the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Oct 2023
Linking biodiversity and geodiversity: Arctic-nesting birds select refuges generated by permafrost degradation
Madeleine-Zoé Corbeil-Robitaille
Éliane Duchesne
Daniel Fortier
Christophe Kinnard
Joël Bêty
Abstract. To gain better insight into the cascading impact of warming-induced changes in the physical landscape on biodiversity, it is crucial to establish stronger links between abiotic and ecological processes governing species distribution. Abiotic processes shaping the physical characteristics of the environment could significantly influence predator movements in the landscape and ultimately affect biodiversity through interspecific interactions. In the Arctic tundra, the main terrestrial predator (Arctic fox) avoids patches of wetlands composed of ponds with islets that can act as refuges for prey. Little is known about the geomorphological processes generating islets selected by prey species. Our study aimed to identify i) the physical characteristics of islets selected by Arctic-nesting birds and ii) the geomorphological processes generating islets available in the landscape. Over two breeding seasons, we determined the occurrence of nesting birds (Glaucous gull, Cackling goose, Red-throated loon) on islets (N=396) found over a 150 km2 area on Bylot Island (Nunavut, Canada). Occupied islets were located further away from the shore (10.6 m ± 7.3 vs 7.4 m ± 6.8) and surrounded by deeper water (33.6 cm ± 10.6 vs 28.1 cm ± 11.5). As expected, all three bird species selected islets less accessible to Arctic foxes, with nesting occurrence increasing (linearly or nonlinearly) with distance to shore and/or water depth around islets. Based on high-resolution satellite image and field observations, we found that ice-wedge polygon degradation generated the majority of islets (71 %) found in the landscape. Those islets were on average farther from the shore and surrounded by deeper water than those generated by other processes. As polygon degradation is projected to accelerate in response to warming, new refuges will likely emerge in the Arctic landscape, but current refuges could also disappear. Changes in the rate of polygon degradation may thus affect Arctic tundra biodiversity by altering predator-prey interactions.
- Preprint
(3993 KB) - Metadata XML
- BibTeX
- EndNote
Madeleine-Zoé Corbeil-Robitaille et al.
Status: open (until 11 Dec 2023)
-
RC1: 'Comment on egusphere-2023-2240', Anonymous Referee #1, 07 Nov 2023
reply
General comments
A very nice article reflecting both interesting and hitherto not fully explored topic. The manuscript is overall well written and interesting, and I did not find any major flaws or inconsistencies. I have few comments for your consideration.
First of all, I especially appreciate extensive fieldwork behind this manuscript. Empirical data is very much needed to gain knowledge of our Arctic landscapes. Well done with that!
Specific comments
Main issue is about the use of the term geodiversity in the Title, Introduction, and Discussion and whether it is used clearly. Geodiversity consists of geological, geomorphological, and hydrological variation of the earth’s surface and subsurface (Gray 2013). I think you should sharpen the message of the manuscript especially in aforementioned sections as you do not assess or use the geodiversity or biodiversity through species richness or georichness, respectively, but rather have a case study of how certain aspects or features of geodiversity (here polygon degradation, glacial boulders, and raised beach crest degradation) are linked to arctic-nesting birds. So, I would see your approach to geodiversity is qualitative, through certain geomorphological features or landforms like f.e. Tukiainen et al. 2019 has done in the Journal of Biogeography.
The first paragraph (Starting from L 34), is about geodiversity and its relevance to the living world. Firstly, please add of what things geodiversity consists of (See Gray 2013). In addition, it should be initialized what kind of approach this manuscript is taking, that isa the qualitative approach to geodiversity-biodiversity relationships.On page 13 L 326 you describe what you have done: linking geomorphological processes and wildlife micro-habitat selection. I would reconsider the title of the manuscript to better fit with the contents of the manuscript f.e. by dropping off the holistic terms geodiversity and biodiversity and adding something more specific f.e. “linking geomorphological processes and wildlife micro-habitat selection”. Geodiversity would fit greatly into keywords of this manuscript.
In Table 1: To emphasize geodiversity, please specify which islets are considered as a part of geodiversity and what is not (biotic process one).
I think these results contribute to our knowledge about Arctic environment and these kinds of studies that bring empirical evidence about the relationship between abiotic and biotic nature are very much needed.
Did you consider adding any other variables into your analyses?Technical corrections
Like said earlier, I find the text easy to follow for a reader not so familiar with birds and I didn’t spot any grammatical errors.
In the appendix D. please present each species’ tables systematically in the same order than in the manuscript figure 4. (1st Glaucous gull, 2nd Cackling goose and 3rd Red-throated loon).
Citation: https://doi.org/10.5194/egusphere-2023-2240-RC1 -
RC2: 'Comment on egusphere-2023-2240', Daniel Ruthrauff, 15 Nov 2023
reply
General comments:
The paper assesses the physical characteristics of small islets in Arctic environments. These islets often serve as nesting platforms for birds, and the authors nicely document the physical characteristics of islands that were selected for nesting compared to unoccupied islands. In a helpful subsequent step, the authors next characterized the geomorphological process by which each islet was formed, finding that ice-wedge polygonal degradation was the primary genesis of islets at their study site. Together these assessments provide a useful overview of the factors that promote the formation of islets and their occupancy by nesting bird. The figures are informative and easy to understand, and the authors employed appropriate analytical approaches to address their study questions. The paper was well written and interesting to read, and the authors provide good context for their findings and discuss the role of climate change in future creation and degradation of Arctic islets. I had very minor suggestions on rewording, syntax, etc., but more substantive suggestions for the authors on ways to improve their analysis of physical factors that promote the occupancy of islets. I hope the authors find my comments to be helpful.
Dan Ruthrauff
US Geological Survey Alaska Science Center
druthrauff@usgs.gov
Specific comments:
In general, your methods and analyses are appropriate for your questions and are clearly presented. I do think, however, that your manuscript would benefit from a more straightforward analytical approach regarding your assessment of factors that promote islet occupancy. For this, you essentially have two model sets, one including measures of islet area and lake area, and one without. You go to a lot of trouble to show results from both sets, which I found a bit confusing…but ultimately base your inference on the model set without measures of area as covariates. You state that your findings regading Distance and Depth do not change with Area as a covariate…which to me begs the question of why you then bother excluding Area? You state that this was due to sample size concerns (n=315 islets with all measures, n=350 with distance and depth), but both model sets employ pretty robust sample sizes. Since islet and lake area seem like biologically relevant covariates, I’d just stick with your ‘larger’ analysis, and not re-run models with area removed. Having the two model sets creates confusion between Table 2 and Appendix 3. I also had some questions about the models in your model set. Unless I’m mistaken, you did not create models that did not include either DISTANCE or DEPTH as a covariate (except for the null model). Having models in your model set with only IsletArea, LakeArea, and an additive model using these same covariates would help better assess the influence of the areal measures on islet occupancy.
Also, you say that models within ≤2 ΔAICc were ‘considered’, but there’s no sign what this actually means. You present only the parameter estimates from your top model, so it doesn’t look like other well-supported models were considered. Given that most of your outputs had pretty equivocal model support, I think you should consider model averaging to estimate parameters. Selecting only the top model is generally not well supported, especially when there’s high model uncertainty.
Technical corrections:
Line 20: what is journal format regarding adding genus and species names at first mention of a species?
Line 25: I assume these are SDs? Please indicate.
Line 27: ‘linearly or nonlinearly’ is confusing; as the reader does not yet know about your distance-weighted function, I’d just drop ‘linearly or nonlinearly’ from the abstract. The truth of the statement remains intact. Also, ‘and/or’ is more clearly just ‘and’.
Line 27: ‘image’ to ‘imagery’
Line 40: changing ‘with’ to ‘and’ would make it clearer that these are the two components being connected
Line 42: consider deleting ‘Nowadays’; unnecessary and a bit vernacular/informal
Line 45: ‘precipitations’ should be singular, ‘precipitation’
Line 52: Caro missing year?
Line 70: colon missing after ‘islets’; maybe replace semi-colons between species with commas.
Line 73: ‘image’ to ‘imagery’
Line 92: maybe ‘polygonal wetland complexes’ is clearer
Line 95: reword to ‘essentially nest only on small islets’
Line 95: consistency with how species names are presented. I see ‘glaucous gull’, ‘Glaucous gull’, and ‘Glaucous Gull’, for instance.
Line 98: for ‘jaegers’, use (Stercorarius sp.); parasitic jaeger is S. parasiticus, not S. parasitus.
Line 104: change ‘carried’ to ‘conducted’
Line 144: not sure you need to mention that you didn’t fit random effects…I guess I only mention it if I do fit random effects.
Line 153: change ‘lesser’ to ‘less’. Also, I’ll await results, but when you say models with deltaAICc ≤2 were considered, how did you consider them? Model averaging? OK, having read more thoroughly, it seems that you only show parameter estimates from your top model (Table 2). In this sense, I’m not sure how you ‘considered’ the other models? A real advantage of AIC modeling is the ability to conduct model averaging for drawing inference; generally, drawing model inference from the best-supported model alone is poorly supported, unless it has overwhelming model weight (which yours do not).
Line 160: so, this is a bit unclear. You present ‘full’ model results in Appendix D, but here state that you removed LakeArea and IsletArea due to missing data. One idea to consider is that if you didn’t include these two measures of area in your final modeling, then you should not mention them at all in the paper. Alternatively, despite the smaller sample sizes, since your model results don’t differ when you do include these area-related variables, I’d probably just keep them in the paper—these seem like biologically relevant measures, even if not collected at all sites. Readers like me would probably wonder about the effects of the lake size and islet size. As it stands, you introduce them and then remove them. I’d advocate for just including them so you can more fully discuss them. But, note that due to the removal of area measures, Appendix D is not really comparable at all to results in Table 2. The AICc values and weights are totally different…so, it’s really an apples-to-oranges comparison to have both. They are different model sets, and not comparable; you sort of walk a middle path between the two sets, which I found confusing. I think it would be clearer were you to base all your results on the ‘full’ results from Appendix D rather than the subset in Table 2.
Line 183: ‘best fitted’ implies some measure of actual fit…so I prefer to use terms like ‘best supported’ in AIC modeling frameworks. This terms does not imply that the model is actually ‘good’, only that it’s the best supported—it’s a more neutral way to frame it.
Lines 188-190: see comments above re. including area. N = 315 is still a pretty robust sample. So, including lake and islet area didn’t really change the relationship between nest occurrence and distance and depth…but what were the relationships to area? As I mention above, I think you’ve got a nice sample size, and restricting your analysis to only islets where you had distance and depth gains you n=35, right (315 v. 350). I’d keep area in your models and discuss this effect. Also, it looks like you didn’t include any models in your model set that did not include either distance or depth (other than your null model and a spatial model)? Why did you not include models with IsletArea, LakeArea, and IsletArea + LakeArea (ie, Areas) alone as models? Seems you haven’t really assessed the influence of area without such models. I see on line 380 you summarize these results (occupied islets tend to have greater IsletArea and LakeArea than unoccupied), but this is not in the main results. I’m also confused why results in Table 2 don’t mimic those in Appendix Dx.2? For instance, in Table 2 for CACG you show DISTANCE* + DEPTH (w=0.2) and DEPTH (w=0.26). In D1.2, which should be the same as what’s presented in Table 2, you show the 2 aforementioned models but also two more models within deltaAICc of 2. Why were the other two models in D1.2 (DISTANCE + DEPTH, DISTANCE* + DEPTH*) not shown in Table 2?
Line 200: nice figure! This clearly shows the relationship between depth and distance across used sites for each species. Also, you previously ordered species in results as CACG, GLGU, RTLO, but here it’s GLGU, CACG, RTLO, might swap them around to maintain order throughout.
Line 240: nest site selectin by loons varies by loon species. In Alaska at least, RTLO breed on small ponds not otherwise occupied by PALO or YBLO. These ponds typically freeze deeply in the winter, so RTLO typically feed in the marine environment. PALO and YBLO, in contrast, nest on deeper lakes with more abundant food resources. Most chick provisioning occurs from within the nest lake itself for PALO and YBLO. So, for ‘loons’, food availability is also a factor in site selection. For RLTOs specifically, this is probably not the case, so you may want to explicitly state ‘red-throated loons’ here instead of ‘loons’ more generally.
Line 258: maybe ‘As with other mammalian mesopredators, Arctic foxes are generally reluctant to swim…’ Reads awkwardly as currently worded.
Line 260: change ‘generate’ to ‘generates’
Line 333: this reads as if the primary way that climate change alters predator-prey interactions and the occurrence and distributions of species in the Arctic is via influencing refuge availability through ice-wedge degradation. Of course, climate change is rapidly and markedly changing predator-prey interactions and the occurrence and distributions of species in the Arctic…but via a multitude of mechanisms, not just ice-wedge degradation. Consider rewording: you’re really just trying to say that climate change is irrevocably altering Arctic ecosystems, and the role of climate change effects on ice-wedge degradation and how this relates to predator-free nest sites had been previously little studied.
Line 384: ‘It seemed adequate to work this way with our variables’ is a bit underwhelming. Maybe restate this more positively and assuredly? ‘Based on our hypotheses about the effects of distance from shore and islet depth on site use, distance weighted functions provided an appropriate model framework for our data structure’ or something similar.
Line 422: as stated previously, there’s not much support for just presenting the results of the model with the smallest AICc, especially when the model support is pretty equivocal. Why do you not model average? I think you’ll need to state why your opted not to implement model averaging.
Appendix E2: this is interesting! I’d love to know about successful vs. unsuccessful nests. At our study site in northern Alaska, the depth of the water around the islets is much greater, but almost without fail these deep-water islands are visiting by swimming foxes who depredate all the nests on the islands (typically black brant nests). In comparison, the depths you measured at your site are quite shallow compared to the sites that I’m familiar with in Alaska.
Citation: https://doi.org/10.5194/egusphere-2023-2240-RC2 -
RC3: 'Comment on egusphere-2023-2240', Anonymous Referee #3, 08 Dec 2023
reply
Dear authors,
It was interesting to read your study, which has an interesting view angle (relation between both types of diversity) and I find well written. I have several main points and more detailed point are given below. (1) To me it is unclear to what extent polygon degradation is a cyclical process (as you say. A long-term process) or sped up by climate change (a short term process via permafrost breakdown). You start from the climate change perspective, but the time scale and relative contribution of processes is unclear. Related to this, if the process of permafrost/polygon degradation is warming-induced, what was then the historic distribution/habitat choice of the species in the landscape? Was it forced to nest in more accessible locations in the past due to absence of distant islets? Is that not the case anymore now? (2) please consider the issue that you seemed to be unable to include failed nests in your analysis. This has important effects on the conclusions that you can draw. (3) please also consider possible effects of bird species on each other.
Best wishes.
TITLE
It is not clear that they select them for nesting.
Also, do you mean permafrost degradation, or polygon degradation? What have you shown?
Also, your article is less concretely linking both types of diversity. You study where birds nest. Maybe the reader expects analysis of a correlation between both types of diversity (comparing areas).
ABSTRACT
- 17: establish stronger = better understand (see also L. 39).
- 25: … than unoccupied islets?
INTRODUCTION
- 34: increasingly gaining recognition as … pivotal = can be removed in my opinion
- 70: characteristics affect nest site selection: this sounds double.
METHODS:
- 97-99: this is about the historic situation, but this may change if the prey nests more and more on islands, where foxes cannot predate them. Any data / discussion on that?
- 119-122: it is important to know whether you were able to detect failed nests. If not, then your analysis of the nest site selection of prey bird species, may in fact not show prey nest site selection, but predator predation success: they removed all nests on easily accessible locations, which led you to conclude that prey only nests on safe locations far away. The subsetting to active nests is tricky in your analysis.
- 119-122: also, how were they observed? How closely did you inspect the islets? Did you step on them to check for sure?
- 119-122: what about a situation where you can have species A in year 1, species B in year 2?
Table 1: since you introduce this topic from the climate change background, it would be good to know which processes are affected by climatic warming. Only 1-2?
Table 1: can the 2 processes in category 6 be distinguished?
Table 1: can you give an indication of the timescale of the processes?
- 133: what about a situation where an islet is in fact the result of a combination of processes? I would guess that is quite common?
- 134: how reliable is that method? Do you have sources/references? Was that validated by soil analysis? Was the categorisation done by one same person? Did you test how consistent the categorisation method is by letting multiple people assign them and see if they come up with the same result? Was the classification done without prior knowledge of where birds nested?
- 144: When analysing selection, you assume that birds have something to choose. Is that the case? Is the area not saturated? How strong is the competition for good islets?
- 151: explain that only additive models were included. Why not interaction effects?
- 156: including just lat+lon is a simple method, which I think is not accurate and does not account for clustering (especially if more clusters occur). It only works if there is 1 cluster which is in the extreme of lon/lat. It does not work when there are multiple clusters spread out over the map, or 1 in the middle of the map. Can you think of a better method, e.g. internest-distances compared to a random situation, or presence of hetero-/conspecifics around compared to a random situation?
RESULTS:
- 166: >41: why no exact value?
- 167: 84 islets were occupied, but what if species A in year 1, species B in year 2?
- 169: why not compare them with unoccupied islets? (instead of all available)
Figure 3: the categories in panel b combine 2 variables with parallel gradients. However, in L. 116 it is said that the variables are not correlated. This is confusing and I cannot evaluate whether the categories in panel b make sense (they would only when there is a correlation between these 2 variables).
- 188 (and also L. 160): should be in methods. The cases with missing data are confusing: you should just be able to look at a satellite image. Or were those island too small? Then you have missing data especially for 1 category (i.e. very small islands, vegetation aggradation/loon).
- 190: It would be good if you could also test the effect of species on each other. If one present, than others around? Hetero/conspecific clustering? And can you describe your data/situation better? I.e. are there more nests/islets in the same pond? (maybe for gulls/geese that is true but not loons, which forage in the pond). How many? Do different species occur in the same pond? Can you give some data about this? Inter-nest distances between con/heterospecifics? Never multiple nests on one islet? Data? Which species?
- 197 (Table 2): so the distance-weighing is because the difference between 101 and 110 m is hypothesised to be less important for the bird than the difference between 1 and 10 meter? Is it not better to log-transform the units distance and depth? Then, it is easier to compare/interpret the effect sizes between species when you have the model output.
- 211-213: to methods.
DISCUSSION:
- 223: coexistence of which species? Predator and prey?
- 225: is = was.
- 227: 3 species: but do they breed at the same time? Do they compete and does one species displace the other? In what order do they arrive and breed, and displace each other?
- 230: Do those islets also become snowfree earliest? Were your years 2018-2019 late or early snowmelt years? Same about the predation: Were your years 2018-2019 low or high predation years? And in the quantitative sense, what was the predation pressure (e.g. foxes per square kilometer)?
- 232: what did that study do?
- 238: very few = how many exactly?
- 241: what about food availability?
- 243: but how much interannual variation is there? And how large could the effect possibly be?
- 246: 3rd and 4th scale = ? I am not familiar with this formulation?
- 248: main nest predator: this is questionable, because your birds nest on areas where foxes cannot predate. Any data on nest success and causes of failure for your birds?
- 248: Tundra species = tundra bird species.
- 251: how would the reduced predator abundance come about?
- 252: quality of islets should thus be based on = unclear.
- 254 (and elsewhere): references should be ordered from old to new, I believe.
- 261: if you make this statement about maximum jump and leg length, can you also give the exact value of what would be too far or too deep? Otherwise, you should stress that this is a hypothesis.
- 264: should be = is.
- 273: cyclical process: please explain, and what is the duration of one cycle?
- 275: you did not show the origin, but assumed/assigned the origin based on visual characteristics. How reliable was that method? (see also L. 134).
- 284: some of them = how many exactly?
- 294: explain = explains.
- 299: should be = is it not possible to test this? And could you say ‘were’?
- 304: various scales = do you mean temporal and spatial? Or different values of those? Please make explicit.
- 313: contingent upon … areas = this is quite cryptically formulated. Can you reformulate it?
- 321: see general comment about the apparent mixing of long-term and short term (climate change) processes in degradation of polygons.
- 329: why mention trait?
- 333-334: but how? How does climate change affect this interaction? Temperature goes up, permafrost disappears, polygons may break down, which could first make more islets, but later they may also break down? Furthermore, climate change has also other effects (earlier snowmelt, which increases breeding propensity and success in geese, but potentially has contrasting mismatch effect on chick growth).
Appendix A
“really small”: please quantify.
“close to shore and in shallow ponds (bottom visible)” = this goes towards circular reasoning. This should be a result of your survey, not a characteristic how to recognise islet types.
Appendix B
Specify DIST to what, to nearest shore?
Specify DEPTH where, maximum depth between islet and closest shore?
Specify that IsletArea and LakeArea were satellite-derived.
- 378-380: this piece of text looks lost. It would fit in the methods (1st sentence) resp. results.
Appendix C
- 383: declining: specify function, a negative exponential?
- 386: effect of what, on what?
Please also see my comment on L. 197.
- 398: surfaces = surface.
Appendix E
“occupied by a nesting bird”: again, how to deal with multiple birds, or spec A in year 1, spec B in year 2? (see also comment about L. 119-122).
Appendix F
It is not clear to me what is the added value of this figure compared with Figure 3.
Appendix G
Indicate time frame of this process (and effect of climate change on that?).
Citation: https://doi.org/10.5194/egusphere-2023-2240-RC3
Madeleine-Zoé Corbeil-Robitaille et al.
Data sets
Dataset_islets_bylotisland Madeleine-Zoé Corbeil-Robitaille https://doi.org/10.5281/zenodo.8395558
Madeleine-Zoé Corbeil-Robitaille et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 185 | 68 | 14 | 267 | 8 | 7 |
- HTML: 185
- PDF: 68
- XML: 14
- Total: 267
- BibTeX: 8
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1