To increase transparency, PeerJ operates a system of 'optional signed reviews and history'. This takes two forms: (1) peer reviewers are encouraged, but not required, to provide their names (if they do so, then their profile page records the articles they have reviewed), and (2) authors are given the option of reproducing their entire peer review history alongside their published article (in which case the complete peer review process is provided, including revisions, rebuttal letters and editor decision letters).
The reviewers were satisfied with your revisions, as was I. Please do consider the comments of reviewer 4 which will improve the paper.
No additional comments
No additional comments
No additional comments
The authors have addressed all of my concerns either through changes to the manuscript or clarification in their rebuttal. I recommend that the article is accepted for publication.
Line 275: I still have a few issues with the implication that chum and pink fisheries are less important to commercial fishers because of their lower price. Yes, pinks and chums are not worth much per pound, but they are the most abundant species of salmon—several times more abundant than Chinook. It’s the price per pound * harvestable biomass that determines the value of the fishery. Further these stocks could be of particular significance to humans for the same reason they are to wildlife, for example if they provide commercial fishing opportunities at unique times of the year, or allow fishers to mitigate for low returns of other salmon stocks.
Comments on Intro:
56: The body size argument isn’t very compelling to me, but that’s OK.
69: I still find this paragraph inaccurate. The paragraph implies that bears and eagles have fundamentally different tactics for exploiting salmon resource pulses. I would argue that they exhibit very similar behavior, just expressed at different spatial and temporal scales due to the constraints of their life-history and physiology. Bears track asynchronous salmon runs to build their fat stores. Suggesting they don’t need to track asynchronous resource pulses because they can store fat is inaccurate. This paragraph would be much clearer if it stated up front that several species exhibit behavioral tactics to exploit resources pulses, tracking asynchronous salmon runs to protract their access to this locally ephemeral food source (Ruff et al. 2010 Ecology, Schindler et al. 2013, additional eagle paper if there is one) then say that the spatial and temporal extent of this behavior depends on the species life-history/physiology/etc.
I don't understand why the importance of the salmon resource wave to bears is downplayed in the intro and emphasized in the discussion.
No additional comments.
I am satisfied with the changes and explanations provided. I think this article will make a valuable contribution. Very interesting. Recommend highly for publication.
Your paper got 3 detailed reviews from people that actively work on bear-salmon interactions using camera traps. A wide variety of concerns are listed in their reviews.
Reviewer 1 suggests the experimental design is fatally flawed, whereas reviewer 2 simply says the design is unclear. I agree with the latter opinion. I also agree with the very positive reviewer 3 who notes that your paper contains a lot of interesting information that is valuable owing to the fact that behavior of bears and other consumers of salmon are poorly known.
So, I require a major revision in which you deal with all of the reviewers' comments. You don't have to agree with every point but you have to say how you dealt with each comment in your rebuttal letter. That will be a rather big job but I think the paper is worth it.
There are a few parts of the manuscript which are unclear, or are missing some detail. Here are a few examples that I noted, with some suggestions:
32- A quibble- Maybe replace marine nutrients with “marine energy and nutrients.” The nutrients are important for vegetation, but most vertebrates are more limited by energy. This is why bears will excrete most of the nitrogen in their urine.
36- Another citation for the diversity of species used by salmon- Shardlow and Hyatt, 2013, Ecological Indicators
38 – …for mobile consumers that move among runs… is maybe more clear
42- Change dispersion to distribution.
58- Although not conclusive, Schindler et al. 2013 suggests that bears also migrate to track resource pulses
76-78- You should note that the tendency for bears to “high grade” and only eat 25% of carcasses only occurs when salmon are both abundant and easy to catch. If not, bears will tend to eat almost everything. This is an important distinction because there is likely a threshold of salmon abundance below which bears will not leave anything for scavengers.
128- weighed-> weighted
225- “integrate over..” is a bit confusing. Maybe something like “bears maximize their annual energy balance by depositing fat during salmon abundance and hibernating during food scarcity”
Fig. 2- Missing the letter G, so the labels don’t match the fig. legend
Other- the spatiotemporal pattern of salmon spawning is important to understanding this paper. It might be helpful to show this with a multi-panel map showing where salmon are available during multiple time periods (e.g., summer, fall, winter). This would highlight the relative scarcity of salmon in late fall/winter.
Although I appreciate the objective of this paper, I don't think that the methods employed allow you to make the conclusions you make. The most important comments that address these issues are below (151, 171).
125-132- You are still overestimating the number of individuals using these areas, unless you are assuming that bears (and the other species) don’t hang around in one spot for more than 2 minutes. This is unreasonable given past research on the length of bear fishing bouts (Gende and Quinn, 2004; mean fishing bout length of 46 min. for dominant bears). If a bear fished for this duration in a location where it will trigger your camera, you might conclude that 23 unique bears visited the sites. However, I don’t think this is a problem if you re-frame what you are measuring. Rather than call them “unique visits” instead treat your data as an index of use of the salmon resource. Thus, a bear that walks past your camera will trigger it once, but is unlikely to stick around if there are no salmon. I think it is safe to assume that the number of detections correlates with the amount of use, even if it is the same few individuals.
133- Were carcasses placed at each camera throughout the entire season, or just for the period corresponding to the nearby run?
142-144- Here, with only a single carcass, the five minute threshold may discriminate between unique groups of bears, but a raven or magpie might take a long time to consume a carcass.
151- These results and interpretation hinge on three important assumptions that are not mentioned in the methods. 1) each species is equally likely to trigger the infrared motion sensor on the cameras; 2) camera trigger rates (detection rates) for a given species are equal across sites; and 3) the viewsheds of each camera are equal. Unfortunately, I think all of these assumptions are violated by the methods used. Assumptions 1 and 2 - If species-specific trigger rates were the same across all sites (a raven was equally likely to trigger a camera at a stream than on a river flat) then the violation of this assumption would not prevent comparing the pattern of use of ravens across habitats. However, when detection varies by habitat differently for different animals, this is a problem. At close range I expect all species to have similar change of triggering the cameras. For this reason I think the stream data is likely the best for comparing among species. At longer distances, however, an eagle may be more likely to trigger the camera than a magpie, making comparisons between species at these sites less reliable. Assumption 3- While this assumption is violated, it does not change your results as long as you stick to comparing among species at a single site, as you do for ravens on line 159.
149- this variation could be because of variation in motion triggers because of different distances to cameras rather than variation in patterns of use.
171- In this paragraph you implicitly use the detection rates of your road-side cameras as estimates of relative abundance of several species within your study area. I don’t think this is valid; relative detection rates may vary as a function of differences in habitat selection among species. For example, black and brown bears may differentially avoid roads due to the risk of running into people. This would be better supported if you had a rough estimate of black vs. brown bear abundances for your study area.
Here are some specific notes.
42-43- Yes, integration of wildlife and salmon management is very important.
153-154 This is likely a function of availability. Even though bears are less efficient when fishing at lake shores, they will if those are the only salmon available. Was this the case? If not, subordinate bears or sows with cubs may choose a poorer fishing habitat to avoid intraspecific competition or infanticide (Ben-David et al. 2004).
163-164- Okay because you are making a relative comparison and you were detecting ravens throughout the period, and I’d expect ravens, gulls, magpies and stellar’s jays to all have the same likelihood of triggering the cameras.
167-170- I don’t think the data supports this interpretation. Yes, there were more brown bears detected at Herman creek, but Herman creek had similar numbers of eagles, more ravens, and more magpies than Klehini flats. I think it would be more accurate to say that there were more bears detected at Herman creek than Klehini flats, however, you still have the problem of differences in the size of camera viewsheds. This would be more clear if the x- axes of figure 3 were all the same. Again, comparisons across species and sites might be due to a confounding relationship between species-specific detection probability and site.
173-177- Yes, and this result agrees with some past research on the subject (Fortin et al. 2007).
197-202- Nice, this addresses my earlier comment.
203- I like that you address the implications of a multi-salmon species system on consumers, but I would be careful about saying that you can fish the less valued species more intensively because they are all equally used by bears. Gende et al. (2004, Oikos) documented bears selecting salmon with higher fat content because they were a more beneficial resource; my personal observation is that bears prefer kings> coho> sockeye> chums> pinks. Assuming this reflects the nutritional benefit of each species, this suggests that you cannot exchange a coho for a pink and not harm bears. Another interpretation of your results is that bears and human fishermen likely benefit from the diversity of salmon species in this area. Schindler et al. (2010) showed how like life history diversity within sockeye salmon populations created a population portfolio that benefited predators and commercial fishers. Run timing diversity extended the duration of salmon availability to wildlife consumers and annual returns were more stable for wildlife and commercial fisheries because of the so called “portfolio effect.” These effects may be even more beneficial to consumers in your system because it involves multiple salmon species.
209- 210- Yes, a perfect example of how multiple salmon species system can extend the duration of access to salmon for bears.
213- This paragraph is interesting.
224- Or, are the other species just less conspicuous?
238-243- Salmon are likely important to mink population productivity (Ben David, 2011), but their use of salmon is more difficult to observe than a bear or eagle.
248-251- I strongly disagree. Salmon species are not equally valuable as a resource, nor are salmon spawning in different areas equally available. You are assuming that the number of detections of an bear at a location is always correlated with the value of the resource. Contrast the number of times you would detect a bear that spends an hour to catch a sockeye salmon spawning on a lake shore versus a bear that can easily catch pink salmon in a shallow stream. The bear at the stream may become satiated in 45 minutes and go take a nap while the other bear fishes for hours. You would detect the lake shore bear many more times, but the pink salmon population is clearly more valuable to the bear.
253-256- Good point.
Because of problems with the experimental design, I don't think that the findings are valid. I think this paper could be rewritten to focus just on contrasting the patterns of detection of bears and eagles across time (the other species distract from the clear contrast between bears and eagles). The problems with the methodology (comment 151 above) can be minimized if you pooled detections across sites with a given period of availability (e.g. late chum, early pink) and normalized the number of detections between zero and one (divide by max daily detections of a species) to account for differences in probability of detection. With these changes I think you could contrast the patterns of use (bears used salmon during all time periods, while eagles focused on late chum, reflecting their different life history strategies).
If this study is ongoing, it would be better to use time lapse photography instead of motion trigger, or, pair a motion trigger camera with a video camera to check whether detection probability varies by species and/or site.
Valuable research question.
Main text needs a bit more proof-reading:
ex., line 146 "weighed" should be "weighted".
ex., line 25-26 change "to determine how different wildlife species use salmon resources" to something along the lines of "to discern potentially different use patterns among consumers"
Line 87: It is not exactly the brains (rather small) bears are selectively eating but more accurately the fat deposits on the top portion of their heads.
Lines 89-90: consuming on 25% of caught salmon is relatively rare. When salmon are very abundant and bears are relatively satiated they do often tend to "high-grade" but quite often come back to carcasses they have high-graded and left. This is more common after about two weeks into the salmon run. Early on "high-grading" is much less common.
Need to be better state what the limitations and assumptions are with methodology & design.
Design and methodology needs more thought and possible improvement.
Why was motion-detection used instead of a more systematic time interval sampling regime? Different consumer species are likely to present different rates of triggering of cameras. This probably biased detection rates and results more than a time lapse photographic approach. What was the layout of the twenty cameras used to monitor wildlife activity on spawning grounds? How were cameras distributed among and within different spawning grounds categories?
How was run timing and salmon abundance measured? General observations or a more systematic approach. Needs to be clearly stated.
The assumption that unique individuals were detected by two or five minute delays is probably not true. .....
Run tests for significance of differences among among consumer types and among different sites or spawning categories.....
Stated that placed cameras along roads and trails in early summer to provide relative index of abundance of various terrestrial carnivores: Limitations of this is that use of trails and roads may not be proportional across species (some species tend to use roads and established trails more than others); only sampling or indexing use in early summer may have biased results because distribution of different terrestrial carnivores varies according to season and trails and roads are not randomly distributed across the study area….. Might want to also see if ADFG or Forest Service may have accurate estimates of terrestrial carnivore abundance. .....
Lines 161-163: The assumption that a five minute delay will confidently unique visitations (individuals?) is likely not true - at least for many consumers.
The authors have a wealth of photographic data that is valuable. I suggest they try using more refined methods and analyses to make the most of this information.
Lines 174-177: Occurrence or frequency of visitation at various sites does not equate to "higher foraging efficiency".
This paper makes natural history observations about the use of salmon resources by terrestrial consumers. The observations, though relatively basic, are valuable to our understanding of the direct effects of salmon subsidies. I look forward to seeing this work published and I hope to see more basic natural history work given some of the holes in our understanding of salmon subsides.
Since this paper is putatively about vertebrate consumers, it would be helpful if it would make a small effort to consider the work that has been done on vertebrate consumers that lack feathers and fur. Stream ecologists have done most of the work on salmon subsidies, and they may be confused by the terminology in this manuscript and its tendency to ignore prior work in freshwater environments. For example, the list of vertebrate consumers in the intro does not include fishes! I realize that PeerJ is about getting the data out there and less about placing results in a broader context, but as long as the format is going to include a results and discussion section, I think it’s reasonable to expect manuscripts to make a decent attempt at placing new material within the context of existing work. See comments below for specific examples.
While I agree that this paper has some management implications, the discussion does not put forward arguments that are either logically strong, or well supported by existing work. The paper makes little attempt to reference existing work that considers how fisheries can affect salmon availability across space and time (i.e. Doctor et al. 2010 Trans. Am. Fish Soc., Boatright et al. 2004 Trans. Am. Fish. Soc.) or how salmon abundance affects consumer foraging opportunities (Bentley et al. 2012 Ecosphere).
12: “distinct life histories” seems like an overstatement, many salmon consumers opportunistic foragers
34: what about fishes?
39: I didn’t totally follow this sentence
48: Smaller birds have higher caloric demands per unit body mass
51: the phrase distinct life-histories is used a few times but I don’t get much meaning out of it
58: Don’t eagles also deposit fat (though not to the same degree) and don’t bears also track asynchronous run-timings? Lisi et al. 2012 Geomorphology showed that substantial phenological variation in salmon can be expressed over small spatial extents, within the range of consumers with modest mobility. In fact due to limited spatial autocorrelation in salmon phenology, the relationship between spatial extent and salmon phenological extent may asymptote within a bear’s home range.
64: I’ve seen >70 bears on a 3 km stream with the last run of salmon in a watershed, I’m not sure that late season aggregations are unique to eagles.
69: the effects of stream size on bear predation is well documented by Quinn, Carlson, Hendre and others and could be cited here.
69: Habitat features have been shown to strongly influence whether fishes can exploit salmon subsidies: Armstrong et al. 2010 Ecology Thermal heterogeneity mediates the effects of pulsed subsidies across a landscape…
83: What is a pool off of a larger river? Floodplain habitat?
89: a sentence reiterating the methodologies would be helpful, so that we know what monitoring an individual carcass means.
93: I’d never heard of lake-spawning coho salmon, are you sure they are spawning along the lake sure and not simply aggregating there? Interesting regardless.
96: How were the salmon breeding phenologies determined?
114: Not a big deal, but usually freshwater scientists don’t call small bodies of flowing water “creeks”, they call them “streams” and only use creek in the name.
116: It’d be nice if you could define what you mean by a pool in terms that would make sense to a stream ecologist – the current definition is likely not what stream ecologists think of as a pool. The results of this paper are relevant to freshwater folks studying salmon subsidies, so to maximize the impact it would help to use clear terminology.
139: What was the condition of the carcass or how were they aquired? Since the energy density of spawning salmon declines over time (Hendry and Berg 1999 Can. J. Zool) the state of the carcass could influence its appeal to scavengers.
146: I’m not sure I find the total aggregate # of images useful, but since it’s listed I’m curious of whether false triggers are included and whether the cameras worked continuously and were never taken out by bears or ran out of batteries.
154: Quinn et al. 2001 do not suggest that bears do not feed along large rivers or lake shores, but instead that they exert lower predation rates on salmon in these habitats. I don’t think that the observation that bears still show up at rivers and lakes shores will be surprising to anyone, and it has also been documented in the literature. I’m not saying that this documentation is not a worthy contribution for natural history, just that it shouldn’t be pitched as contrary to the results of Quinn et al. 2001 at least not without more nuance.
170: The tricky part about making inferences from these data is that we don’t know all the salmon foraging habitats that were within the foraging neighborhoods of these consumers—if the birds didn’t show up at a small stream with a camera on it, was that because they don’t like small streams, or because they were on another small stream where a camera was not present? I’m not saying that these data don’t provide insights, but if the authors could address this potential issue it would make their arguments stronger.
185: If brown bears are quickly grabbing many of the carcasses, how does that affect the inference of scavenging by other consumers? Would you see potentially different results if you surrounded the carcasses with electric fences permeable to small carnivores?
192: What would prevent eagles from foraging in forested habitats? I’m just curious since they seem invulnerable to predation from all but bears, but they do seem to prefer open habitats.
197: I’ve noticed that the time of year strongly influences whether bears will consume carcasses left as bait. In unpublished studies of maggot scavenging dynamics, bears would grab a carcass within a couple hours at the very beginning of the salmon runs. In contrast (and intuitively) during the peak of the runs, most carcasses left out were not scavenged.
203: I am skeptical that managers would get much traction by arguing that a recent study showed bears fed on a variety of salmon runs, and therefore fishing levels should decline. The argument for backing off of chum and pink runs is interesting and perhaps ecologically valid, but those aren’t necessarily “low value” species. They may be on a per pound basis, but certainly the aggregate value of pink and chum fisheries can be substantial. Further, many Alaskans (whether native, Caucasian, “urban”, and rural) have little interest in maximizing brown bear productivity, but are very concerned about salmon yields.
213: Could the absence of consumers on early runs be related to other factors, such as constraints due to reproduction, or use of alternative food sources? For example some herbivores catch the “green wave” late and leave it early, presumably due to trade-offs between resource tracking and arrival at breeding sites (Kolzsch et al. 2015 J Anim. Ecol.)
228: it’s not clear whether this is speculation or backed by existing data or studies
234: the term integrate is confusing when applied to energy allocation
242: Seems worth noting that salmon carcasses are likely much lower in energy density than the alternative prey of small carnivores—a spawned out salmon is about ~3 kJ/g (Hendry and Berg 1999) whereas a rodent is about 2-3-times as energy dense (Cox and Secor 2007, Comp. Biochem. Phys A). Foraging theory suggests that the abundance of preferred prey (not alternative prey) influences prey switching, which in turn suggests that small carnivores might only switch to salmon when rodent cycles are in their low phase. See Lisi et al. 2013 Ecology of Freshwater Fish for twist on fish switching to rodent prey during years with low salmon abundance.
248: Combining all salmon biomass when managing for salmon misses the point that certain salmon populations have disproportionate importance to consumers, which I thought was a key point of this paper. I’m not saying it’s feasible to manage salmon fisheries for specific populations, but it would certainly seem important to avoid management practices that are likely to hammer specific populations, for example the common practice of meeting escapement goals and then heavily harvesting late arriving fish.
255: these results seem overstated, I would say “would likely” but “will” is not demonstrated by these data.
Figure 1: Can you color or symbol the sites by run timing and or species
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.