Drone Monitoring of Breeding Waterbird Populations: The Case of the Glossy Ibis

Round 1

Reviewer 1 Report

Please take care to clearly define the differences between the 'visual expert counts', the 'manual counts' and 'automatic counts' in the introduction and method sections.

Also, in the Discussion section, it is worth mentioning that increasing the number of small sampling areas to perhaps three in each habitat type would either increase confidence in the ability of the automatic counter to consistently detect c.50% of the individuals in the image or establish the range of variation associated with the automatic counter.

Comments for author File: Comments.pdf

Author Response

Comments and Suggestions for Authors

#1- Please take care to clearly define the differences between the 'visual expert counts', the 'manual counts' and 'automatic counts' in the introduction and method sections.

#1- “Visual expert count” was replaced for “manual count” in the text to avoid misconception.

#2- Also, in the Discussion section, it is worth mentioning that increasing the number of small sampling areas to perhaps three in each habitat type would either increase confidence in the ability of the automatic counter to consistently detect c.50% of the individuals in the image or establish the range of variation associated with the automatic counter.

#2- Following this interesting comment, we have performed a new validation analysis, including three sampling areas in each habitat, to increase the confidence ability and to obtain a range of variation of the automatic counter. New results are in Table 2. We have also used these new validation plots for recalculating nest density and distance parameters obtained from manual counts (Table 1). Methods have been rewritten in order to reflect the new validation analysis, and results and discussion have modified in accordance with the results (lines 236-241, 267-174, 283-286, 309-313, 336-337).

#3- peer-review-3348314.v1.pdf

#3- Thank you so much for the thorough style revision of the text. We have modified all suggestions that appeared in the peer-review.pdf.

Author Response File: Author Response.docx

Reviewer 2 Report

In general, I find the study interesting and it shows very well the limitation of automatic counts in a cryptic species. The introduction could be improved to better "sell" the study, and there are a few things in the methodology that could be clarified. Another major problem is the phrasing and grammar. I am myself not a native english speaker, but I found way too many errors or poor choices of words and phrases were sometimes hard to understand. I reported some of them further in my comments, but there were too many to report them all. I strongly recommend that a native english speaker takes a look at the text.

General comments:

- Some of your sources are not cited in chronological order in the text. Is that in accordance with the style of the journal?

- Concerning the intro: You do a nice job of describing the general problems with waterbird surveys and UAVs, but I think there is a link missing between the description of the problems and your study species. I would like to know why exactly you chose the glossy ibis. Is it only because of its conservation status? Is it because nest counts are typically difficult to be conducted on the field? What in particular makes this study species interesting other than it being a waterbird? For me, the fact that this species is dark and cryptic makes this study very interesting, as most attempts to implement automatic counts to nest monitoring have usually been done on conspicuous species. 

I would also maybe liked to get a word about those automatic counting methods in the intro. Which one are you going to use? How does it work? What exactly are you going to do? Going into your methodology, I have only a vague idea of what you are going to do.

- I would maybe put more details in the supplementary material on the mechanics of your automatic detection algorithm (decision tree, etc.)

line 3: Should read "Case of the Glossy Ibis"

line 11: No comma before an "and"

line 11: I don't understand the part "consequently, objective of ecological monitoring programs"

line 18: should be "nest counting"?

line 27-28: the first sentence of an intro is usually a widely accepted fact, and therefore you should have a little bit more than one source from 2018 to support it. 

line 29-30: two statements in a row which are not cited. Again, you should start by showing that the ideas you rely on for your study are widely accepted.

line 35: remove comma after "both" and after "conditions"

line 37: I would maybe replace "flagship" with a more used term like "indicator species" or "umbrella species", depending on what you want to emphasize

line 38: "Nevertheless" should be replaced by "However", as "Nevertheless" implies that you go from something good to talk about something bad, while "however" is the opposite.  

line 42: Should read "Furthermore, the breeding places regularly contain several species breeding in sympatry", and you might want to change "place" with something else

line 43: Again, don't put two adverbs next to each other

line 46-50: Revise wording. 

line 46-50: This is the key sentence of you intro, justifying your research. Again, I would maybe add one or two more citations to show that drone surveys are really popular nowadays (e.g.: Brisson-Curadeau et al., etc.)

Line 53: change "truth"

line 68: change "during time"

line 76: I don't think the "Spanish Red Data Book" is going to appeal to anyone outside of Spain so I would just say that there are local concerns about the conservation of the species and then I would cite the Red Data Book.

line 77: Change "nevertheless" again

Line 81: change "densely"

line 89-98: Some of this information would have been more relevant in your intro.

line 108: how did the pilot follow the transect? Did you program the UAV beforehand?

line 110: 22 transects covering 16 hectares in only 30 minutes? Sorry to be skeptical but I find that very surprising. 

line 110: Check GPS location. It should be 6 degrees not 62

line 114: where was the operator positioned during the flights? Did he have to move to maintain that maximum distance of 500m?

line 123: adding a scale would be interesting 

line 132-136: I don't know much about mosaic building using software, but does the fact that all controll points are situated more or less in the same region provide a good enough base for the software to avoid distorsion for the rest of the mosaic?  

line 157: what about ibises not siting on their nests?

line 179: I am not too sure how you evaluated an omission or a commission, especially since the visual classification is vector based and the automatic classification is pixel-based. Did you overlay the two maps and visually identified when nests were matching? If so, how close would a visually-identified nest be to an automatically-identified nest for you to consider that there were a match?

line 203: I understand later that these birds are all nesting? Were there non nesting individuals also found?

line 205: Why only use 30 nests for calculating this metric? How did you choose those 30 nests?

line 212: you should include units in the table

line 216: You say at line 160 that you don't try to differentiate the two species. How exactly were you expecting the purple heron nests to be classified?

Line 237: I would add nuances to this statements. You think the breeding estimation was good because of the image quality, and it's fair to think that. However, you cannot state it as a fact until you have compared it with other methods, like ground-based counts.

 

 

 

 

Author Response

Comments and Suggestions for Authors

In general, I find the study interesting and it shows very well the limitation of automatic counts in a cryptic species. The introduction could be improved to better "sell" the study, and there are a few things in the methodology that could be clarified. Another major problem is the phrasing and grammar. I am myself not a native english speaker, but I found way too many errors or poor choices of words and phrases were sometimes hard to understand. I reported some of them further in my comments, but there were too many to report them all. I strongly recommend that a native english speaker takes a look at the text.

Thank you so much for your helpful comments. We have modified the text according to general and specific comments proposed by the reviewer. Particularly we have better focused the introduction into the objectives of our work, we have extended some parts of the methodological section that were unclear, and we have made a thorough revision of the grammar and English style of the text.

General comments:

#1- Some of your sources are not cited in chronological order in the text. Is that in accordance with the style of the journal?

#1-The style of the bibliography has been automatically assigned by the Endnote bibliographic manager according to MDPI style that we downloaded from the web. We have revised the style and it coincides with the journal requirements.

 

#2- Concerning the intro: You do a nice job of describing the general problems with waterbird surveys and UAVs, but I think there is a link missing between the description of the problems and your study species. I would like to know why exactly you chose the glossy ibis. Is it only because of its conservation status? Is it because nest counts are typically difficult to be conducted on the field? What in particular makes this study species interesting other than it being a waterbird? For me, the fact that this species is dark and cryptic makes this study very interesting, as most attempts to implement automatic counts to nest monitoring have usually been done on conspicuous species. 

I would also maybe liked to get a word about those automatic counting methods in the intro. Which one are you going to use? How does it work? What exactly are you going to do? Going into your methodology, I have only a vague idea of what you are going to do.

#2- We have added more information about the choice of the target species and the methodology in the introduction section (lines 93-105). The referee is right about the reasons for choosing the Glossy ibis as a species study, so we have developed them in the intro.

 

#3- I would maybe put more details in the supplementary material on the mechanics of your automatic detection algorithm (decision tree, etc.)

#3- As we have no longer any information for supplementary material in addition to what the referee proposes, we have developed the information about the detection algorithm in the methodological section (lines 194-199).

 

#4- line 3: Should read "Case of the Glossy Ibis"

#4- Amended

 

#5- line 11: No comma before an "and"

#5- Amended

 

#6- line 11: I don't understand the part "consequently, objective of ecological monitoring programs"

#6- We have reworded the sentence by “consequently, a potential object of ecological monitoring programs”. The study of species indicators (as waterbirds) offer valuable information on the ecosystem status. So, and consequently as we said, ecological monitoring of these species offers us an indirect measure of changes in the ecosystem.

 

#7- line 18: should be "nest counting"?

#7- We have replaced “nest counting” by “nest counts that would be”.

 

#8- line 27-28: the first sentence of an intro is usually a widely accepted fact, and therefore you should have a little bit more than one source from 2018 to support it. 

#8- We have added more citations to support this assertion.

 

#9- line 29-30: two statements in a row which are not cited. Again, you should start by showing that the ideas you rely on for your study are widely accepted.

#9- We have added more citations to support this assertion

 

#10- line 35: remove comma after "both" and after "conditions"

#10- Removed

 

#11- line 37: I would maybe replace "flagship" with a more used term like "indicator species" or "umbrella species", depending on what you want to emphasize

#11- We have replaced the term “flagship” by “indicator species”

 

#12- line 38: "Nevertheless" should be replaced by "However", as "Nevertheless" implies that you go from something good to talk about something bad, while "however" is the opposite.

#12- Done

 

#13- line 42: Should read "Furthermore, the breeding places regularly contain several species breeding in sympatry", and you might want to change "place" with something else

#13- We have replaced the sentence by “Furthermore, the breeding places often contain several species breeding in sympatry”.

 

#14- line 43: Again, don't put two adverbs next to each other

#14- We have removed one adverb. The sentence now reads as follow: “Consequently, aerial surveys have become…”

 

#15- line 46-50: Revise wording. 

#15- We have reworded the sentence by “Therefore, the application of UAVs has been a qualitative leap in bird monitoring, enabling rapid, low disturbance surveys of inaccessible areas while delivering repeatable images with a fine spatial resolution”.

 

#16- line 46-50: This is the key sentence of your intro, justifying your research. Again, I would maybe add one or two more citations to show that drone surveys are really popular nowadays (e.g.: Brisson-Curadeau et al., etc.)

#16- We have added two more citations: Brisson-Curadeau et al. 2017 and Han Y.-G. et al. 2017.

 

#17- Line 53: change "truth"

#17- We have replaced it by “ground-truth”

 

#18- line 68: change "during time"

#18- We have removed during time, because we have considered that this assumption is implicit when we said “patterns fluctuate interannual and seasonally”, at the beginning of this sentence.

 

#19- line 76: I don't think the "Spanish Red Data Book" is going to appeal to anyone outside of Spain so I would just say that there are local concerns about the conservation of the species and then I would cite the Red Data Book.

#19- We have added the IUCN Red Data Book citation.

 

#20- line 77: Change "nevertheless" again

#20- Done

 

#21- Line 81: change "densely"

#21- We have added “densely populated” by a better comprehension of the term.

 

#22- line 89-98: Some of this information would have been more relevant in your intro.

#22- We have moved some of the species information to the intro, in order to focus better the choice of the target species for the work.

 

#23- line 108: how did the pilot follow the transect? Did you program the UAV beforehand?

#23- The pilot did not follow the transect during the flight. The UAV was programmed beforehand with the path to assure that all the area was covered with the front and side overlap required to obtain a proper orthomosaic. We have added that the flight was automatic and programmed in the methods section (line 144). It is also mentioned in Figure 2 caption (“Automatic flight”).

 

#24- line 110: 22 transects covering 16 hectares in only 30 minutes? Sorry to be skeptical but I find that very surprising. 

#24- The UAV covered a total of 6.29 km, obtaining an image of 16 hectares, at a 12 km/h speed. It is easily verifiable now with the scale we have added in Figure 2. The effective time of flight was around 30 minutes (6.29 km / 12 km/h), but the total duration of the census was ca 1 hour (line 151). We had to change the battery during the route because it was not possible for the battery to cover the flight in one run.

 

#25- line 110: Check GPS location. It should be 6 degrees not 62

#25- Thank you for the notice. We have changed the coordinates.

 

#26- line 114: where was the operator positioned during the flights? Did he have to move to maintain that maximum distance of 500m?

#26- The operator was located in the asphalt area located at the north of the image (the parking). It was not necessary to move to another location during the flight because the maximum distance reached by the UAV was 325 meters (see Figure 2).

 

#27- line 123: adding a scale would be interesting 

#27- Scale has been added.

 

#28- line 132-136: I don't know much about mosaic building using software, but does the fact that all controll points are situated more or less in the same region provide a good enough base for the software to avoid distorsion for the rest of the mosaic?  

#28- This is a good point. Ground Control Points (GCP) were situated to evaluate the absolute accuracy of the georeferenced image, rather than the relative accuracy. Absolute accuracy concerns to the degree to which the measured position of a point on a map corresponds to its actual position in the real world, whereas relative accuracy refers to the degree to which the distances between points on a map correspond to the actual distance between those points in the real world. This means that GCP participate only in the identification of common points between frames, and were not taken into account in collinearity equations as in the orthophotographs georeferencing process. In this case, the number and location of points was enough to accomplish the absolute georeferencing objective. The term “absolute” has been added to the Figure 2 caption.

 

#29- line 157: what about ibises not siting on their nests?

#29- Glossy ibis individuals that were not hatching in their nests were foraging. The flight took place during the late hatching period when most breeding individuals are incubating or handling the chickens during their first days because nest occupancy is highly synchronous. During this stage, only one individual remains at the nest. Early and at the end of the day, individuals exchange their activities (hatching vs. feeding [Cramp, S. & Simmons, K.E.L. (eds.) 1977. The Birds of the Western Palearctic, Vol.I.]. Furthermore, foraging areas are located out of the “Lucio de la FAO”, in the marshland main area of the Doñana National Park, so not-breeding individuals do not interfere in the colony. Therefore, this is a very good time for censusing the colony, since there are no individuals out of the nests in the colony. We have added some of this information in the materials section (lines 153-157).

 

#30- line 179: I am not too sure how you evaluated an omission or a commission, especially since the visual classification is vector based and the automatic classification is pixel-based. Did you overlay the two maps and visually identified when nests were matching? If so, how close would a visually-identified nest be to an automatically-identified nest for you to consider that there were a match?

#30- The referee is right in that we have only mentioned and “spatial overlap” in the methodological section without details. We have added more details in the Validation section (lines 232-234, 242-245). To perform the spatial overlap, we used the vectorized layer of the raster classification (lines 226-230). In the other hand, we created a buffer area with 15 cm of radius around the nests detected manually. These two vector layers were then overlapped. Finally, we visually checked that no nests were located out but near the overlapped area between the automatic classification and the buffer area generated around the nests counted manually.

 

#31- line 203: I understand later that these birds are all nesting? Were there non nesting individuals also found?

#31- See #29 response.

 

#32- line 205: Why only use 30 nests for calculating this metric? How did you choose those 30 nests?

#32- After measuring 30 nests, and due to the small deviation achieved between data (0.09 m), we considered that 30 was a sufficient sample for this descriptive parameter. The objective of this estimate was only to show the reader nest size (0.6 m) in relation to spatial resolution of the image (0.013 m). Furthermore, nest size is within the range known for this species (0.3-0.7 m, Cramp, S. & Simmons, K.E.L. 1977). Nevertheless, as Purple heron nests showed higher variability in size, we have increase the sample up to 50, with similar results.

 

#33- line 212: you should include units in the table

#33- We have included units in Table 1 and 2

 

#34- line 216: You say at line 160 that you don't try to differentiate the two species. How exactly were you expecting the purple heron nests to be classified?

#34- The main problem during the classification was not in the Purple heron individuals, but in the Glossy ibis. When we performed a classification with four training classes (Glossy ibis, Purple heron, Lemna areas and white areas around nests), the algorithm confused the grey colors of Glossy ibis with Purple heron individuals. Consequently, ibis individuals appear as part classified as ibis and part as a heron. That’s why we decided not to include Purple heron individuals in the classification, because they share spectral characteristics with Glossy ibis. We have specified this in lines 205-207.

 

#35- Line 237: I would add nuances to this statements. You think the breeding estimation was good because of the image quality, and it's fair to think that. However, you cannot state it as a fact until you have compared it with other methods, like ground-based counts.

#35- Yes, ground-truth counts would be ideal, but were not possible to accomplish. The colony is located in a flooded area of 40-50 cm of depth, with a swampy terrain, with dense and high vegetation between nests. It is not possible to census the nests without generating a great disturbance to the individuals of the colony and even the nests. That’s why we have tried to count the nests by the aerial census.

 

Author Response File: Author Response.doc