Evaluation of Early Bark Beetle Infestation Localization by Drone-Based Monoterpene Detection

Round 1

Reviewer 1 Report

The paper describes novel technology with great potential in study of volatiles emitted from forest stands. The paper is technically very well written. Used technology is impressive.

However, α-pinene is a standard component of forest air. It can not be used for detection of bark beetle attack. The highest concentration of α-pinene is usually near to fresh logging debris or wind-broken trees.  The only way how to use this technology is the detection of bark beetle pheromone. However, it could be possibly more difficult due to lover concentration of the compounds in air.

I suggest to change title to: Method of mapping α-pinene concentration in the forest air with the use of UAV, rewrite introduction, discussion and conclusion. It would be good to involve expert on bark beetle chemical ecology to authors team.   

Author Response

The paper describes novel technology with great potential in study of volatiles emitted from forest stands. The paper is technically very well written. Used technology is impressive.

Reply: Thank you very much for appreciating our methodological approach.

However, α-pinene is a standard component of forest air. It can not be used for detection of bark beetle attack. The highest concentration of α-pinene is usually near to fresh logging debris or wind-broken trees.  

Reply: Basically we agree with you, as we pointed this out in Line 405 – 411 when referring to intense drought stress emissions. We added a paragraph in the discussion in Line 422 – 443 with an additional table 1 and Line 461 – 465 to highlight the limitations of the sensor system, also relating to the emission sources you mentioned. However, we addressed these sources as “false positives” and a false positive detection does not imply that the sensor concept cannot work. It only means that the sensor system can also create misleading information. We reviewed the literature critically in relation to the comparison of infestation stages and drought stress induced emissions (Table 3 and Line 422 – 343).

The only way how to use this technology is the detection of bark beetle pheromone. However, it could be possibly more difficult due to lover concentration of the compounds in air.

Reply: We absolutely agree that the detection of the pheromone would be better, but as you state, the concentration is very low and the sensors are not selective enough to detect the pheromone mixed with higher concentrations of e.g. alpha-Pinene. This was added in Line 368 – 377 including a respective literature source.

I suggest to change title to: Method of mapping α-pinene concentration in the forest air with the use of UAV,

Reply: This paper deals with the evaluation of the possibility to detect bark beetle infestation at an early stage by means of α-pinene. Many studies are cited that confirm that this approach is basically reasonable, however, the limitations are discussed accordingly, including the aspect given by the reviewers. Therefore, we changed the title to “Evaluation of early bark beetle infestation detection by drone-based monoterpene detection”. We consider youre comment as very helpful because it broadened our perspective. For instance, it is realistic to use our sensor at larger distances above the canopy to assess forest / atmosphere interactions. Respective literature sources are added in the manuscript to support this idea (Line 405 – 408). However, the general scope of the manuscript is kept on bark beetle detection with respective discussion of false positive signals.

rewrite introduction, discussion and conclusion.

Reply: Introduction, discussion and conclusion were adapted as stated above, adding changes in Line 55 – 56, 87 - 100 (Introduction), Line 368 – 377, 405 – 411, 422 – 443, 461 – 465, 468 - 472 (Discussion) and Line 511 – 514 (Conclusion)

It would be good to involve expert on bark beetle chemical ecology to authors team. 

Reply: The corresponding author did his PhD in chemical ecology in a forest department, working on electrophysiology and odorant binding proteins of flies and beetles. Therefore, the chemical ecology of bark beetle pheromones was shortly reviewed and included in the Introduction and discussion in Line 87 – 100 and Line 368 – 377. As stated above, the pheromones are of no technical use, as they are undetectable for the sensors.

Reviewer 2 Report

Manuscript Review: Project Protect Forest: Early detection of bark beetle infestation by drone-based monoterpene detection

Paczkowski et al.

The authors describe the creation of a drone based olfactometer that can detect alpha-pinene from above tree crowns.  This is a novel and interesting approach.  The goal of this work is to have a system that could detect alpha-pinene and then be used to determine the beginning of a bark beetle outbreak.  I find this an interesting approach, but as it stands this is all very preliminary and future studies are planned.  I understand that this is a “communication”, but my suggestion is to wait to publish until after tests are completed during the spring 2021 where new modifications are tested.  Right now, the manuscript describes a system that is already changed through the steps one would take to roll out a new method.  So, the manuscript stops one step to early.  Once the next series of tests are added, it will be a much stronger manuscript and would not have to worry about duplicative publishing.

My expertise is in bark beetle ecology and management and I have no experience with sensors and the technology outlined here for detecting alpha-pinene in the environment.  That said, I do have some comments that I hope will help strengthen the manuscript from the bark beetle perspective:

• Is there a characteristic of terpenes that make them more easily detectable than bark beetle pheromones? I imagine there is because you chose a generic host volatile over a specific bark beetle pheromone.  If you could use pheromones that would make this a much stronger technology.  You could use Ips or Dendroctonus pheromones, depending on what systems you were working in.  This way, the detection of the chemical would definitely relate to the presence of the bark beetle (it would not mean a successful attack was ongoing, just that the beetles were there at the time of flight) instead to the presence of tree stress, which could be drought influenced (as mentioned by authors), from fire, windstorms, or ice damage.
• General statements about bark beetle biology need to be cited and corrected. As an example, L35-40: D. ponderosae never has four generations, fertility is not related to generations, and sister broods are not in addition to generations as they are from the same generation (i.e., emergent parent beetles after laying brood in one tree).  Further, for a beetle that has one generation per year, detecting trees months or almost a year later is useful.
• Authors mention detection methods within “spectral range of 450-890, but do not contextualize this. Visual surveys using aerial detection methods are very effective at mapping bark beetle infestations.  Mention of these alternative methods should be added here – visual as well as satellite based which is suggested with near IR by authors
• The paragraph (L47-56) on host defenses needs improvement. I would suggest the authors edit this to follow common host defense literature, i.e., preformed (=constitutive) and induced host defenses of conifers.  It’s not just about pitching the beetles out.  In this section you could also touch on how drought and other stresses affect host defenses since that is discussed later in the manuscript.
• The paragraph about sensors is technical compared to what most readers will understand. Perhaps reduce the text here or put it in a more easily digestible format.  This could also just be my view since I am mostly unfamiliar with these techniques.  However, I do not think the details at this level are needed for this manuscript.
• L129: What do you mean by “four m heights”? Is this a pole broken out in 4 m increments?  Be specific here.
• Why was such a small stand selected? I would think a better test in forest would be to use a larger stand.  Alternatively, did you consider testing this under more uniform conditions in the field, like over a young plantation or nursery where you could easily survey the trees and create damage if you wanted?
• Need citations for many statements
• There are many steps taken and a lot of integrated technology. I think since you are proposing a new method for survey, adding the cost range of putting this equipment together as well as the hours and expertise it took to do a single survey would be really useful to allow readers to determine if this is something to explore or out of their budget/expertise.  In addition, add information on how many times a stand would need to be surveyed in a season.  Even if beetles disperse once in the spring, alpha-pinene in the environment would vary, so I imagine multiple flights would be necessary.
• L308-: It’s important to note that many bark beetles disperse relatively early in the spring when avg temperatures are still relatively low. They attack trees and then while temperatures continue to rise through the spring/summer, much of the development occurs.  I do not know if that matters too much to your method, but the damage and volatile release could be lower than expected because it is emitting over a longer period under cooler temperatures.  Also, after a tree responds to a bark beetle attack, the terpene profile does not stay consistent.  Consequently, alpha could be replaced by another terpene as the most common terpene emitting from trees.
• L318-: drought is just one factor in releasing alpha-pinene into the environment. Trees damaged by ice storms, wind, and fire would all be releasing alpha.  These are all situations that one would be interesting in detecting bark beetles in the surrounding forests, so losing the signal with your methodology during these circumstances is problematic.  If you could modify the technique being developed to detect bark beetle pheromones you would have a useful tool.
• L320: Another general statement here that all drought-stricken stands will suffer bark beetle attack. This is much more nuanced than this, with many stands like what you describe not colonized by bark beetles at all.
• L334: If I am following this correctly, this would be a severe limitation to using this method for survey. A single non-host tree through off a false positive?  Many conifer bark beetle systems have hardwoods at least in the understory if not also in the intermediate and codominant canopy layer.  Would those trees often release semiochemicals that would be problematic?

Author Response

The authors describe the creation of a drone based olfactometer that can detect alpha-pinene from above tree crowns.  This is a novel and interesting approach.  The goal of this work is to have a system that could detect alpha-pinene and then be used to determine the beginning of a bark beetle outbreak.  I find this an interesting approach, but as it stands this is all very preliminary and future studies are planned. 

Reply: Thank you very much for your appreciation of our approach.

I understand that this is a “communication”, but my suggestion is to wait to publish until after tests are completed during the spring 2021 where new modifications are tested.  Right now, the manuscript describes a system that is already changed through the steps one would take to roll out a new method.  So, the manuscript stops one step to early.  Once the next series of tests are added, it will be a much stronger manuscript and would not have to worry about duplicative publishing.

Reply: We agree that our results are preliminary and basically show the potential of the application. However, we believe that the novelty of the approach justifies an early publication, as the scientific community will be interested in the approach. This communication can induce a scientific discussion on further applications in forestry, as well as agriculture.

My expertise is in bark beetle ecology and management and I have no experience with sensors and the technology outlined here for detecting alpha-pinene in the environment.  That said, I do have some comments that I hope will help strengthen the manuscript from the bark beetle perspective:

• Is there a characteristic of terpenes that make them more easily detectable than bark beetle pheromones? I imagine there is because you chose a generic host volatile over a specific bark beetle pheromone.  If you could use pheromones that would make this a much stronger technology.  You could use Ips or Dendroctonus pheromones, depending on what systems you were working in.  This way, the detection of the chemical would definitely relate to the presence of the bark beetle (it would not mean a successful attack was ongoing, just that the beetles were there at the time of flight) instead to the presence of tree stress, which could be drought influenced (as mentioned by authors), from fire, windstorms, or ice damage.

• Reply: We added a discussion on this issue in the manuscript comprising chemical ecology and biosensor aspects and we agree with you that detecting the pheromone would be ideal. Future research might reveal a solution for pheromone detection, but we believe this will be a matter of many years or even decades -> Line 87 – 100, Line 368 – 377. We also addressed the drought induced VOC emissions adding Line 81  84, Line 422 – 443 and Table 1. In general we believe that more data is needed, as the literature sources are partly contradictory or not comparable due to different methods. Our drone solution can allow an easier VOC sampling, thereby increasing research studies in this area.

• General statements about bark beetle biology need to be cited and corrected. As an example, L35-40: D. ponderosae never has four generations,

• Reply: We deleted D. ponderosae (Line 37)

• fertility is not related to generations

• Reply: We only stated that hot and dry seasons increase the fertility, as stated by Rouault el al and McCollum and Lundquist. We change the grammar of the respective sentence to avoid misunderstanding (Line 37)

• and sister broods are not in addition to generations as they are from the same generation (i.e., emergent parent beetles after laying brood in one tree). 

• Reply: Yes, this is another misunderstanding that needs clarification. We deleted the term sister brood (Line 37)

• Further, for a beetle that has one generation per year, detecting trees months or almost a year later is useful.

• Reply: That is correct, however we refer particularly to the recent situation, where several generations are the usual case. Therefore, we focused on Ips typographus and Pityogenes chalcographus to avoid misunderstandings (Line 36)

• Authors mention detection methods within “spectral range of 450-890, but do not contextualize this. Visual surveys using aerial detection methods are very effective at mapping bark beetle infestations.  Mention of these alternative methods should be added here – visual as well as satellite based which is suggested with near IR by authors

• Reply: We clarified the text passage on visual detection of bark beetle. Our statements are that 1) early detection in the green stage is necessary to prevent an outbreak of Ips typographus and 2) early detection in the green stage cannot be done by optical drone or satellite sensors (Line 45 – 49)

• The paragraph (L47-56) on host defenses needs improvement. I would suggest the authors edit this to follow common host defense literature, i.e., preformed (=constitutive) and induced host defenses of conifers.  It’s not just about pitching the beetles out.  In this section you could also touch on how drought and other stresses affect host defenses since that is discussed later in the manuscript.

• Reply: We added literature on drought stress emission from plant, although there seem to be only few studies with concise emission values (Line  81 – 84) and especially discussed this issue in Line 412 – 421 together with Table 1. Drought stress emission is one more point that can already be evaluated with our drone solution, which we addressed in Line 461 – 465. The paragraph on bark beetle defense was improved (Line 65 – 86) by adding several literature sources. However, it was not the goal of the study to review constitutive and induced host plant defense mechanisms, but to focus on induced host plant defense in relation to VOC emissions.

• The paragraph about sensors is technical compared to what most readers will understand. Perhaps reduce the text here or put it in a more easily digestible format.  This could also just be my view since I am mostly unfamiliar with these techniques.  However, I do not think the details at this level are needed for this manuscript.

• Reply: We would like to keep the description, because we assume that part of the readers who are interested in the state of the development of our approach are not foresters or chemical ecologists, but sensor scientists and we like to show that we understood the basics of the technology that we use.

• L129: What do you mean by “four m heights”? Is this a pole broken out in 4 m increments?  Be specific here.

• Reply: Our choice of words was not good here. We meant a pole that is four m long and in one piece. (Change in Line 184)

• Why was such a small stand selected? I would think a better test in forest would be to use a larger stand.  Alternatively, did you consider testing this under more uniform conditions in the field, like over a young plantation or nursery where you could easily survey the trees and create damage if you wanted?

• Reply: There was a simple technical reason for using such a small stand: We were restricted in the amount of way points for the height profile and could not cover more area and we flew very slow, so using an area to be covered with one accu pack seemed feasible. The final application will not have these limits and then larger areas are possible.

• Need citations for many statements

• Reply: We added many citations to the text, increasing the total number from 30 to 51.

• There are many steps taken and a lot of integrated technology. I think since you are proposing a new method for survey, adding the cost range of putting this equipment together as well as the hours and expertise it took to do a single survey would be really useful to allow readers to determine if this is something to explore or out of their budget/expertise. 

• It is interesting that you think of this, because in the frame of the project we designed a feasibility study to assess the costs of the application. This is rather complex and will be content of a follow up paper. For now, we added some cost assumptions in Line 497 – 502, hoping that we will not distort a possible future market with our statement.

• In addition, add information on how many times a stand would need to be surveyed in a season.  Even if beetles disperse once in the spring, alpha-pinene in the environment would vary, so I imagine multiple flights would be necessary.

• Reply: We agree and added Line 500 – 502.

• L308-: It’s important to note that many bark beetles disperse relatively early in the spring when avg temperatures are still relatively low. They attack trees and then while temperatures continue to rise through the spring/summer, much of the development occurs.  I do not know if that matters too much to your method, but the damage and volatile release could be lower than expected because it is emitting over a longer period under cooler temperatures.  Also, after a tree responds to a bark beetle attack, the terpene profile does not stay consistent.  Consequently, alpha could be replaced by another terpene as the most common terpene emitting from trees.

• We added this valuable information in the discussion in Line 391 - 395

• L318-: drought is just one factor in releasing alpha-pinene into the environment. Trees damaged by ice storms, wind, and fire would all be releasing alpha.  These are all situations that one would be interesting in detecting bark beetles in the surrounding forests, so losing the signal with your methodology during these circumstances is problematic.  If you could modify the technique being developed to detect bark beetle pheromones you would have a useful tool.

• Reply: We shortly discussed this issue relating to “false positives” from other sources than drought in Line 418 – 423.

• L320: Another general statement here that all drought-stricken stands will suffer bark beetle attack. This is much more nuanced than this, with many stands like what you describe not colonized by bark beetles at all.

• Reply: We found it very hard to discuss the details and still being able to write in a concise style not to confuse readers who are not so much in VOC and chemical ecology. However, we relativized especially this point in Line 412. Virtually any statement in such a complex situation (vapor pressure, temperature, plume dynamics, cause of emission at which stage of infestation) will be a little inexact.

• L334: If I am following this correctly, this would be a severe limitation to using this method for survey. A single non-host tree through off a false positive?  Many conifer bark beetle systems have hardwoods at least in the understory if not also in the intermediate and codominant canopy layer.  Would those trees often release semiochemicals that would be problematic?

• Reply: Yes, another “false positive” issue that cannot be resolved easily. For instance, as stated, flowering will lead to odors that can lead to signals on the sensor surface, which are hardly differentiable. Future studies will determine the detection rate for the proposed system and this rate will include the detected “false positives”. We might add optical data from the drone itself to exclude obvious “false positives”, e.g. broadleave trees, flowering trees, broken branches or trees etc. Line 468 - 473

Reviewer 3 Report

The laboratory part of the work is impressive and test results are fine.

I am not sure, the paper needs to include the field trial with the problems described.

But it clearly shows, that the development of a simple tool for foresters will take some time and a lot of consideration during interpretation of data, especially if the wind conditions are not good.

The most concerning about the field result is the lack of specificity of the sensor and false positive results above the prunus domestica tree. The differentiation of isoprene and α-pinene should definitely be resolved before another field trial.

I have added several comments and suggestions in text - see attached PDF.

Comments for author File: Comments.pdf

Author Response

The laboratory part of the work is impressive and test results are fine.

Reply: Thank you very much for the appreciation of our work.

I am not sure, the paper needs to include the field trial with the problems described.

Reply: We believe that the field try shows the recent limitations of the system and is ideal to give a practical outlook for future development. Also, the measuring concept can be used for different approaches already, e.g. forest / atmospheric interactions or drought stress mapping, as we added in Line 461 – 465.

But it clearly shows, that the development of a simple tool for foresters will take some time and a lot of consideration during interpretation of data, especially if the wind conditions are not good.

Reply: We absolutely agree and we work to push the method beyond a proof of principle in the field

The most concerning about the field result is the lack of specificity of the sensor and false positive results above the prunus domestica tree. The differentiation of isoprene and α-pinene should definitely be resolved before another field trial.

Reply: The capabilities of the sensors selectivity in the field are limited, but there are possibilities, e.g. using temperature cycles or more sensors to rule out isoprene or other plant VOCs.

I have added several comments and suggestions in text - see attached PDF.

note prosmissing results from spectral trajectories eg.: Hais M., Wild J., Berec L., Brůna J., Kennedy R., Braaten J. & Brož Z. (2016) Landsat Imagery Spectral Trajectories — Important Variables for Spatially Predicting the Risks of Bark Beetle Disturbance. Remote Sensing 8: 1–22. https://doi.org/10.3390/rs8080687

Reply: Thank you for this literature source that we added and elaborated in Line 49 – 54. Such “holistic” approaches can narrow down possible areas of infestation and it is good to add it to the manuscript.

the concentration range is wider, than in the graphs below (10 -8  - 10-4) vs (10 -6  - 10-4)

Reply: Yes, this is true and we forgot to address this issue in Figure 1, although it is mentioned in the text. Therefore, we added “Concentrations below 10^-6 were not shown as they did not elicit a sensor response.” in the caption of Figure 3 (Line 260), as this is an important information.

“4”

Reply: We change the term “four” to “4” in Line 184.

same height as pole, but with 55 cm capilary - or is the flight line counted to the inlet of the capilary

Reply: We see that we did not describe the method in detail and added information in Line 189 and Line 194 – 195.

what method/settings - kriging as bellow?

Reply: Again, we added information on the interpolation of the first heat map, which was not done in the same way as in the forest field test (Line 208 – 209).

how was this done? GPS positioning?

Reply: Again, we highlighted this information in Line 213 – 214. We used an inverted water shed algorithm developed by ESRI.

following the note above - the prepared concentrations ranged from 10-8 to 10-4 - part is missing

Reply: We added an explanation in the caption, as stated above (Line 263 - 264)

modify scale for this sensor to show response, or modify whole to use log scale

Reply: We would like to keep the scale, as this allows a visual comparison of the three sensor responses, which is important for the discussion to understand the difference of the reaction depending on the metal type. However, we agree that apart from this comparison, the respective part of the Figure 1 is non-explanatory and therefore we have added Table 1 with concise information on sensor response and linearity of response development at increasing concentrations.

???

Reply: We see that the respective paragraph is not explanatory and even contradictory and changed the text in Line 265 – 267

these connections do not make sense - mixing distance and velocity chnage

Reply: We agree from a mathematical point of view, but would like to keep the lines, as it simply improves the readability of the Figure 2. Without lines readers would find it hard to understand the discussion of Figure 2, which is already rather complex. Usually, such lines should be avoided in graphs, as they suggest interpolation of the results, which is not the case here. We added this to the caption of Figure 2 for clarification (Line 303 – 304)

indicate with arrow in map - this wuld increase readability of the map

Reply: The average wind direction was added to the map in Figure 6 (Line 336).

“4”

            Reply: “4” was changed to “four” in Line 333

possible additional sources here (out of map)?

Reply: We discussed this remark accordingly in Line 468 – 473 and in general in Line 488 -490 and give an outlook on possible applications in Line 461 – 465.

indicate wind direction and speed

Reply: We indicated wind direction and speed in Figure 6, Line 336

how were they localized

Reply: This was added to the method in Line 207 – 208.

SnO2/SnO3 ?

Reply: Thank you, we missed the O here (Line 329)

Reviewer 4 Report

This paper describes preliminary studies to test the concept of using drones fitted with sensors to detect elevated levels of a-pinene as an early indication of bark-beetle infested trees.  This is an interesting new technique and the results suggest that it could be promising.  I believe the paper would be suitable for publication with major revisions to address the comments below:

1. Overall the paper is fairly-well written, but it could use some editing to improve minor English grammatical issues throughout.
2. It would be helpful to provide more background information in the introduction about what is currently known in the scientific literature about monoterpene levels in healthy and infested stands, in the vicinity of healthy trees and infested trees, and at different heights in the tree canopy. Some studies have sampled monoterpene emissions from the trunks of trees near ground level and near the trees.  What are typical background levels of a pinene in stands?  Is there seasonal variation in monoterpene emissions?
3. Were the sensors selective for a-pinene or do they detect other volatiles as well?
4. A figure with a photo or schematic diagram of the set up would be helpful. How was gas flow led over the surface of the sensors?  Were they connected by tubes?
5. How were the a-pinene dilution levels selected for the calibration study? Are they representative of realistic a-pinene concentrations reported near infested trees in the field from published literature?
6. Since this is a preliminary feasibility study to test a new concept, I am not terribly concerned that only a single sensor of each of the 3 types was tested. However, ideally multiple sensors of each type would have been evaluated.  The sensor calibration trial was well-replicated with each a-pinene concentration tested 13 times.  The wind-tunnel experiment also appeared to be replicated 6 times.  It is unclear how many replicates were conducted for the artificial field test with the drone flights over  the 4-m pole and for the field test with drone flights over a forest stand.  How many flights were made over the artificial test site and the forest stand? Please indicate the number of replicates.  The tables of results give means and standard deviation, but no N values.  The statistical analyses appear to be appropriate.
7. Were all 3 sensors mounted on the drone and sampled simultaneously for each drone flight?
8. It would be helpful to include more information about the forest stand that was tested. What was the tree species composition, stand density, basal area, tree height, any signs of bark beetle infestation?
9. The discussion brings up many logistical considerations including low spring temperatures that could limit detection of a-pinene for early season detection, background levels of a-pinene due to drought or other factors, and wind-transport of volatiles that interferes with point location of the individual trees from which they are emitted. Ground truthing would be necessary to locate infested trees and confirm results.  It is unclear whether ground surveys would actually be more efficient.
10. Could sensors be adjusted to be specific to bark beetle pheromones?

Author Response

This paper describes preliminary studies to test the concept of using drones fitted with sensors to detect elevated levels of a-pinene as an early indication of bark-beetle infested trees.  This is an interesting new technique and the results suggest that it could be promising. 

Reply: Thank you very much for positively recognizing our new approach.

I believe the paper would be suitable for publication with major revisions to address the comments below:

1. Overall the paper is fairly-well written, but it could use some editing to improve minor English grammatical issues throughout.
1. Reply: The paper was completely edited and partly rewritten.
2. It would be helpful to provide more background information in the introduction about what is currently known in the scientific literature about monoterpene levels in healthy and infested stands, in the vicinity of healthy trees and infested trees, and at different heights in the tree canopy. Some studies have sampled monoterpene emissions from the trunks of trees near ground level and near the trees.  What are typical background levels of a pinene in stands?  Is there seasonal variation in monoterpene emissions?
1. Reply: We added several literature sources, respective discussions and especially Table 1 in the introduction to provide more insights in this complex field. However, we believe that the studies are hard to compare and a final conclusion on VOC emissions in forest stands can not be drawn. In detail please see Line 55 – 66 for general information, Line 87 – 100 on bark beetle pheromone dispersion and especially Line 422 – 443 on VOC emission literature and 468 – 473 as an example of “false positive” emission sources.
3. Were the sensors selective for a-pinene or do they detect other volatiles as well?
1. Reply: As added in Line 114 – 117 it is possible to detect monoterpenes, but by now not a specific one, e.g. alpha pinene. However, we added promising results of Schüler et al. in Line 353 – 355 and of Leidinger et al. in Line 357 – 359. We also clarified e.g. in Line 364 that only the class of monoterpenes are detectable.
4. A figure with a photo or schematic diagram of the set up would be helpful. How was gas flow led over the surface of the sensors?  Were they connected by tubes?
1. Reply: A new Figure was added to show the sensor setup (Line 154).
5. How were the a-pinene dilution levels selected for the calibration study? Are they representative of realistic a-pinene concentrations reported near infested trees in the field from published literature?
1. Reply: The selection was done on the basis of the lowest possible concentration of the sensor. The concentration detection limit of the sensor is now extensively discussed in Line 444 – 454 together with the literature in Table 1.
6. Since this is a preliminary feasibility study to test a new concept, I am not terribly concerned that only a single sensor of each of the 3 types was tested. However, ideally multiple sensors of each type would have been evaluated.  The sensor calibration trial was well-replicated with each a-pinene concentration tested 13 times.  The wind-tunnel experiment also appeared to be replicated 6 times.  It is unclear how many replicates were conducted for the artificial field test with the drone flights over  the 4-m pole and for the field test with drone flights over a forest stand.  How many flights were made over the artificial test site and the forest stand? Please indicate the number of replicates.  The tables of results give means and standard deviation, but no N values. The statistical analyses appear to be appropriate.
1. Reply: The number of replicates were added in all cases (Table 2, Figure 2, Table 3, Figure 3, Figure 4, Figure 5, Line 195, Line 231. The field tests were based on only one replication.
7. Were all 3 sensors mounted on the drone and sampled simultaneously for each drone flight?
1. Reply: Yes, this was the case and is clarified in Figure 1.
8. It would be helpful to include more information about the forest stand that was tested. What was the tree species composition, stand density, basal area, tree height, any signs of bark beetle infestation?
1. Reply: The information relevant for the VOC emission and infestation status is given in Figure 5. The part of the stand that was assessed by the drone and by foot consisted only of spruce trees at an age between 40 and 80 years. This info was added in Line 211 – 212.
9. The discussion brings up many logistical considerations including low spring temperatures that could limit detection of a-pinene for early season detection, background levels of a-pinene due to drought or other factors, and wind-transport of volatiles that interferes with point location of the individual trees from which they are emitted. Ground truthing would be necessary to locate infested trees and confirm results.  It is unclear whether ground surveys would actually be more efficient.
1. Reply: This is true. We are working on an evaluation whether this statement is correct or not and for the moment add a discussion on the economic feasibility in Line 497 – 502 with one of the results being that steep areas, areas with dense understory and large areas will probably be more feasible than conventional assessment by foot.
10. Could sensors be adjusted to be specific to bark beetle pheromones?
1. Reply: Unfortunately not, we are quite sure about that. We added an intense discussion and also surplus information in Line 87 – 100 and Line 368 – 377.

Round 2

Reviewer 1 Report

The new version of manuscript is considerably improved. It is a really good paper now.

Best regards

Reviewer 2 Report

Nice work addressing my previous comments.  The revised manuscript is improved and better contextualizes the methodology.  I look forward to reading future work that integrates drones and volatile detection.