problems (the title is misleading and throughout the MS more is claimed from the records the survey generated than is responsible) despite some revision; but these problems can be solve largely through rewording and some excision of material.  Although references to 'abundance' have mostly been removed (but not in all places, including most unfortunately the title!), the wider issue, that encounter rates (without a lot of contextual info) don't allow confident conclusion about patterns of abundance has not been addressed. There remain many potentially spurious conclusions in the MS - they're also potentially correct, but the thing is that it's not possible to tell, and therefore such conclusions should not be presented. I may not have flagged them all.

The overall summing up is weak. What it should be saying is that the evidence of these species living in anthropogenic vegetation assemblages is important, and what is still required is much greater clarity on the roles these habitats could play in the conservation of the species in question. It's an absolute priority to know whether these habitats can support these species on their own, or whether they're only overflow areas from adjacent native habitats. The small/tiny size of the areas surveyed here and the lack of information provided about the wider landscape in which they sit precludes any clarity on that from these results (but the second issue should be addressed in the MS). The MS should call for this further information because it has profound implications for conservation planning. If these animals' populations are fully viable independent of native habitat, and especially if (as the MS postulates but with, as far as I can see, absolutely no credible basis to do so) populations are higher in anthropogenic areas than in native habitats, then it's quite correct that - as stated in the MS, but implicitly as a 'problem' - these species are ignored in PA planning: the PAs need to focus on the species that won't survive, or at least will have major declines, independent of them. Whereas if presence of these species in anthropogenic habitats depends on adjacent native habitat, then the PAs do need to consider their needs, and, where possible it'd be good to site these land-uses adjacent to PAs to allow larger and thus more resilient populations of these species. If the MS wishes to extend to discussion and recommendations, rather than 'just' (note, that this is its primary function) presenting the records, it needs to be along these lines, not drawing premature and thus possibly incorrect conclusions on aspects for which the available information is insufficient.

Comments for author File: Comments.pdf

Author Response

This MS retains fundamental problems (the title is misleading and throughout the MS more is claimed from the records the survey generated than is responsible) despite some revision; but these problems can be solve largely through rewording and some excision of material.  Although references to 'abundance' have mostly been removed (but not in all places, including most unfortunately the title!), the wider issue, that encounter rates (without a lot of contextual info) don't allow confident conclusion about patterns of abundance has not been addressed. There remain many potentially spurious conclusions in the MS - they're also potentially correct, but the thing is that it's not possible to tell, and therefore such conclusions should not be presented. I may not have flagged them all.

We thank the reviewer for their careful reading of the manuscript. We had a limited time for revision and were largely working in a remote location in Bangladesh. We are embarrassed that the title did not retain a revision we thought we had made and apologise. We have now shortened the title and removed the inappropriate terms. We also did a search for the term abundance in the entire manuscript and did our best to keep it for when it referred to studies that did calculate actual abundance.

The overall summing up is weak. What it should be saying is that the evidence of these species living in anthropogenic vegetation assemblages is important, and what is still required is much greater clarity on the roles these habitats could play in the conservation of the species in question. It's an absolute priority to know whether these habitats can support these species on their own, or whether they're only overflow areas from adjacent native habitats. The small/tiny size of the areas surveyed here and the lack of information provided about the wider landscape in which they sit precludes any clarity on that from these results (but the second issue should be addressed in the MS).

We have edited the summing up substantially. We focussed more on Bangladesh and made the implications of the study far less broad, and highlighted the issues stated above.

The MS should call for this further information because it has profound implications for conservation planning. If these animals' populations are fully viable independent of native habitat, and especially if (as the MS postulates but with, as far as I can see, absolutely no credible basis to do so) populations are higher in anthropogenic areas than in native habitats, then it's quite correct that - as stated in the MS, but implicitly as a 'problem' - these species are ignored in PA planning: the PAs need to focus on the species that won't survive, or at least will have major declines, independent of them. Whereas if presence of these species in anthropogenic habitats depends on adjacent native habitat, then the PAs do need to consider their needs, and, where possible it'd be good to site these land-uses adjacent to PAs to allow larger and thus more resilient populations of these species. If the MS wishes to extend to discussion and recommendations, rather than 'just' (note, that this is its primary function) presenting the records, it needs to be along these lines, not drawing premature and thus possibly incorrect conclusions on aspects for which the available information is insufficient.

We hope that, as the reviewer suggested, by deleting any possible conjecture, we have made these sightings more useful to address further study. We also feel it is very important for these data to be published in the Bangladeshi context as currently there is no management of the agroforestry systems, which are growing substantially. By showing their potential for conservation this can have important implications for Bangladesh, but because these small species are normally not studied this issue is not raised. This paper would really help support work outside the (as you note very tiny) remaining protected areas.

Round 3

Reviewer 1 Report

parts of the MS have been well improved but the improvement has been patchy, as if rushed. There is still confusion over encounter rates vs abundances. Parts of it remain highly ambiguous / inconsistent, such as the habitat terminology, even though many of the points in the attached commented-on MS have been raised before. I see the authors wrote "We had a limited time for revision"; it come across as rather rude and arrogant to send it back in the half-baked state in which it currently is; doesn't it occur to you that maybe reviewers might also be short of time and to have to go through and make many of the same points as last time round is not apt? If you don't have enough time to do the revision properly, then tell the journal you need more time!

Comments for author File: Comments.pdf

Author Response

We again thank the reviewer for his/her comments and suggestions - we appreciate the time and effort invested in reviewing our paper. We have addressed all the comments and suggestions, often by including the suggestions / caveats as worded by the reviewer in the text.

We have also made it clear that the four forest reserves we worked in in Bangladesh are all continuous with small amounts of forest mostly to the north, and, importantly, with forest to the south, i.e., in Tripura in India. As such they are less isolated than may appear by judging them on the gazetted sizes. 

We have checked the references and we noticed that for a series of them they were 2 off - we have now corrected that. 

We checked the consistency of the terms and terminology we used. 

Throughout we have changed Adampur to Rajkandi -- Adampur is one of three beats in Rajkandi and occupies 3/4th of it, but the two other beats are continuous with it and hence all is better referred to as Rajkandi (as indeed indicated on our map).

We hope the paper is now acceptable for publication

 

 

 

Round 4

Reviewer 1 Report

this is a good revision. I noted one point for clarification - line 158, was it really 'animal count'? not 'animal encounter rate'?

Author Response

We thank the reviewer for granting that the current version is suitable for publication. Regarding the comment at line 158, we used animal count as dependent variable and transect length as an offset. This is a common technique to handle count data on transects with different length (we also added a reference). So we do not need to change the content of the paper.