the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Mar 2023
The Clam Before the Storm: A Meta Analysis Showing the Effect of Combined Climate Change Stressors on Bivalves
Abstract. Impacts of a range of climate change on marine organisms have been analysed in laboratory and experimental studies. The use of different taxonomic groupings, and assessment of different processes, though, makes identifying overall trends challenging, and may mask phylogenetically different responses. Bivalve molluscs are an ecologically and economically important data-rich clade, allowing for assessment of individual vulnerability and across developmental stages. We use meta-analysis of 203 unique experimental setups to examine how bivalve growth rates respond to increased water temperature, acidity, deoxygenation, changes to salinity, and combinations of these drivers. Results show that anthropogenic climate change will affect different families of bivalves disproportionally but almost unanimously negatively. Almost all drivers and their combinations have significant negative effects on growth. Combined deoxygenation, acidification, and temperature shows the largest negative effect size. Eggs/larval bivalves are more vulnerable overall than either juveniles or adults. Infaunal taxa, including Tellinidae and Veneridae, appear more resistant to warming and oxygen reduction than epifaunal or free-swimming taxa but this assessment is based on a small number of datapoints. The current focus of experimental set-ups on commercially important taxa and families within a small range of habitats creates gaps in understanding of global impacts on these economically important foundation organisms.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(988 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-287', Lisa Levin, 28 Mar 2023
This is a valuable, well-written meta analysis of climate change impacts on bivalves, focusing largely on experimental work. The authors do a good job of explaining what is new about their work - illustrating differential responses at the family-level, among life stages and lifestyles, complexity of responses, and revealing biases in study taxa. There is thoughtful discussion of life stage differences, mechanisms underlying impacts, and implications for aquaculture, fisheries, conservation and restoration. For those not working with bivalves, it might be interesting for the authors to speculate in the discussion which of their findings might be general across classes (or phyla) , and which might be specific to bivalves (or possibly other calcifying taxa) and why.
It took me a while to find the link to the 79 articles used in the analysis. Perhaps indicate this earlier in the methods. I could not find the original data derived from these articles that formed the basis of the analysis. These should be provided to the reader as a supplement or separate database. I think this is standard practice for this type of work.
Below are comments, questions and suggestions intended to strengthen the presentation.
Introduction
Line 25-32. These average environmental changes do not really tell the whole story since the regional variability is huge (e.g. some areas have lost 40% of their oxygen).
Methods – It is hard from the text to evaluate what magnitude of stressors were applied (under specific RCP or SSP scenarios?) Were studies limited to coastal species? Perhaps provide a species list - or did I miss this?
Line 90 By selecting one trait (growth) but not including survivorship – how many studies were excluded? There are a number of papers that might have contributed to this study if survivorship were addressed. I recognize growth is sublethal and the survivorship is lethal ; would you expect results to be very different?
Line 100 Why was deoxygenation not a search term? Did this come up under oxygen or not?
Line 131 preformed or performed?
Natural variability in seawater associated with diel cycles, upwelling/seasonality , respiration as well as warming effects on gas solubility causes stressors to change together. So warmer temperatures are usually associated with lower oxygen, nighttime respiration draws down pH and O2 together – etc. Thus bivalves would be expected to be adapted to related changes. Perhaps discuss how this does or does not manifest in your results?
Line 187-88 The results are stated to be unexpected. This means that there were specific expectations. Could these be presented as hypotheses? Is it the different responses among families that were unexpected? Or the contrasting response to combinations of treatments?
This is stylistic – but for sentences where you give the statistical result at the start of the sentence and the scientific meaning at the end of the sentence – I suggest reversing these and leading with the science, which is what the reader cares most about (e.g., lines 204-208).
Line 307 – typo in via
Citation: https://doi.org/10.5194/egusphere-2023-287-RC1 -
AC1: 'Reply on RC1', George Hoppit, 29 May 2023
"This is a valuable, well-written meta analysis of climate change impacts on bivalves, focusing largely on experimental work. The authors do a good job of explaining what is new about their work - illustrating differential responses at the family-level, among life stages and lifestyles, complexity of responses, and revealing biases in study taxa. There is thoughtful discussion of life stage differences, mechanisms underlying impacts, and implications for aquaculture, fisheries, conservation and restoration. For those not working with bivalves, it might be interesting for the authors to speculate in the discussion which of their findings might be general across classes (or phyla) , and which might be specific to bivalves (or possibly other calcifying taxa) and why."
We thank Prof Levin for their time and positive comments. We appreciate they feel our manuscript adds value to the wider field. We are delighted they felt our manuscript would prompt other researchers to consider the taxonomic granularity of their respective study groups considering our findings and interpretations.
"It took me a while to find the link to the 79 articles used in the analysis. Perhaps indicate this earlier in the methods. I could not find the original data derived from these articles that formed the basis of the analysis. These should be provided to the reader as a supplement or separate database. I think this is standard practice for this type of work."
We provided the data and code in a temporary Github folder https://github.com/georgehoppit/Bivalve-meta-analysis, until finalisation of the manuscript. This repository contains the full dataset used for analysis, and the code we used to do so. It is our understanding that the journal will provide a clear link to our data once the manuscript has been accepted which is easily navigable for researchers interested. Additionally, we propose uploading the data to www.pangaea.de as a lot of other similar data can be found there.
"Below are comments, questions and suggestions intended to strengthen the presentation."
"Introduction"
"Line 25-32. These average environmental changes do not really tell the whole story since the regional variability is huge (e.g. some areas have lost 40% of their oxygen)."
We propose adding “with some areas, such as the Gulf of Mexico, already suffering frequent, severe deoxygenation events (Breitburg, et al., 2018). The heterogeneric nature of change in each parameter will result in a larger number of combinations of stressors and extent of the effect of each stressor on communities. Such interaction of stressors have been increasingly recognised in the literature over the last few decades though this, and the important regional variability in response, remain underexplored overall (add ref for an additive/synergistic paper).”
"Methods – It is hard from the text to evaluate what magnitude of stressors were applied (under specific RCP or SSP scenarios?) Were studies limited to coastal species? Perhaps provide a species list - or did I miss this?"
We will update the text with a line: When extracting data from papers we opted for experiments which explored plausible climate change conditions for end of century for the location and taxon. Data were extracted from studies which maintained constant experimental conditions thereby not exploring the natural diurnal variability which is large in coastal settings (ref) especially for species which are exposed to air at low tide. Data for individual species is available in our accompanying data which includes water depth estimates taken from the published literature and available online databases (e.g., WORMS, SeaLifeBase). We will also include an additional table housing information of organism habitat types.
"Line 90 By selecting one trait (growth) but not including survivorship – how many studies were excluded? There are a number of papers that might have contributed to this study if survivorship were addressed. I recognize growth is sublethal and the survivorship is lethal ; would you expect results to be very different?"
The reviewer raises and interesting point about whether survivorship would create different results, however this is beyond the scope of our study.
The number of excluded papers is shown in figure 1, where we excluded 347 paper not meeting the requirements for our analysis. As we focussed on growth not lethal conditions, we cannot quantify what the difference would without repeating the screening, as this information was not recorded due to the different focus.
Many studies do report survivorship often as a percentage, but these frequently do not contain all the information needed to preform meta-analysis. The meta-analysis approach required studies reporting sample mean, variance, and sample size in order to calculate the effect sizes that fed into the meta-analytical models. Thus, studies just reporting survivorship as a percentage cannot be used in the same approach, and hence part of our decision to exclude this trait response and focus on growth.
"Line 100 Why was deoxygenation not a search term? Did this come up under oxygen or not?"
“deoxygenation” was a permutation of oxygen that was covered by our search terms when identifying articles. We will include deoxygenation in our search terms section clarify, however running deoxygenation as a search term, together with Bivalvia or bivalve produces zero experimental papers on growth.
"Line 131 preformed or performed?"
Edit: performed
"Natural variability in seawater associated with diel cycles, upwelling/seasonality , respiration as well as warming effects on gas solubility causes stressors to change together. So warmer temperatures are usually associated with lower oxygen, nighttime respiration draws down pH and O2 together – etc. Thus bivalves would be expected to be adapted to related changes. Perhaps discuss how this does or does not manifest in your results?"
We will add this point to the introduction to introduce the information early on (see comment above). We will update our methods to specify that data come from experiments that maintained constant conditions, and update our discussion to reflect the reviewers valuable point and how this is not accounted for in our results but is an important consideration.
"Line 187-88 The results are stated to be unexpected. This means that there were specific expectations. Could these be presented as hypotheses? Is it the different responses among families that were unexpected? Or the contrasting response to combinations of treatments?"
We will edit the text to better phrase the point we wished to convey in this line: Based on available data, families do not all respond in the same way as the whole class Bivalvia. Stressors affect different families in ways that cannot be predicted from analysis of Bivalvia as a whole (Fig. 5).
"This is stylistic – but for sentences where you give the statistical result at the start of the sentence and the scientific meaning at the end of the sentence – I suggest reversing these and leading with the science, which is what the reader cares most about (e.g., lines 204-208)."
We will consider this throughout the text. For the specific example, we propose an edit e.g.: Publications with significant results are published more often than would be expected by chance, suggesting negative observations are less frequently reported (see Appendix A; Table A1), as shown by highly significant Egger’s regression test (P < 0.001) results for every stressor.
"Line 307 – typo in via"
Edited
Citation: https://doi.org/10.5194/egusphere-2023-287-AC1
-
AC1: 'Reply on RC1', George Hoppit, 29 May 2023
-
RC2: 'Comment on egusphere-2023-287', Anonymous Referee #2, 10 May 2023
The manuscript of Kruft Welton et al. reports on a meta-analysis focused on the effects of climate change stressors (pH, O2, temperature and salinity) on bivalves. Many papers have been published in the last decade showing that these stressors (isolated and/or combined) have significant effects on these species. The present study and manuscript follow a bunch of meta-analyses not necessarily focusing on this group but that performed separated analyses on them: Kroecker et al. (2013) showed that based on 45 datasets bivalve growth is significantly impacted by pH, Harvey et al. (2013) showed that the combination of warming and acidification is synergistic, Sampaio et al. (2021) focused on the effects of pH, temperature and O2, the most recent Leung et al. (2022) restricted their study on pH but on a very comprehensive database. The main “new” approach here is to perform meta-analyses on different families of bivalves that were studied so far. Although I believe that identifying some bias in the literature towards economically important species is important (but that a narrative review would have done), I do not recommend considering this manuscript for publication for the three following reasons:
- Apart from separating by families (but I will come back on that after), I do not see what is the real novelty presented here, I am sorry but we already knew that bivalves are impacted by climate changes, that larval stages are certainly the most sensitive stage etc…
- I have strong concerns about the use of meta-analyses when
- protocols are so different from each other (level of perturbation applied, exposure time, amount of food provided (artificial vs natural, 1 phytoplankton species or a cocktail of species etc…), parameter or process measured: tissue growth, shell growth, dry mass etc…). Nothing is said here. Even the most important to me, the level of perturbation: L121. “Climate stressor values were based on realistic end of century projections based on author’s determination for that experimental setup or study location.” In the data table (thanks for sharing), pH offsets can be as low as -0.8 (maybe more, as the table is difficult to read), or as small as -0.2 (actually where do come from the 0.3-1 pH units mentioned in the introduction for the end-century projected changes in pH based on the IPCC?). It is very hard to see what is the range of O2 changes considered here, since units are different in the table and not easy to read, are they realistic? For temperature, the same, what is the range of offsets considered? Were these factors considered as constant or were they varying? Very important to consider in the coastal ocean where strong dynamics are very common, and placing an organism under stable conditions is already a source of changes compared to their natural environment. What about salinity? When you mention pH, you refer to acidification (decrease in pH), when you mention T, you refer to warming (increase in T), when you mention O2, you refer to deoxygenation (right?), when you mention salinity, it is not clear if you refer to decrease or increase in salinity. Actually, from the data table, I see that both low salinity and high salinity effects are mixed together, isn’t that weird?
- the number of available datasets is very low. We have here on several cases the extreme situation where the “meta-analysis” is done on 1 experimental set-up… To me, it is very difficult to compare pH effects based on quite a lot of experimental setups (90), oxygen (18) and salinity (8). It is getting even worse when focusing on the combination of factors (e.g. pH + S = 2 experimental setups = 1 publication). Of course, when considering families separately, the number of experimental results is, in my opinion, way too low to apply any confidence level on the results (except for Mytilidae maybe).
- I have already mentioned that that because of important differences in the protocols used between the different datasets, the low number of datasets in many cases etc.., I believe that the results presented here have a low confidence level, I will push a bit further, and will take 2 examples showing that I even believe using this method in such conditions can lead to false conclusions.
- First of all, L168-169: “The effect of pH + salinity is intermediate between that of the two single stressors, dampening the salinity effect”. This assessment is based on the observation that salinity (8 datasets, increase, decrease of salinity???) has a stronger effect than pH+S (2 datasets). These 2 datasets are from the work of Dickinson (Dickinson et al., Interactive effects of salinity and elevated CO2 levels on juvenile eastern oysters, Crassostrea virginica. Journal of Experimental Biology 215, 29-43 (2012); Dickinson, O. B. Matoo, R. T. Tourek, I. M. Sokolova, E. Beniash, Environmental salinity modulates the effects of elevated CO2 levels on juvenile hard-shell clams, Mercenaria mercenaria. Journal of Experimental Biology 216, 2607-2618 (2013)). On oysters, they found that each factor leads to decreased growth (soft tissue) but that there is no interaction between these factors. On clams, they also showed that low salinity combined with hypercapnia leads to decrease in growth as compared to the growth measured following exposure to one of these stressors in isolation. So, we are far for the finding that a decrease in pH leads to a dampening of salinity effect….
- The second example is connected to the main conclusion of the article: L367-368: “We demonstrate that synergistic combinations of stressors (e.g., effects of combined temperature + O2 + pH change) cause greater reductions in bivalve growth then individual stressors.”. This assessment is based on 4 experimental setups, but from one single publication (Stevens, C. J. Gobler, Interactive effects of acidification, hypoxia, and thermal stress on growth, respiration, and survival of four North Atlantic bivalves. Marine Ecology Progress Series 604, 143-161 (2018)). They indicated that for 4 species of bivalves, “The combination of low DO and low pH often interacted antagonistically to yield growth rates higher than would be predicted from either individual stressor (…) Elevated temperature and low pH interacted both antagonistically and synergistically, producing outcomes that could not be predicted from the responses to individual stressors. Collectively, this study revealed species- and size-specific vulnerabilities of bivalves to coastal stressors along with unpredicted interactions among those stressors». Again, we are far from a “greater reduction in growth than individual stressors” with a much more complicated story.
Citation: https://doi.org/10.5194/egusphere-2023-287-RC2 -
AC2: 'Reply on RC2', George Hoppit, 29 May 2023
"The manuscript of Kruft Welton et al. reports on a meta-analysis focused on the effects of climate change stressors (pH, O2, temperature and salinity) on bivalves. Many papers have been published in the last decade showing that these stressors (isolated and/or combined) have significant effects on these species. The present study and manuscript follow a bunch of meta-analyses not necessarily focusing on this group but that performed separated analyses on them: Kroecker et al. (2013) showed that based on 45 datasets bivalve growth is significantly impacted by pH, Harvey et al. (2013) showed that the combination of warming and acidification is synergistic, Sampaio et al. (2021) focused on the effects of pH, temperature and O2, the most recent Leung et al. (2022) restricted their study on pH but on a very comprehensive database. The main “new” approach here is to perform meta-analyses on different families of bivalves that were studied so far. Although I believe that identifying some bias in the literature towards economically important species is important (but that a narrative review would have done), I do not recommend considering this manuscript for publication for the three following reasons:"
We thank the reviewer for taking time to consider our manuscript. We disagree with the conclusion that our manuscript should not be considered for publication as supported by the positive comments from reviewer 1. We present the most comprehensive known overview of datasets extracted from the literature recording the response of bivalves to climate change. Our manuscript explores a greater range and combination of climate change stressors than other known meta-analyses. Further, the value added by performing our meta-analysis at a family level highlights the variability of responses in what is often considered a well-studied group. Thereby our results show that previous work suggesting that bivalves/ molluscs respond poorly to climate change is not an assessment of bivalves overall but in reality only 3-4 families driving the signal. Our approach encourages future work to explore higher taxonomic granularity, as well as address the other biases we identify, which will support conservation and management strategies in light of climate change by emphasising disparity in organismal responses, which is needed by practitioners.
"Apart from separating by families (but I will come back on that after), I do not see what is the real novelty presented here, I am sorry but we already knew that bivalves are impacted by climate changes, that larval stages are certainly the most sensitive stage etc…"
Our analysis explores a much greater combination of climate change stressors then past work, specifically 10 stressor combinations, while the largest previous work in terms of stressors (Sampaio et al., 2021) explored 4. We consider this a strong development toward assessing the complexity needed and facilitated by a growing number of multi-stressor experiments.
We present the first known meta-analysis for marine climate change which explores the response of an individual group, and thereby identify that even a very well-studied group such as bivalves has large gaps in taxonomic understandings of how individual organisms respond to climate change. Our analysis was not intended to specifically point out that bivalves respond negatively to climate change, but highlight the taxonomic, geographic, and ecological variability experienced by the group, and provide an overview of the current state of experimental data. Therefore our findings should encourage a new approach to meta-analysis by moving towards more differentiated taxonomic understanding of a group’s responses to future conditions, as well as hopefully stimulate additional data collection. Only this approach is able to show the level of variability that can be present within individual groups while still providing a summative response above the individual species.
"I have strong concerns about the use of meta-analyses when protocols are so different from each other (level of perturbation applied, exposure time, amount of food provided (artificial vs natural, 1 phytoplankton species or a cocktail of species etc…), parameter or process measured: tissue growth, shell growth, dry mass etc…). Nothing is said here. Even the most important to me, the level of perturbation: L121. “Climate stressor values were based on realistic end of century projections based on author’s determination for that experimental setup or study location.” In the data table (thanks for sharing), pH offsets can be as low as -0.8 (maybe more, as the table is difficult to read), or as small as -0.2 (actually where do come from the 0.3-1 pH units mentioned in the introduction for the end-century projected changes in pH based on the IPCC?). It is very hard to see what is the range of O2 changes considered here, since units are different in the table and not easy to read, are they realistic? For temperature, the same, what is the range of offsets considered?"
"Were these factors considered as constant or were they varying? Very important to consider in the coastal ocean where strong dynamics are very common, and placing an organism under stable conditions is already a source of changes compared to their natural environment."
"What about salinity? When you mention pH, you refer to acidification (decrease in pH), when you mention T, you refer to warming (increase in T), when you mention O2, you refer to deoxygenation (right?), when you mention salinity, it is not clear if you refer to decrease or increase in salinity. Actually, from the data table, I see that both low salinity and high salinity effects are mixed together, isn’t that weird?"
The experimental literature we collected our database from is disparate, with papers having a wide range of experimental protocols to collect their data. Such a summative analysis across setups is the power of meta-analysis and the basis of all published analyses of this type. While we see the concern, the reviewer is seemingly challenging an entire field of well-established research without seeing the benefits of the summative power of this approach.
Sharing some of the concerns, we opted for ‘growth’ broadly as a measure of organism physiological response to climate stressors (and not more specific) for multiple reasons. Disentangling specific growth measurements would dilute our analysis greatly due to the wide disparity in the experimental approaches for measuring growth responses to climate stressors. Furthermore, we wanted to keep in line with the previously cited meta-analyses that we and the reviewer refer to (Krocker et al., 2013; Harvey et al., 2013; Sampaio et al., 2021), who all used growth and not more specific measurements. As such, analysing growth in this manner has become standardised in the synthesis literature for comparable research.
When extracting data from papers we also opted for experimental setups which are considered to represent plausible climate change values for end-of-century conditions in any given location/for any given taxon examined (please see our response to a similar point from reviewer 1). Data were only extracted from studies which maintained constant experimental conditions (see response to reviewer 1) and we will update the methods to better signpost this comment. Salinity can both increase and decrease because of climate change (given location specific factors) in contrast to oxygenation, temperature and acidification. Hence we opted for both increasing and decreasing under the umbrella term of ‘change’. We observed negative growth responses for both increased or decreased salinity; hence we felt justified in including salinity data in this manner. As salinity was not included in previous analyses we considered this an important addition in our analysis for highlighting the variability of climate stressors and the need to explore them, and for identifying potentially novel trends not present in other meta-analyses which focus largely on pH.
"the number of available datasets is very low. We have here on several cases the extreme situation where the “meta-analysis” is done on 1 experimental set-up… To me, it is very difficult to compare pH effects based on quite a lot of experimental setups (90), oxygen (18) and salinity (8). It is getting even worse when focusing on the combination of factors (e.g. pH + S = 2 experimental setups = 1 publication). Of course, when considering families separately, the number of experimental results is, in my opinion, way too low to apply any confidence level on the results (except for Mytilidae maybe)."
We fully agree with the reviewer that for some families, life stages, and combination of stressors the sample size is indeed low, but ultimately this reflects the current state of the experimental literature and is something that we point out as a major conclusion and recommendation for future work. We highlight the abundance of experiments focusing on pH as a literature bias relative to other stressors. We did not perform meta-analysis specifically on one experimental set up. Upon reviewing the code we provide in addition to our data, the reviewer can see that we used ‘stressor’ as a grouping moderator in the meta-analysis. (Code line 186 families_experiment_counts <- map(list_of_families, function (family) metadata_bivalves |> filter(Family == family) |> group_by(Stressor) |> summarize(count = n()))). This approach does not perform meta-analysis on any stressor individually, but places it within the context of all others, as a work around to biases in experimental data.
We recognise that indicating statistical significance on small sample sizes in instances of low data limits the confidence that can be applied to the specific cases, but within the context of the meta-analytical approach our conclusions are still valid. We specify in the paper that we are just working with what is available in the data. Our hope is that by drawing attention to these deficits and biases, it will encourage experimental work on a wider range of bivalves and climate stressors.
"I have already mentioned that that because of important differences in the protocols used between the different datasets, the low number of datasets in many cases etc.., I believe that the results presented here have a low confidence level, I will push a bit further, and will take 2 examples showing that I even believe using this method in such conditions can lead to false conclusions."
"First of all, L168-169: “The effect of pH + salinity is intermediate between that of the two single stressors, dampening the salinity effect”. This assessment is based on the observation that salinity (8 datasets, increase, decrease of salinity???) has a stronger effect than pH+S (2 datasets). These 2 datasets are from the work of Dickinson (Dickinson et al., Interactive effects of salinity and elevated CO2 levels on juvenile eastern oysters, Crassostrea virginica. Journal of Experimental Biology 215, 29-43 (2012); Dickinson, O. B. Matoo, R. T. Tourek, I. M. Sokolova, E. Beniash, Environmental salinity modulates the effects of elevated CO2 levels on juvenile hard-shell clams, Mercenaria mercenaria. Journal of Experimental Biology 216, 2607-2618 (2013)). On oysters, they found that each factor leads to decreased growth (soft tissue) but that there is no interaction between these factors. On clams, they also showed that low salinity combined with hypercapnia leads to decrease in growth as compared to the growth measured following exposure to one of these stressors in isolation. So, we are far for the finding that a decrease in pH leads to a dampening of salinity effect…."
We fear our point might have been misunderstood by our choice of language regarding the sentence “The effect of pH + salinity is intermediate between that of the two single stressors, dampening the salinity effect”. We will restructure the text to remove “dampening” and just point out that its intermediate between the two other stressors. But again, our approach with meta-analysis was not to drill down to specific results, but try and place them within the wider narrative of the collective experimental data we extracted and analyses.
While we consider the inclusion of the salinity information important, we do not place large narrative importance on the impacts of salinity due to the low number of experimental studies. Our suggestion to restructure the text that referred to “dampening” will aid this. The point of our text is to draw attention to the state of the published data, while also emphasising the ‘direction of travel’ for bivalves as a group based on these data. We like to convey that salinity is a lesser explored stressor but may have interesting interactions with others, e.g., pH, temp, and oxygen.
"The second example is connected to the main conclusion of the article: L367-368: “We demonstrate that synergistic combinations of stressors (e.g., effects of combined temperature + O2 + pH change) cause greater reductions in bivalve growth then individual stressors.”. This assessment is based on 4 experimental setups, but from one single publication (Stevens, C. J. Gobler, Interactive effects of acidification, hypoxia, and thermal stress on growth, respiration, and survival of four North Atlantic bivalves. Marine Ecology Progress Series 604, 143-161 (2018)). They indicated that for 4 species of bivalves, “The combination of low DO and low pH often interacted antagonistically to yield growth rates higher than would be predicted from either individual stressor (…) Elevated temperature and low pH interacted both antagonistically and synergistically, producing outcomes that could not be predicted from the responses to individual stressors. Collectively, this study revealed species- and size-specific vulnerabilities of bivalves to coastal stressors along with unpredicted interactions among those stressors». Again, we are far from a “greater reduction in growth than individual stressors” with a much more complicated story."
We disagree with the reviewer who clearly rejects the approach of meta-analysis. Our conclusions map across with the most directly comparable work (Sampaio et al). Most importantly, our point is not in sole reference to Stevens and Gobler (2018) but refers to the fact that adding different climate stressors amplifies negative responses as shown in detail in our results. The statement is derived from the combined temperature, O2, and pH change detected in our work. We fully agree that the “the story may indeed be more complicated”, the idea of a meta-analysis is to find commonalties and we are confident that our work adds to the understanding.
Citation: https://doi.org/10.5194/egusphere-2023-287-AC2
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-287', Lisa Levin, 28 Mar 2023
This is a valuable, well-written meta analysis of climate change impacts on bivalves, focusing largely on experimental work. The authors do a good job of explaining what is new about their work - illustrating differential responses at the family-level, among life stages and lifestyles, complexity of responses, and revealing biases in study taxa. There is thoughtful discussion of life stage differences, mechanisms underlying impacts, and implications for aquaculture, fisheries, conservation and restoration. For those not working with bivalves, it might be interesting for the authors to speculate in the discussion which of their findings might be general across classes (or phyla) , and which might be specific to bivalves (or possibly other calcifying taxa) and why.
It took me a while to find the link to the 79 articles used in the analysis. Perhaps indicate this earlier in the methods. I could not find the original data derived from these articles that formed the basis of the analysis. These should be provided to the reader as a supplement or separate database. I think this is standard practice for this type of work.
Below are comments, questions and suggestions intended to strengthen the presentation.
Introduction
Line 25-32. These average environmental changes do not really tell the whole story since the regional variability is huge (e.g. some areas have lost 40% of their oxygen).
Methods – It is hard from the text to evaluate what magnitude of stressors were applied (under specific RCP or SSP scenarios?) Were studies limited to coastal species? Perhaps provide a species list - or did I miss this?
Line 90 By selecting one trait (growth) but not including survivorship – how many studies were excluded? There are a number of papers that might have contributed to this study if survivorship were addressed. I recognize growth is sublethal and the survivorship is lethal ; would you expect results to be very different?
Line 100 Why was deoxygenation not a search term? Did this come up under oxygen or not?
Line 131 preformed or performed?
Natural variability in seawater associated with diel cycles, upwelling/seasonality , respiration as well as warming effects on gas solubility causes stressors to change together. So warmer temperatures are usually associated with lower oxygen, nighttime respiration draws down pH and O2 together – etc. Thus bivalves would be expected to be adapted to related changes. Perhaps discuss how this does or does not manifest in your results?
Line 187-88 The results are stated to be unexpected. This means that there were specific expectations. Could these be presented as hypotheses? Is it the different responses among families that were unexpected? Or the contrasting response to combinations of treatments?
This is stylistic – but for sentences where you give the statistical result at the start of the sentence and the scientific meaning at the end of the sentence – I suggest reversing these and leading with the science, which is what the reader cares most about (e.g., lines 204-208).
Line 307 – typo in via
Citation: https://doi.org/10.5194/egusphere-2023-287-RC1 -
AC1: 'Reply on RC1', George Hoppit, 29 May 2023
"This is a valuable, well-written meta analysis of climate change impacts on bivalves, focusing largely on experimental work. The authors do a good job of explaining what is new about their work - illustrating differential responses at the family-level, among life stages and lifestyles, complexity of responses, and revealing biases in study taxa. There is thoughtful discussion of life stage differences, mechanisms underlying impacts, and implications for aquaculture, fisheries, conservation and restoration. For those not working with bivalves, it might be interesting for the authors to speculate in the discussion which of their findings might be general across classes (or phyla) , and which might be specific to bivalves (or possibly other calcifying taxa) and why."
We thank Prof Levin for their time and positive comments. We appreciate they feel our manuscript adds value to the wider field. We are delighted they felt our manuscript would prompt other researchers to consider the taxonomic granularity of their respective study groups considering our findings and interpretations.
"It took me a while to find the link to the 79 articles used in the analysis. Perhaps indicate this earlier in the methods. I could not find the original data derived from these articles that formed the basis of the analysis. These should be provided to the reader as a supplement or separate database. I think this is standard practice for this type of work."
We provided the data and code in a temporary Github folder https://github.com/georgehoppit/Bivalve-meta-analysis, until finalisation of the manuscript. This repository contains the full dataset used for analysis, and the code we used to do so. It is our understanding that the journal will provide a clear link to our data once the manuscript has been accepted which is easily navigable for researchers interested. Additionally, we propose uploading the data to www.pangaea.de as a lot of other similar data can be found there.
"Below are comments, questions and suggestions intended to strengthen the presentation."
"Introduction"
"Line 25-32. These average environmental changes do not really tell the whole story since the regional variability is huge (e.g. some areas have lost 40% of their oxygen)."
We propose adding “with some areas, such as the Gulf of Mexico, already suffering frequent, severe deoxygenation events (Breitburg, et al., 2018). The heterogeneric nature of change in each parameter will result in a larger number of combinations of stressors and extent of the effect of each stressor on communities. Such interaction of stressors have been increasingly recognised in the literature over the last few decades though this, and the important regional variability in response, remain underexplored overall (add ref for an additive/synergistic paper).”
"Methods – It is hard from the text to evaluate what magnitude of stressors were applied (under specific RCP or SSP scenarios?) Were studies limited to coastal species? Perhaps provide a species list - or did I miss this?"
We will update the text with a line: When extracting data from papers we opted for experiments which explored plausible climate change conditions for end of century for the location and taxon. Data were extracted from studies which maintained constant experimental conditions thereby not exploring the natural diurnal variability which is large in coastal settings (ref) especially for species which are exposed to air at low tide. Data for individual species is available in our accompanying data which includes water depth estimates taken from the published literature and available online databases (e.g., WORMS, SeaLifeBase). We will also include an additional table housing information of organism habitat types.
"Line 90 By selecting one trait (growth) but not including survivorship – how many studies were excluded? There are a number of papers that might have contributed to this study if survivorship were addressed. I recognize growth is sublethal and the survivorship is lethal ; would you expect results to be very different?"
The reviewer raises and interesting point about whether survivorship would create different results, however this is beyond the scope of our study.
The number of excluded papers is shown in figure 1, where we excluded 347 paper not meeting the requirements for our analysis. As we focussed on growth not lethal conditions, we cannot quantify what the difference would without repeating the screening, as this information was not recorded due to the different focus.
Many studies do report survivorship often as a percentage, but these frequently do not contain all the information needed to preform meta-analysis. The meta-analysis approach required studies reporting sample mean, variance, and sample size in order to calculate the effect sizes that fed into the meta-analytical models. Thus, studies just reporting survivorship as a percentage cannot be used in the same approach, and hence part of our decision to exclude this trait response and focus on growth.
"Line 100 Why was deoxygenation not a search term? Did this come up under oxygen or not?"
“deoxygenation” was a permutation of oxygen that was covered by our search terms when identifying articles. We will include deoxygenation in our search terms section clarify, however running deoxygenation as a search term, together with Bivalvia or bivalve produces zero experimental papers on growth.
"Line 131 preformed or performed?"
Edit: performed
"Natural variability in seawater associated with diel cycles, upwelling/seasonality , respiration as well as warming effects on gas solubility causes stressors to change together. So warmer temperatures are usually associated with lower oxygen, nighttime respiration draws down pH and O2 together – etc. Thus bivalves would be expected to be adapted to related changes. Perhaps discuss how this does or does not manifest in your results?"
We will add this point to the introduction to introduce the information early on (see comment above). We will update our methods to specify that data come from experiments that maintained constant conditions, and update our discussion to reflect the reviewers valuable point and how this is not accounted for in our results but is an important consideration.
"Line 187-88 The results are stated to be unexpected. This means that there were specific expectations. Could these be presented as hypotheses? Is it the different responses among families that were unexpected? Or the contrasting response to combinations of treatments?"
We will edit the text to better phrase the point we wished to convey in this line: Based on available data, families do not all respond in the same way as the whole class Bivalvia. Stressors affect different families in ways that cannot be predicted from analysis of Bivalvia as a whole (Fig. 5).
"This is stylistic – but for sentences where you give the statistical result at the start of the sentence and the scientific meaning at the end of the sentence – I suggest reversing these and leading with the science, which is what the reader cares most about (e.g., lines 204-208)."
We will consider this throughout the text. For the specific example, we propose an edit e.g.: Publications with significant results are published more often than would be expected by chance, suggesting negative observations are less frequently reported (see Appendix A; Table A1), as shown by highly significant Egger’s regression test (P < 0.001) results for every stressor.
"Line 307 – typo in via"
Edited
Citation: https://doi.org/10.5194/egusphere-2023-287-AC1
-
AC1: 'Reply on RC1', George Hoppit, 29 May 2023
-
RC2: 'Comment on egusphere-2023-287', Anonymous Referee #2, 10 May 2023
The manuscript of Kruft Welton et al. reports on a meta-analysis focused on the effects of climate change stressors (pH, O2, temperature and salinity) on bivalves. Many papers have been published in the last decade showing that these stressors (isolated and/or combined) have significant effects on these species. The present study and manuscript follow a bunch of meta-analyses not necessarily focusing on this group but that performed separated analyses on them: Kroecker et al. (2013) showed that based on 45 datasets bivalve growth is significantly impacted by pH, Harvey et al. (2013) showed that the combination of warming and acidification is synergistic, Sampaio et al. (2021) focused on the effects of pH, temperature and O2, the most recent Leung et al. (2022) restricted their study on pH but on a very comprehensive database. The main “new” approach here is to perform meta-analyses on different families of bivalves that were studied so far. Although I believe that identifying some bias in the literature towards economically important species is important (but that a narrative review would have done), I do not recommend considering this manuscript for publication for the three following reasons:
- Apart from separating by families (but I will come back on that after), I do not see what is the real novelty presented here, I am sorry but we already knew that bivalves are impacted by climate changes, that larval stages are certainly the most sensitive stage etc…
- I have strong concerns about the use of meta-analyses when
- protocols are so different from each other (level of perturbation applied, exposure time, amount of food provided (artificial vs natural, 1 phytoplankton species or a cocktail of species etc…), parameter or process measured: tissue growth, shell growth, dry mass etc…). Nothing is said here. Even the most important to me, the level of perturbation: L121. “Climate stressor values were based on realistic end of century projections based on author’s determination for that experimental setup or study location.” In the data table (thanks for sharing), pH offsets can be as low as -0.8 (maybe more, as the table is difficult to read), or as small as -0.2 (actually where do come from the 0.3-1 pH units mentioned in the introduction for the end-century projected changes in pH based on the IPCC?). It is very hard to see what is the range of O2 changes considered here, since units are different in the table and not easy to read, are they realistic? For temperature, the same, what is the range of offsets considered? Were these factors considered as constant or were they varying? Very important to consider in the coastal ocean where strong dynamics are very common, and placing an organism under stable conditions is already a source of changes compared to their natural environment. What about salinity? When you mention pH, you refer to acidification (decrease in pH), when you mention T, you refer to warming (increase in T), when you mention O2, you refer to deoxygenation (right?), when you mention salinity, it is not clear if you refer to decrease or increase in salinity. Actually, from the data table, I see that both low salinity and high salinity effects are mixed together, isn’t that weird?
- the number of available datasets is very low. We have here on several cases the extreme situation where the “meta-analysis” is done on 1 experimental set-up… To me, it is very difficult to compare pH effects based on quite a lot of experimental setups (90), oxygen (18) and salinity (8). It is getting even worse when focusing on the combination of factors (e.g. pH + S = 2 experimental setups = 1 publication). Of course, when considering families separately, the number of experimental results is, in my opinion, way too low to apply any confidence level on the results (except for Mytilidae maybe).
- I have already mentioned that that because of important differences in the protocols used between the different datasets, the low number of datasets in many cases etc.., I believe that the results presented here have a low confidence level, I will push a bit further, and will take 2 examples showing that I even believe using this method in such conditions can lead to false conclusions.
- First of all, L168-169: “The effect of pH + salinity is intermediate between that of the two single stressors, dampening the salinity effect”. This assessment is based on the observation that salinity (8 datasets, increase, decrease of salinity???) has a stronger effect than pH+S (2 datasets). These 2 datasets are from the work of Dickinson (Dickinson et al., Interactive effects of salinity and elevated CO2 levels on juvenile eastern oysters, Crassostrea virginica. Journal of Experimental Biology 215, 29-43 (2012); Dickinson, O. B. Matoo, R. T. Tourek, I. M. Sokolova, E. Beniash, Environmental salinity modulates the effects of elevated CO2 levels on juvenile hard-shell clams, Mercenaria mercenaria. Journal of Experimental Biology 216, 2607-2618 (2013)). On oysters, they found that each factor leads to decreased growth (soft tissue) but that there is no interaction between these factors. On clams, they also showed that low salinity combined with hypercapnia leads to decrease in growth as compared to the growth measured following exposure to one of these stressors in isolation. So, we are far for the finding that a decrease in pH leads to a dampening of salinity effect….
- The second example is connected to the main conclusion of the article: L367-368: “We demonstrate that synergistic combinations of stressors (e.g., effects of combined temperature + O2 + pH change) cause greater reductions in bivalve growth then individual stressors.”. This assessment is based on 4 experimental setups, but from one single publication (Stevens, C. J. Gobler, Interactive effects of acidification, hypoxia, and thermal stress on growth, respiration, and survival of four North Atlantic bivalves. Marine Ecology Progress Series 604, 143-161 (2018)). They indicated that for 4 species of bivalves, “The combination of low DO and low pH often interacted antagonistically to yield growth rates higher than would be predicted from either individual stressor (…) Elevated temperature and low pH interacted both antagonistically and synergistically, producing outcomes that could not be predicted from the responses to individual stressors. Collectively, this study revealed species- and size-specific vulnerabilities of bivalves to coastal stressors along with unpredicted interactions among those stressors». Again, we are far from a “greater reduction in growth than individual stressors” with a much more complicated story.
Citation: https://doi.org/10.5194/egusphere-2023-287-RC2 -
AC2: 'Reply on RC2', George Hoppit, 29 May 2023
"The manuscript of Kruft Welton et al. reports on a meta-analysis focused on the effects of climate change stressors (pH, O2, temperature and salinity) on bivalves. Many papers have been published in the last decade showing that these stressors (isolated and/or combined) have significant effects on these species. The present study and manuscript follow a bunch of meta-analyses not necessarily focusing on this group but that performed separated analyses on them: Kroecker et al. (2013) showed that based on 45 datasets bivalve growth is significantly impacted by pH, Harvey et al. (2013) showed that the combination of warming and acidification is synergistic, Sampaio et al. (2021) focused on the effects of pH, temperature and O2, the most recent Leung et al. (2022) restricted their study on pH but on a very comprehensive database. The main “new” approach here is to perform meta-analyses on different families of bivalves that were studied so far. Although I believe that identifying some bias in the literature towards economically important species is important (but that a narrative review would have done), I do not recommend considering this manuscript for publication for the three following reasons:"
We thank the reviewer for taking time to consider our manuscript. We disagree with the conclusion that our manuscript should not be considered for publication as supported by the positive comments from reviewer 1. We present the most comprehensive known overview of datasets extracted from the literature recording the response of bivalves to climate change. Our manuscript explores a greater range and combination of climate change stressors than other known meta-analyses. Further, the value added by performing our meta-analysis at a family level highlights the variability of responses in what is often considered a well-studied group. Thereby our results show that previous work suggesting that bivalves/ molluscs respond poorly to climate change is not an assessment of bivalves overall but in reality only 3-4 families driving the signal. Our approach encourages future work to explore higher taxonomic granularity, as well as address the other biases we identify, which will support conservation and management strategies in light of climate change by emphasising disparity in organismal responses, which is needed by practitioners.
"Apart from separating by families (but I will come back on that after), I do not see what is the real novelty presented here, I am sorry but we already knew that bivalves are impacted by climate changes, that larval stages are certainly the most sensitive stage etc…"
Our analysis explores a much greater combination of climate change stressors then past work, specifically 10 stressor combinations, while the largest previous work in terms of stressors (Sampaio et al., 2021) explored 4. We consider this a strong development toward assessing the complexity needed and facilitated by a growing number of multi-stressor experiments.
We present the first known meta-analysis for marine climate change which explores the response of an individual group, and thereby identify that even a very well-studied group such as bivalves has large gaps in taxonomic understandings of how individual organisms respond to climate change. Our analysis was not intended to specifically point out that bivalves respond negatively to climate change, but highlight the taxonomic, geographic, and ecological variability experienced by the group, and provide an overview of the current state of experimental data. Therefore our findings should encourage a new approach to meta-analysis by moving towards more differentiated taxonomic understanding of a group’s responses to future conditions, as well as hopefully stimulate additional data collection. Only this approach is able to show the level of variability that can be present within individual groups while still providing a summative response above the individual species.
"I have strong concerns about the use of meta-analyses when protocols are so different from each other (level of perturbation applied, exposure time, amount of food provided (artificial vs natural, 1 phytoplankton species or a cocktail of species etc…), parameter or process measured: tissue growth, shell growth, dry mass etc…). Nothing is said here. Even the most important to me, the level of perturbation: L121. “Climate stressor values were based on realistic end of century projections based on author’s determination for that experimental setup or study location.” In the data table (thanks for sharing), pH offsets can be as low as -0.8 (maybe more, as the table is difficult to read), or as small as -0.2 (actually where do come from the 0.3-1 pH units mentioned in the introduction for the end-century projected changes in pH based on the IPCC?). It is very hard to see what is the range of O2 changes considered here, since units are different in the table and not easy to read, are they realistic? For temperature, the same, what is the range of offsets considered?"
"Were these factors considered as constant or were they varying? Very important to consider in the coastal ocean where strong dynamics are very common, and placing an organism under stable conditions is already a source of changes compared to their natural environment."
"What about salinity? When you mention pH, you refer to acidification (decrease in pH), when you mention T, you refer to warming (increase in T), when you mention O2, you refer to deoxygenation (right?), when you mention salinity, it is not clear if you refer to decrease or increase in salinity. Actually, from the data table, I see that both low salinity and high salinity effects are mixed together, isn’t that weird?"
The experimental literature we collected our database from is disparate, with papers having a wide range of experimental protocols to collect their data. Such a summative analysis across setups is the power of meta-analysis and the basis of all published analyses of this type. While we see the concern, the reviewer is seemingly challenging an entire field of well-established research without seeing the benefits of the summative power of this approach.
Sharing some of the concerns, we opted for ‘growth’ broadly as a measure of organism physiological response to climate stressors (and not more specific) for multiple reasons. Disentangling specific growth measurements would dilute our analysis greatly due to the wide disparity in the experimental approaches for measuring growth responses to climate stressors. Furthermore, we wanted to keep in line with the previously cited meta-analyses that we and the reviewer refer to (Krocker et al., 2013; Harvey et al., 2013; Sampaio et al., 2021), who all used growth and not more specific measurements. As such, analysing growth in this manner has become standardised in the synthesis literature for comparable research.
When extracting data from papers we also opted for experimental setups which are considered to represent plausible climate change values for end-of-century conditions in any given location/for any given taxon examined (please see our response to a similar point from reviewer 1). Data were only extracted from studies which maintained constant experimental conditions (see response to reviewer 1) and we will update the methods to better signpost this comment. Salinity can both increase and decrease because of climate change (given location specific factors) in contrast to oxygenation, temperature and acidification. Hence we opted for both increasing and decreasing under the umbrella term of ‘change’. We observed negative growth responses for both increased or decreased salinity; hence we felt justified in including salinity data in this manner. As salinity was not included in previous analyses we considered this an important addition in our analysis for highlighting the variability of climate stressors and the need to explore them, and for identifying potentially novel trends not present in other meta-analyses which focus largely on pH.
"the number of available datasets is very low. We have here on several cases the extreme situation where the “meta-analysis” is done on 1 experimental set-up… To me, it is very difficult to compare pH effects based on quite a lot of experimental setups (90), oxygen (18) and salinity (8). It is getting even worse when focusing on the combination of factors (e.g. pH + S = 2 experimental setups = 1 publication). Of course, when considering families separately, the number of experimental results is, in my opinion, way too low to apply any confidence level on the results (except for Mytilidae maybe)."
We fully agree with the reviewer that for some families, life stages, and combination of stressors the sample size is indeed low, but ultimately this reflects the current state of the experimental literature and is something that we point out as a major conclusion and recommendation for future work. We highlight the abundance of experiments focusing on pH as a literature bias relative to other stressors. We did not perform meta-analysis specifically on one experimental set up. Upon reviewing the code we provide in addition to our data, the reviewer can see that we used ‘stressor’ as a grouping moderator in the meta-analysis. (Code line 186 families_experiment_counts <- map(list_of_families, function (family) metadata_bivalves |> filter(Family == family) |> group_by(Stressor) |> summarize(count = n()))). This approach does not perform meta-analysis on any stressor individually, but places it within the context of all others, as a work around to biases in experimental data.
We recognise that indicating statistical significance on small sample sizes in instances of low data limits the confidence that can be applied to the specific cases, but within the context of the meta-analytical approach our conclusions are still valid. We specify in the paper that we are just working with what is available in the data. Our hope is that by drawing attention to these deficits and biases, it will encourage experimental work on a wider range of bivalves and climate stressors.
"I have already mentioned that that because of important differences in the protocols used between the different datasets, the low number of datasets in many cases etc.., I believe that the results presented here have a low confidence level, I will push a bit further, and will take 2 examples showing that I even believe using this method in such conditions can lead to false conclusions."
"First of all, L168-169: “The effect of pH + salinity is intermediate between that of the two single stressors, dampening the salinity effect”. This assessment is based on the observation that salinity (8 datasets, increase, decrease of salinity???) has a stronger effect than pH+S (2 datasets). These 2 datasets are from the work of Dickinson (Dickinson et al., Interactive effects of salinity and elevated CO2 levels on juvenile eastern oysters, Crassostrea virginica. Journal of Experimental Biology 215, 29-43 (2012); Dickinson, O. B. Matoo, R. T. Tourek, I. M. Sokolova, E. Beniash, Environmental salinity modulates the effects of elevated CO2 levels on juvenile hard-shell clams, Mercenaria mercenaria. Journal of Experimental Biology 216, 2607-2618 (2013)). On oysters, they found that each factor leads to decreased growth (soft tissue) but that there is no interaction between these factors. On clams, they also showed that low salinity combined with hypercapnia leads to decrease in growth as compared to the growth measured following exposure to one of these stressors in isolation. So, we are far for the finding that a decrease in pH leads to a dampening of salinity effect…."
We fear our point might have been misunderstood by our choice of language regarding the sentence “The effect of pH + salinity is intermediate between that of the two single stressors, dampening the salinity effect”. We will restructure the text to remove “dampening” and just point out that its intermediate between the two other stressors. But again, our approach with meta-analysis was not to drill down to specific results, but try and place them within the wider narrative of the collective experimental data we extracted and analyses.
While we consider the inclusion of the salinity information important, we do not place large narrative importance on the impacts of salinity due to the low number of experimental studies. Our suggestion to restructure the text that referred to “dampening” will aid this. The point of our text is to draw attention to the state of the published data, while also emphasising the ‘direction of travel’ for bivalves as a group based on these data. We like to convey that salinity is a lesser explored stressor but may have interesting interactions with others, e.g., pH, temp, and oxygen.
"The second example is connected to the main conclusion of the article: L367-368: “We demonstrate that synergistic combinations of stressors (e.g., effects of combined temperature + O2 + pH change) cause greater reductions in bivalve growth then individual stressors.”. This assessment is based on 4 experimental setups, but from one single publication (Stevens, C. J. Gobler, Interactive effects of acidification, hypoxia, and thermal stress on growth, respiration, and survival of four North Atlantic bivalves. Marine Ecology Progress Series 604, 143-161 (2018)). They indicated that for 4 species of bivalves, “The combination of low DO and low pH often interacted antagonistically to yield growth rates higher than would be predicted from either individual stressor (…) Elevated temperature and low pH interacted both antagonistically and synergistically, producing outcomes that could not be predicted from the responses to individual stressors. Collectively, this study revealed species- and size-specific vulnerabilities of bivalves to coastal stressors along with unpredicted interactions among those stressors». Again, we are far from a “greater reduction in growth than individual stressors” with a much more complicated story."
We disagree with the reviewer who clearly rejects the approach of meta-analysis. Our conclusions map across with the most directly comparable work (Sampaio et al). Most importantly, our point is not in sole reference to Stevens and Gobler (2018) but refers to the fact that adding different climate stressors amplifies negative responses as shown in detail in our results. The statement is derived from the combined temperature, O2, and pH change detected in our work. We fully agree that the “the story may indeed be more complicated”, the idea of a meta-analysis is to find commonalties and we are confident that our work adds to the understanding.
Citation: https://doi.org/10.5194/egusphere-2023-287-AC2
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 526 | 225 | 27 | 778 | 18 | 17 |
- HTML: 526
- PDF: 225
- XML: 27
- Total: 778
- BibTeX: 18
- EndNote: 17
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Rachel A. Kruft Welton
Daniela N. Schmidt
James D. Witts
Benjamin C. Moon
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(988 KB) - Metadata XML