Salamander Demography at Isolated Wetlands within Mature and Regenerating Forests
Round 1
Reviewer 1 Report
Church and coauthors describe the demography (which includes dispersal) of salamander populations in natural regenerating (after clear cut) forests. They argue that habitat quality is lower in regenerating forest than in mature forest.
I enjoyed reading the manuscript. It is interesting and well written. Methods and conclusions are sound.
My main comment is related to dispersal. I liked the result that salamanders preferentially move towards mature forests. This nicely corroborates previous results on “informed” dispersal. Amphibians disperse towards better habitat. For a review, see Cayuela et al. 2020 Quarterly Review of Biology. There are two recent empirical studies which may be of interest. Cayuela et al. 2018 Journal of Animal Ecology showed that newts leave low-density ponds and disperse to high density ponds. Unglaub et al. 2021 Molecular Ecology showed that newts preferentially disperse to ponds with high habitat quality. I think it would be important to discuss these ideas of “informed” dispersal in the manuscript.
It was interesting to see that there were differences among the three populations. Variation in the demography among seems to be common in amphibians (see e.g. Muths et al. 2017 Scientific Reports) and the variation cannot always be explained. This clearly limits our ability to generalize.
Other comments.
Line 47. Petrovan and Schmidt 2016 Plos One report declines of common toads in Europe.
Line 103 (or elsewhere). How far does the species disperse? What is know about dispersal distances?
The inset in figure 1 which shows the model should be a figure on its own. The model is quite complex. It’s not easy to understand the model and all the states and transitions. For example, it would be nice to show which transitions/movements are “local” and which ones are “landscape”.
Figure 1 (the aerial photo) needs a scale bar such that readers know the distances between ponds.
Line 126. Please provide more detail on the drift fence. There is some information in Church et al. 2007 Ecology.
Line 139. Looks like nice software!
Line 195. How can local movements be estimated? I understand the notation and which type of movement it is but I didn’t understand what kind of information in the data allows to estimate this type of movement.
Line 243. The probability of successful matches was 0.96. How does this affect the construction of capture histories and parameter estimates?
Line 293. A direct comparison of breeding season survival and non-breeding season survival is difficult because the time periods are different. Could you provide monthly survival? This would facilitate comparisons. Male breeding survival (1 month) ranges from 0.87 to 1. Male non-breeding survival ranges from 0.52 to 1. The eleventh root of 0.52 is 0.94 which may suggest that non-breeding survival could be slightly higher than breeding survival.
Figure 3. Please shift symbols horizontally to increase readability. Some estimates are hard to read (e.g., the ones for 2001-2002 NB).
Figure 4. “Breeding probabilities for … nonbreeders” sounds awkward. Consider rewording.
Line 444-445. What does “the interaction between salamander demography and forest structure remains dynamic” mean?
Line 455. Same question as above. Are dispersal distances known?
Line 485. See Petrovan and Schmidt 2019 Biological Conservation for a review on the importance of the juvenile stage.
Line 555. Syntopic or sympatric?
Line 641. Where will the data be deposited? (Replace XXX)
Author Response
Reviewer #1 Comments
Church and coauthors describe the demography (which includes dispersal) of salamander populations in natural regenerating (after clear cut) forests. They argue that habitat quality is lower in regenerating forest than in mature forest. I enjoyed reading the manuscript. It is interesting and well written. Methods and conclusions are sound.
My main comment is related to dispersal. I liked the result that salamanders preferentially move towards mature forests. This nicely corroborates previous results on “informed” dispersal. Amphibians disperse towards better habitat. For a review, see Cayuela et al. 2020 Quarterly Review of Biology. There are two recent empirical studies which may be of interest. Cayuela et al. 2018 Journal of Animal Ecology showed that newts leave low-density ponds and disperse to high density ponds. Unglaub et al. 2021 Molecular Ecology showed that newts preferentially disperse to ponds with high habitat quality. I think it would be important to discuss these ideas of “informed” dispersal in the manuscript.
Authors Response (AR): We thank the reviewer for these references and have integrated them, and the ideas of ‘informed’ dispersal, into our manuscript. See lines 76-77; 519-520; 645-647; 698-700 and 710. We were unable to find the Journal of Animal Ecology (2018) paper involving newts, but cited Cayuela et al. 2018 Oecologia instead as it seemed to reflect the work the reviewer was highlighting. Our manuscript is not focuses on dispersal, though it is certainly a component of our findings, but we did mention the idea of informed movements and dispersal to make the connection to the vast body of contemporary work by Cayuela and colleagues.
It was interesting to see that there were differences among the three populations. Variation in the demography among seems to be common in amphibians (see e.g. Muths et al. 2017 Scientific Reports) and the variation cannot always be explained. This clearly limits our ability to generalize.
AR: We completely agree!
Other comments.
Line 47. Petrovan and Schmidt 2016 Plos One report declines of common toads in Europe.
AR: We have included this reference.
Line 103 (or elsewhere). How far does the species disperse? What is know about dispersal distances?
AR: We have included information about published dispersal distances for this species (Gamble et al. 2007). See lines 106-112.
The inset in figure 1 which shows the model should be a figure on its own. The model is quite complex. It’s not easy to understand the model and all the states and transitions. For example, it would be nice to show which transitions/movements are “local” and which ones are “landscape”.
AR: We have removed the inset from Figure 1 and created a new figure (now Figure 3 – lines 210-215) that specifically denotes: (1) ‘local’ movements, both within the pond basin and outside the pond basin – the later have been termed migrations by some authors (Semlitsch 2008), and (2) ‘landscape’ movements or breeding dispersal > 0.00 (Semlitsch 2008, Cayuela et al. 2020).
Figure 1 (the aerial photo) needs a scale bar such that readers know the distances between ponds.
AR: Distances between ponds have been added in the figure legend. Exact distances were given because the low angle of the photo would distort a scale bar. See lines 133-135.
Line 126. Please provide more detail on the drift fence. There is some information in Church et al. 2007 Ecology.
AR: We have added more information on the drift fence capture methods (lines 136-142).
Line 139. Looks like nice software!
AR: Yes, it was invaluable for our study.
Line 195. How can local movements be estimated? I understand the notation and which type of movement it is but I didn’t understand what kind of information in the data allows to estimate this type of movement.
AR: We have revised the manuscript to include an additional figure (Figure 3) to clarify our ‘local’ movements and their respective timing as it relates to the ‘local’ movements estimated via our modified robust-design multistate model. Our study contains 4 primary periods, each with 2 secondary sessions (T=8 occasions). Our model is a modified robust-design because mortality and/or movement between selected observable states is allowed within the primary period. Stated differently, we can estimate survival and movement probabilities between habitat states for breeders that are entering and exiting the pond, within the primary period (i.e., between the two sessions in each primary period). The drift fences are monitoring daily during the breeding season each year. Adult breeders migrating to the pond basin are captured by the drift fence and collected in pitfall traps. Each captured individual was digitally photographed and assigned a state based on the habitat associated with the pitfall trap they were captured in (i.e., mature or regenerating forest). Each captured individual is then released on the opposite side of the fence and proceeds to the pond. Males generally leave the pond basin after breeding, but females will remain at the nest until late fall or winter rains fill the pond and flood the nest. As these individual exit the pond basin they are again captured by the drift fence and collected in pitfall traps on the inside of the fence. Each captured individual was digitally photographed and assigned a state based on the habitat associated with habitat they were exiting into (i.e., mature or regenerating forest) based on the location of the pitfall trap they were captured in. We have included additional information on this process and the drift fence as requested by both reviewers (lines 226-228).
Capture histories, created from the detection data described above, include 8 encounter occasions and 12 states (Figure 3). Within primary periods, we only estimate ‘local’ movement probabilities that occur within the pond basin between two observable states at each pond (i.e., transition probabilities denoted with dotted lines in Figure 3). Between primary periods, i.e., between breeding seasons, all movements and transitions between breeding and non-breeding (unobservable) states shown in Figure 3 are estimable from our data. We verified parameter identifiability in our models using two numerical methods - an analytic-numeric method (Burnham et al. 1987, pp. 214–217, 292–295, referred to as the simulation method in Gimenez et al. 2004) and a numerical version of the formal derivative matrix, termed the numerical Catchpole Morgan Freeman (numerical CMF),included in program M-SURGE (Choquet et al. 2005). We added a sentence at the end of section 2.3 to inform readers of this process: “We verified that all parameters in the models described below were identifiable using two numeric methods (see Bailey et al. 2009, 2010 for method details).” Bailey et al. 2009 described the process for one pond, but we verified that all parameters in our models were identificable for our three pond system using the same numeric methods. Hopefully, this answers the reviewer’s question.
Line 243. The probability of successful matches was 0.96. How does this affect the construction of capture histories and parameter estimates?
AR: It does not affect the construction of the capture histories. We conducted the small test of the matching process using a set of 500 individuals just to verify that failing to identify a previously captured individual using our software was very low. Our probability of successful match represents a lower bound and only applies to individuals that were previously captured only once; most previously captured individuals in our study had more than one photograph in the library after the first year, given the high capture probabilities.
Line 293. A direct comparison of breeding season survival and non-breeding season survival is difficult because the time periods are different. Could you provide monthly survival? This would facilitate comparisons. Male breeding survival (1 month) ranges from 0.87 to 1. Male non-breeding survival ranges from 0.52 to 1. The eleventh root of 0.52 is 0.94 which may suggest that non-breeding survival could be slightly higher than breeding survival.
AR: The focus of our study was on comparing survival probabilities between mature and regenerating habitat during the breeding and nonbreeding season, so we did not include these monthly estimates in the text or figures of the manuscript, but we did add provide monthly estimates to Appendix C using the average residence time reported in Appendix A for each pond, year, and sex.
Figure 3. Please shift symbols horizontally to increase readability. Some estimates are hard to read (e.g., the ones for 2001-2002 NB).
AR: This is now Figure 4 in the revised manuscript. We are not sure we understand the reviewers comment or what symbols to shift. We feel that the figure is clear in the formatted version provided by the journal and the exact survival estimates are also given in Table C.1 in Appendix C.
Figure 4. “Breeding probabilities for … nonbreeders” sounds awkward. Consider rewording.
AR: This is now Figure 5. We have revised this figure legend to read: ‘Breeding probabilities for individuals that breed the previous year (B) and those that skipped one or more breeding opportunities (nonbreeders, NB)….’ See lines 447-448.
Line 444-445. What does “the interaction between salamander demography and forest structure remains dynamic” mean?
AR: We removed the phrase ‘the interaction between’ from this sentence. See lines 521-522.
Line 455. Same question as above. Are dispersal distances known?
AR: See our response to this question above – i.e., response to Reviewer 1 Line 103 comment. We have included information about published dispersal distances for this species (Gamble et al. 2007). See lines 106-112
Line 485. See Petrovan and Schmidt 2019 Biological Conservation for a review on the importance of the juvenile stage.
AR: We added this citation to support this sentence. See lines 567-568.
Line 555. Syntopic or sympatric?
AR: Syntopic has replaced sympatric throughout the manuscript now.
Line 641. Where will the data be deposited? (Replace XXX)
AR: We have now included the mark-recaptured data as supplemental material. See lines738-739.
Reviewer 2 Report
This is a solid analysis of a large high-quality data set from a recapture study. The presentation of the results is clear and well structured.
The description of the methods, however, is scanty. Drift fences with pitfall traps were used to capture salamanders in four consecutive years. All reported results refer to adults. It is unclear whether juveniles were not recorded at all, or not used in this analysis because the color pattern is not complete when the metamorphs leave the ponds, or for some other reason. Juveniles and subadults certainly are important stages in demography. The paper could benefit from including more background information on the ecology of the studied species, and from providing more details on the field methods.
Discussion and conclusions are focused on comparing mature with regenerating forest. I wonder why the contrast of broadleaf versus coniferous forest is not touched at all. With the pine plantations mentioned in the text (lines 112f) and visible in Figure 1 one might expect that these habitats do not recover towards their previous state during succession after logging. Changes in leaf litter composition may have pervasive consequences for soil structure and biota; thus ecological characteristics in these regrowth areas perhaps diverge from states in the mature forest, rather than converging on them.
Only five of the 46 cited papers were published in the last ten years. It may improve the discussion to compare the findings of this work with recent results and syntheses related to the questions studied here. There has been some progress in demography and spatial ecology of amphibians in the last decade.
Author Response
Reviewer #2 Comments
This is a solid analysis of a large high-quality data set from a recapture study. The presentation of the results is clear and well structured.
The description of the methods, however, is scanty. Drift fences with pitfall traps were used to capture salamanders in four consecutive years. All reported results refer to adults. It is unclear whether juveniles were not recorded at all, or not used in this analysis because the color pattern is not complete when the metamorphs leave the ponds, or for some other reason. Juveniles and subadults certainly are important stages in demography. The paper could benefit from including more background information on the ecology of the studied species, and from providing more details on the field methods.
AR: We have now included more information on the species ecology (lines 102-113) and drift fence capture methods (lines 136-142). Also see related responses to Reviewer #1 comments above regarding dispersal distances. Juveniles (metamorphs) were captured at the drift fences but were not included in this analysis until they returned to breed as adults – we include this information in lines 181-186.
Discussion and conclusions are focused on comparing mature with regenerating forest. I wonder why the contrast of broadleaf versus coniferous forest is not touched at all. With the pine plantations mentioned in the text (lines 112f) and visible in Figure 1 one might expect that these habitats do not recover towards their previous state during succession after logging. Changes in leaf litter composition may have pervasive consequences for soil structure and biota; thus ecological characteristics in these regrowth areas perhaps diverge from states in the mature forest, rather than converging on them.
AR: We added a sentence to the Discussion acknowledging that the planted pines in the regenerating forest may change leaf litter composition and the terrestrial invertebrate prey abundance (Cifuentes-Croquevielle et al. 2020). See lines 596-598.
Only five of the 46 cited papers were published in the last ten years. It may improve the discussion to compare the findings of this work with recent results and syntheses related to the questions studied here. There has been some progress in demography and spatial ecology of amphibians in the last decade.
AR: We have updated the references with the papers recommended by Reviewer #1 and several other recent papers related to the questions studied here. See additional papers in References section.
