the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Jul 2026
Biological modulation of shell δ13C in Mytilus galloprovincialis and Phorcus sp.: Divergent responses between CO2 vents and laboratory mesocosms
Abstract. Stable isotopes are fundamental proxies for deciphering past environmental conditions and carbon cycle perturbations. However, applying these geochemical tools with a forward-looking perspective can also help predict the resilience of marine biota in future acidified oceans. This study investigates carbon uptake and biomineralization pathways under reduced pH in two ecologically keystone calcifying species: the bivalve Mytilus galloprovincialis and the gastropod Phorcus sp. To unravel these mechanisms, we adopted a dual approach, comparing isotopic trajectories (δ13C) under strictly controlled laboratory mesocosms with observations from natural CO2 vent systems in Ischia, Italy. Our results reveal a striking discrepancy between settings. Under laboratory conditions, both species exhibited a significant depletion in shell δ13C as pH decreased, directly tracking seawater carbonate chemistry. Conversely, specimens from the volcanic vents displayed a paradoxical δ13C enrichment at low-pH sites (pH ~7.4). These divergent trajectories indicate that in complex natural environments, the geochemical signal of ocean acidification can be overridden by metabolic overprints and trophically driven vital processes. Consequently, while laboratory experiments are essential to isolate kinetic fractionation, natural analogues remain crucial to capture the biological resilience and complex ecological feedback of future oceans. Future research combining these geochemical trends with direct, in vivo physiological assessments will be essential to precisely constrain the underlying metabolic kinetics and refine predictive metabolic models.
- Preprint
(1159 KB) - Metadata XML
- BibTeX
- EndNote
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2026-4016', Anonymous Referee #1, 10 Aug 2026
-
AC1: 'Reply on RC1', Annalisa Capasso, 02 Sep 2026
We thank the referee #1 for the rigorous and constructive evaluation of our manuscript. The issues raised identify substantive concerns that we have taken very seriously. In response, we have carefully re-examined our raw data and analysis scripts, discussed the overall structure of the study with all co-authors, and reviewed and clarified the information concerning sampling permits for the Marine Protected Area.
We address the reviewer's comments individually below.
Structural problems in the experimental design and in the comparability of the field and laboratory arms:
We accept and agree with the reviewer's criticism regarding the differences between the laboratory and field arms of the study, such as source populations, life stage, exposure duration and history, and diet. We recognise these as important limitations of the original framing and we worked to address them throughout the revision. Rather than presenting the laboratory and field datasets as a direct, mechanistically interpretable comparison, we now present them as separate case studies, each examining whether and how shell δ¹³C in marine calcifiers tracks seawater carbonate chemistry under ocean acidification, under two different environmental contexts (controlled laboratory conditions and natural CO₂ vent systems). We believe this reframing more accurately reflects what the data can support. We have revised the Title, Abstract, Introduction, and Discussion accordingly.
We propose revising the title to: “Shell δ¹³C responses to ocean acidification across contrasting biological and ecological contexts”
R) Regarding the pH data, following the reviewer’s comment, we were able to retrieve the continuous pH records collected at the deployment stations throughout the 2024 field experiment. We have therefore incorporated these data into the revised manuscript and added a new Supplementary Table reporting the mean ± SD and range of pH recorded at each station over the deployment period. This information provides a more robust characterization of the actual pH conditions experienced by the organisms during the field exposure.
Regarding the absence of TA, DIC, pCO₂, or Ω measurements: the manuscript, as submitted, characterises the acidification gradient only in terms of pH classes and we rely on biogeochemical characterization available for each provided in the previous studies.
For Castello Aragonese vent system, we draw on as reference at the well-established biogeochemical characterization, derived from continuous in-situ monitoring reported in previous studies (Hall-Spencer et al., 2008 - dataset: PANGAEA, https://doi.org/10.1594/PANGAEA.728719); Hofmann et al., 2011 [https://doi.org/10.1371/journal.pone.0028983]; Kroeker et al., 2011 [https://doi.org/10.1073/pnas.1107789108]; Foo et al., 2018; Gambi et al., 2020 [Art7]; Teixidò et al., 2018, 2024 [https://doi.org/10.1111/gcb.17105]) some of these already cited in the manuscript. These confirm the pattern already assumed in the manuscript (L104–106): total alkalinity is essentially stable across the gradient, while pH decline is driven by DIC/pCO₂ increase.
For Vullatura, published characterization is more limited than for Castello Aragonese, but not entirely absent (Gambi et al., 2019/2020; Mecca et al., 2020 [https://doi.org/10.12681/mms.20795]; Lee et al., 2022 [https://doi.org/10.1111/gcb.16265]), and a three-zone classification consistent with our own (Ambient 8.2–8.0; Low 7.8 ± 0.3; Extremely low 7.2 ± 0.5 - Low and Extremely low categories correspond respectively to our own Medium and Low classes; Barruffo et al., 2021 [https://doi.org/10.3390/d13070311]). We have added these citations and values at the paragraph “2.1.1 Study Sites and Sample collection” and to the relevant supplementary table.
These are historical characterization data for the same vent systems, not measurements taken during our own 2024 deployment, and vent chemistry at these sites is known to fluctuate both seasonally and sub-daily. They therefore can’t substitute for site- and time-specific measurements, and we no longer imply that our nominal pH classes are equivalent to a full carbonate system characterization of the 2024 field experiment. We agree that full, deployment-specific carbonate system characterization, for both sites, especially for Vullatura, remains a priority for future work, and we now state this explicitly as a limitation in the Discussion: in the absence of a directly measured characterization of the surrounding carbonate chemistry, the interpretation of our results based on carbonate system dynamics should be considered accordingly constrained.
R) Regarding the taxonomic indeterminacy of Phorcus sp., we thank the reviewer for raising this important point. We agree that genus-level identification does not allow us to unequivocally assign all sampled individuals to a single Phorcus species.
Photographic documentation of the collected individuals is available; however, after re-examination this material, we concluded that species-level identification based exclusively on shell morphology would not be sufficiently reliable, particularly considering the morphological plasticity of the genus and the degree of shell erosion observed at the vent sites. Unfortunately, no soft tissue samples were preserved for subsequent molecular identification. We have therefore revised the manuscript to explicitly acknowledge this taxonomic uncertainty and to avoid species-specific interpretations of the observed patterns. In particular, the inter-individual variability in shell δ¹³C can no longer be attributed exclusively to individual physiological plasticity or distinct calcification strategies, as taxonomic heterogeneity may also contribute to this variability. We have revised Section 4.2 to explicitly flag this alternative explanation.
Similarly, we have revised our interpretation of the lower abundance and smaller size of individuals observed at the most acidified station. Rather than interpreting this pattern as evidence of a demographic response to acidification, we now acknowledge that differences in the occurrence and distribution of Phorcus taxa along the environmental gradient may also contribute to the observed pattern. Throughout the revised manuscript we therefore treat the field material conservatively as Phorcus sp. and restrict our conclusions to patterns observed at the genus-level assemblage represented by our samples. We do not attempt to attribute these patterns to the response of any particular Phorcus species. This taxonomic uncertainty is now explicitly acknowledged as a limitation when interpreting the biological mechanisms underlying the observed isotopic variability.
R) Regarding the pseudoreplication point: we have re-analyzed both the mussel and Phorcus laboratory experiment datasets accordingly, as detailed below.
Mussel laboratory experiment. We re-fitted the model as nested design, with Tank as a random effect (linear mixed-effects model: δ¹³C ~ Condition + (1|Tank), fitted via "lme4/lmerTest" packages), in which each tank belongs to a single condition and is therefore not crossed with it. We additionally cross-checked this against the simplest possible correction, aggregating each tank to its mean δ¹³C and analyzing the resulting n = 3 vs. n = 3 tank means as experimental unit, following the recommendation of Harrison et al. (2018) [https://doi.org/10.7717/peerj.4794] for designs with very few clusters, since mixed-model variance-component estimates can be unstable when the number of random-effect levels is this small. Both approaches confirm that the treatment effect remains robust once pseudoreplication is corrected: the mixed model yields F₁,₃.₉ = 160.3, p = 0.0002 (Satterthwaite-approximated degrees of freedom), and the tank-aggregated analysis yields F₁,₄ = 175.2, p = 0.0002 (control mean = −2.51‰, treatment mean = −4.11‰, both expressed as means of tank means), compared to the original pseudoreplicated result of F₁,₇₃ = 169.7. We have replaced the original one-way ANOVA in the Methods and Table 1 with this tank-corrected analysis.
Phorcus laboratory experiment. We have re-analyzed the Phorcus laboratory dataset accordingly, following the same tank-corrected approach applied to the mussel laboratory experiment (mixed-effects model with Tank as a random effect, δ¹³C ~ Condition + (1|Tank), cross-checked against a tank-aggregated t-test on the n = 3 vs. n = 3 tank means). Unlike the mussel dataset, where both corrected approaches converged on the same conclusion, here the two approaches disagree, and we report this transparently rather than selecting the more favourable result. The mixed model returns a non-significant effect (F₁,₁₅ = 3.01, p = 0.103) and converges to a singular fit, with the between-tank variance component estimated at zero; the tank-aggregated t-test, in contrast, returns a significant effect (t₄ = 7.42, p = 0.0018, control mean = −1.60‰, treatment mean = −2.14‰, both as means of tank means).
We investigated the source of this disagreement and traced it to the small and uneven number of individuals per tank in this experiment (2–4 individuals, compared to 20 in the mussel experiment), and in particular to one tank (Treatment replicate 1, n = 2) with very high internal dispersion between its two individuals. With only three tanks per condition, the mixed model cannot reliably distinguish genuine between-tank variability from the noise contributed by this single, sparsely sampled tank. The tank-aggregated test, by construction, does not attempt to estimate this variance component and is therefore more stable, but it is correspondingly more sensitive to the fact that tank means themselves are based on as few as two individuals. We have revised the manuscript to present the Phorcus laboratory result descriptively, reporting both corrected analyses and explicitly stating that the tank-level replication in this experiment is insufficient to draw a statistically robust conclusion about the treatment effect. We have added this as an explicit limitation in the Discussion.
R) We thank the reviewer for highlighting this point and we agree that the role of San Pietro in the study required clearer description. San Pietro has already been used as an external ambient-pH reference site in several previous studies conducted at the Castello Aragonese vent system, based on its comparable hydrodynamic conditions, temperature and salinity (Ricevuto et al., 2015a [http://dx.doi.org/10.1016/j.marenvres.2015.09.005]; Migliaccio et al., 2019 [https://doi.org/10.1016/j.scitotenv.2019.04.005]; Palombo et al., 2023 [https://doi.org/10.1016/j.marpolbul.2023.114700]; Signorini et al., 2025 [https://doi.org/10.1016/j.envres.2025.121874];). We have added these references at and revised the corresponding text to make clear that the environmental comparability of San Pietro is supported by previous studies rather than by the shell δ¹³C measurements obtained here. We nevertheless acknowledge the reviewer’s point that site-specific carbonate-system measurements, including δ¹³C_DIC, are not available for San Pietro. We have therefore revised the manuscript to avoid implying that San Pietro is independently validated as equivalent to the Castello Ambient station in terms of carbonate chemistry. Instead, we identify it explicitly as an external ambient-pH reference site for the Phorcus dataset and acknowledge the absence of contemporaneous carbonate-system and δ¹³C_DIC measurements as a limitation when interpreting the isotopic patterns.
Finally, regarding the statistical test cited in the manuscript (F₁,₁₂ = 0.45, p = 0.514, San Pietro vs. Castello Ambient shell δ¹³C): we agree that the absence of a significant difference in shell δ¹³C between San Pietro and Castello Ambient cannot independently demonstrate environmental equivalence between the two sites. We have therefore removed this interpretation from the revised manuscript and now present San Pietro strictly as an external site with prior-documented hydrodynamic/physical comparability.
San Pietro was used exclusively as an external reference site for the Phorcus field dataset and was not part of the mussel transplantation experiment, which instead included the Castello Aragonese and Vullatura vent systems. We have clarified which sites pertain to which dataset in the Methods, and revised Fig. 1 and its caption to describe San Pietro explicitly and to assign it a distinct symbol, differentiating it visually from the ambient stations of the vent gradient. The revised figure is included in the Supplementary Material.
R) Regarding the shell mineralogy: we acknowledge that it was not directly characterised in this study. Shells of M. galloprovincialis and Phorcus sp. were consistently collected from the growing shell margin and the shell aperture, respectively, following the same powder-drilling procedure described in the Methods. The resulting powders therefore represent an integrated signal from the mineral phases present in these shell regions (a mixture of calcite and aragonite for the bimineralic growing margin of M. galloprovincialis, and aragonite alone for the exclusively nacreous shell aperture of Phorcus sp.). The standardised sampling protocol should have minimised differences in mineralogical composition among samples; if the relative proportions of the two mineral phases remained constant, their different carbon-isotope fractionation factors would primarily produce a systematic offset rather than affect the observed δ¹³C trends.
-
AC1: 'Reply on RC1', Annalisa Capasso, 02 Sep 2026
-
RC2: 'Comment on egusphere-2026-4016', Anonymous Referee #2, 14 Sep 2026
General assessment
The manuscript addresses a potentially interesting topic, namely the effects of ocean acidification on Mytilus galloprovincialis and Phorcus sp., using the carbon isotopic composition of shells combining two approaches: from natural CO₂ vent systems and from laboratory experiments. However the manuscript presents substantial methodological, experimental, and statistical limitations that compromise the robustness of the discussion and conclusion. In particular, the comparison between organisms belonging to inadequately characterized taxonomic groups, different exposure approaches, different size classes, non-comparable exposure times, and different environmental conditions introduces numerous sources of variability that are not adequately controlled. These issues are further compounded by important gaps in the description of the methods, potential pseudoreplication in the laboratory experiments, and an interpretation of the results that, in several instances, goes substantially beyond what the data can demonstrate. I also consider the lack of species-level identification of the Phorcus specimens and the use of individuals from natural populations without excluding differences in population composition among sites to be particularly problematic. Indeed, at least 3-4 species belonging to the Phorcus genus are present in the Mediterraenan Sea. This issue is especially relevant considering that the authors themselves emphasize in the Introduction the importance of species-specific responses to acidification. Overall, the issues identified do not appear to be resolvable through a simple revision of the manuscript or the Discussion. Some of them would require new experimental information, improved characterization of the organisms and study sites, and a substantial revision of the statistical approach. For these reasons, I recommend rejection of the manuscript.
Main concerns
Materials and Methods – Natural acidified conditions
The comparison between Mytilus galloprovincialis and Phorcus sp. does not allow any robust biological conclusions. The authors compare a single species, M. galloprovincialis, with a group of organisms belonging to the genus Phorcus, without providing species-level taxonomic identification of the Phorcus specimens collected in the field. This approach is problematic because it prevents the assessment of species-specific responses to acidification. Considering that many Phorcus species occur in the Mediterranean, the presence of different species at different sites could introduce a substantial source of biological variability and represent an important bias in the comparison. It is possible that the difficulty of taxonomically identifying Phorcus specimens motivated this choice, whereas M. galloprovincialis is relatively easy to identify. However, this methodological difficulty cannot be considered negligible, as it fundamentally compromises the comparability of the organisms used in the study.
The Mussel Watch field experiment is also insufficiently described. The manuscript does not clearly state how many cages were used at each site, how many animals were present in each cage, or whether mortality was monitored throughout the experiment. Furthermore, considering that mussels are filter-feeding organisms, the characteristics of the cages and their degree of permeability to water should be described, particularly to assess whether individuals located in the inner parts of the cages had comparable access to water flow. Moreover, the depth at which sampling was conducted at San Pietro is also not reported, and it is unclear whether it was comparable to the depths at Vullatura and Castello Aragonese. Although the authors state that the San Pietro site has biogeochemical parameters comparable to those of the ambient areas at Castello and is unaffected by volcanic inputs, this statement is not supported by any biogeochemical characterization presented in the manuscript. The authors describe pH, DIC, and TA gradients, but the actual values measured at different sites are not reported.
Another important gap concerns the origin of the M. galloprovincialis individuals used in the transplant experiment. The authors emphasize that the two species have different ecological histories; however, the origin of the mussels used in the field experiment is not adequately described. It is essential to specify which population the mussels originated from and whether they were collected from the wild, particularly if they were obtained from a site not shown on the map of the study area. This information is important because the previous environmental history of the transplanted individuals may influence their physiological and isotopic responses during the experiment. There is also insufficient information regarding the initial size of the animals and their growth during the 60-day exposure period. Mussel size is particularly important because it is directly related to growth and therefore to the amount of carbonate deposited during the experiment. It would therefore be important to know whether animal growth was measured and to what extent the environmental conditions allowed normal growth.
The statement that the Phorcus individuals represent populations chronically exposed to acidified conditions “from the larval stage through the juvenile, sub-adult, and adult phases” is particularly problematic. This statement cannot be demonstrated based on the data presented. Given the geographical proximity of the different sites, larval dispersal or recruitment from other areas cannot be excluded. No evidence is provided to demonstrate that the individuals were born and spent their entire life cycle at the site where they were collected. Furthermore, only three individuals were collected at some sites. Such a low sample size appears insufficient to represent the natural variability of a population, particularly when the authors collected up to 15 individuals at other sites. I understand the difficulty of obtaining large sample sizes from natural populations; however, with such a low “n”, it is impossible to exclude the possibility that the sampled individuals were transient rather than representative of a resident population. This issue is further aggravated by the absence of species-level taxonomic identification. Finally, the two species were not sampled at the same sites: M. galloprovincialis was used across the different sites of the gradient, whereas Phorcus was absent from Vullatura. This further complicates the overall comparison between the two organisms.
Materials and Methods – Laboratory experiments
The comparison between the natural and laboratory experiments also presents important limitations. The animals used in the two approaches differ in size, exposure duration, environmental conditions, and probably origin (lacking information). In the case of M. galloprovincialis, the laboratory animals were juvenile specimens (mean length of approximately 1.5 cm) collected from the northern Adriatic coast near Chioggia, whereas the organisms used in the field experiment were probably adults (again no info available). Size therefore represents another potentially important variable that could explain differences between the two experiments. Moreover, the two environmental seawaters used can be highly different, one from Nort Italy from a lagoon and the other from South Italy near an island, performed also in different years. The manuscript does not report that the water parameters used in the laboratory were comparable to those present in the natural environment at Ischia. Differences in water chemistry, salinity, temperature, nutrient availability, and other environmental parameters could contribute to the differences observed. The exposure durations also differ: approximately 60 days in the field experiment and 90 days in the laboratory experiment for M. galloprovincialis, whereas the laboratory experiment with Phorcus lasted eight weeks. These differences make impossible a rigorous comparison between the different experiments. The laboratory experiments also lack sufficient information on the actual amount of food available, and, in the case of mussels, the concentration of microalgae supplied ( cells/mL). Similarly, the actual pH values measured in the different treatments are not adequately reported. It is therefore unclear whether the nominal values set by the system actually corresponded to the conditions to which the animals were exposed throughout the experiment. For Phorcus, the presence of several species within the genus makes taxonomic identification of individuals from the different populations particularly important. Comparing populations from southern and northern Italy without verifying their species identity may introduce additional variability into the dataset and makes it difficult to attribute the observed differences specifically to acidification.
Isotopic analyses
The description of the shell sampling procedures is insufficient to ensure study reproducibility. It is unclear how many individuals were selected from each cage, how they were selected, and which specific portion of the shell was analyzed. Although the authors generally identify the growing shell margin for M. galloprovincialis and the aperture for Phorcus, they do not provide sufficiently detailed information on the size and precise location of the analyzed shell portion. For example, it is unclear how the newly deposited shell portion was identified, whether it represented a defined percentage of the total shell length, and whether an initial shell size measurement was available before the experiment. Did the authors use image analysis to quantify the newly deposited shell portion? Similar information is missing for Phorcus. These details are particularly important for the laboratory experiments, where the objective is to interpret the isotopic signal as a consequence of the experimental exposure. For the field experiment, it is also difficult to understand how the shell portion corresponding to growth during the 60-day exposure period was identified in a standardized manner. The statement “All samples consisted of multiple shell specimens” is particularly problematic. It is unclear whether “sample” refers to a single isotopic sample obtained by pooling several shells or to multiple shells analyzed separately. This distinction is fundamental because the two approaches result in completely different experimental units and sample sizes. If several shells were pooled into a single sample, inter-individual variability cannot be assessed. If the shells were analyzed individually, the number of individuals analyzed and how these measurements were subsequently used in the statistical analysis must be clearly reported. As currently written, this statement makes it impossible to determine the actual n used in the analyses and the true experimental units. This is an important issue both for study reproducibility and for the correct interpretation of the statistical results.
Statistical analysis
The main statistical concern is pseudoreplication. In the laboratory experiments, multiple individuals were maintained within the same tank, yet the statistical analyses appear to treat individual animals as independent observations. However, the experimental unit is the tank rather than the individual animal. Individuals maintained within the same tank share the same experimental conditions and therefore cannot be considered fully independent replicates of the treatment. In the mussel experiment, 20 individuals were maintained in each tank, with three replicate tanks per treatment. Therefore, the effective experimental sample size is n = 3 tanks per treatment, rather than the total number of individuals analyzed. The reported ANOVA degrees of freedom (df = 1,73), however, suggest that individual mussels were treated as independent observations. This may substantially inflate the degrees of freedom and result in artificially high statistical significance. The same concern applies to the Phorcus laboratory experiment. The analysis should therefore account for the hierarchical structure of the experimental design, either by considering the tank as the experimental unit or by using an appropriate mixed-effects model with tank included as a random effect. Treating individuals within the same tank as independent replicates is not appropriate for this experimental design.
Discussion – Natural acidified conditions
Section 4.1 presents a potentially interesting result, namely the enrichment in δ13C observed in Mytilus galloprovincialis shells at low-pH sites. However, I believe that the interpretation proposed by the authors goes substantially beyond what the data can demonstrate. In particular, the conclusion that biological modulation is capable of “overriding” the ambient geochemical signal is not adequately supported by the dataset. The main issue is the absence of δ13C-DIC measurements along the environmental gradient and of measurements of the main organic carbon sources. Without these data, it is not possible to distinguish the contribution of ambient DIC from that of metabolic carbon or to quantify the relative importance of the different carbon sources. Consequently, the statement that respiratory CO₂ represents the dominant carbon source for calcification remains purely speculative. Similarly, the hypothesis that acidification induces prolonged valve closure, isolation of the extrapallial fluid, and consequently an increased metabolic contribution has not been experimentally tested. No behavioral, physiological, respiratory, or calcification measurements are available to verify this mechanism. The section concerning primary productivity is also particularly problematic. The statement that photosynthesis produces biomass enriched in 13C that is subsequently assimilated by the organisms is not adequately supported. Again, measurements of the δ13C of organic matter and its sources are lacking, making it impossible to substantiate the proposed mechanism. Similar caution is required for the ontogenetic interpretation of Phorcus sp., where the δ13C pattern is attributed to a progressive reduction in vital effects without independent measurements of growth or calcification rates. Several alternative explanations remain possible, including changes in diet, metabolism, microhabitat, and seasonality. Finally, I do not consider the statement that seasonal variability in δ13C-DIC cannot explain the observed pattern to be sufficiently justified, since this variable was not directly measured at the study sites. Comparison with values reported in the literature is not sufficient to exclude this possibility.
Overall, an isotopic observation is used to construct a highly specific physiological and biogeochemical explanation without measuring the variables required to test the proposed mechanisms.
Discussion – Laboratory experiments
The interpretation of the laboratory experiment is also more speculative than the data allow. The authors observe a decrease in δ13C in Mytilus galloprovincialis under reduced pH and interpret this result as a direct geochemical response to the reduction in CO₃²⁻ availability. However, the experiment does not allow the authors to identify CO₃²⁻ reduction as the specific cause of the observed isotopic change. Lowering pH by adding CO₂ affects several components of the carbonate system simultaneously, and the study does not provide evidence that the observed δ13C shift can be specifically attributed to changes in CO₃²⁻. The authors further conclude that, under controlled laboratory conditions, the shell acts as a relatively passive recorder of environmental chemistry. However, controlled conditions do not eliminate physiological effects. The mussels were still actively feeding, growing, respiring, and calcifying, and no measurements of growth, calcification, respiration, or metabolic activity were provided. Therefore, the study cannot determine whether the observed isotopic change was driven primarily by environmental carbonate chemistry, physiological responses, or a combination of both. The data support an association between reduced pH and changes in shell δ13C, but they do not demonstrate the specific mechanism proposed by the authors. The interpretation of Phorcus sp. is even weaker. A non-significant trend is used to support the existence of high physiological plasticity, different calcification strategies, changes in energy allocation, and stochastic calcification events. None of these hypotheses is directly supported by measurements of growth, metabolism, respiration, or calcification. Inter-individual variability alone does not demonstrate physiological plasticity or different calcification strategies. The laboratory result for Phorcus is in fact not statistically significant (p = 0.144), and therefore cannot be used to support an isotopic response to acidification. Likewise, the significant treatment effect observed in natural Phorcus populations cannot simply be interpreted as an effect of acidification because the individuals originate from different natural populations, the species were not taxonomically resolved, sample sizes can be variable, and several environmental factors may be confounded with pH. The strong ontogenetic effect observed in the Phorcus data (F₂,₃₉ = 29.67; p < 0.001) further demonstrates that δ¹³C is strongly influenced by growth stage. However, the criteria used to assign individuals to the different ontogenetic stages (juvenile, sub-adult, and adult) are not described. It is unclear whether these categories were based on shell size, morphological characteristics, or another criterion. This information is particularly important because shell size and growth rate may directly influence the isotopic signal. Moreover, without independent measurements of growth or calcification rates, the subsequent interpretation in terms of changes in vital effects remains speculative.
Overall, also in this section the authors transform descriptive isotopic patterns into untested mechanistic explanations. The experiments demonstrate, at most, an association between acidification and changes in shell carbonate δ13C; they do not demonstrate either predominantly geochemical control of the signal in Mytilus or metabolic control in Phorcus.
Conclusions
The methodological and statistical issues described above, considered together, substantially compromise the robustness of the conclusions of the manuscript. The problems are not limited to the overinterpretation of some results but also concern the difficulty of reliably identifying the experimental units, assessing their independence, determining the taxonomic composition of the Phorcus populations, verifying the actual exposure conditions, and identifying the sources of carbon incorporated into the shells. These issues cannot be resolved through a simple revision of the text or a reformulation of the Discussion. Additional experimental information and, for some of the main issues, new data would be required to adequately address these concerns. Furthermore, although the authors compare natural and laboratory approaches to study ocean acidification, the two experiments differ in several additional factors, including species, animal size, origin, and exposure duration. Therefore, differences in isotopic responses cannot be attributed specifically to the experimental approach (natural versus laboratory), as they may instead result from these uncontrolled differences. Overall, I consider these limitations to be fundamental to the experimental design and interpretation of the results. Therefore, I recommend rejection of the manuscript in its present form.
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2026-4016', Anonymous Referee #1, 10 Aug 2026
General comments
This manuscript addresses a relevant and interesting question: whether shell δ¹³C in marine calcifiers tracks seawater carbonate chemistry under acidification, and whether laboratory and natural CO₂ vent systems yield consistent isotopic responses. The paired vents/mesocosm comparison on two taxa with contrasting life histories, complemented by an ontogenetic component in Phorcus, is in principle a valuable approach, and the underlying dataset has genuine documentary value for the Ischia vent system. I therefore consider the scientific significance of the topic to be good.
However, I regret to conclude that the manuscript in its current form suffers from major deficiencies in scientific quality that cannot be remedied through revision. My concerns fall into two categories: 1: structural problems in the experimental design and in the comparability of the field and laboratory arms, and 2: serious issues of internal consistency and traceability of the reported results.
Regarding (1), the central claim of the paper, that biological/trophic processes override the geochemical signal in complex natural environments, requires that the field and laboratory arms differ only in ecological complexity. Instead, the two approaches differ simultaneously in: source populations (Adriatic vs. Tyrrhenian; the origin of the transplanted mussels is never stated), life stage (juvenile mussels in the lab vs. adults in the field; lifetime-exposed native Phorcus vs. 8-week exposed lab adults), exposure duration and history, diet (cultured microalgae of unmeasured δ¹³C for lab mussels vs. natural seston; field-collected epiphytes for lab gastropods), thermal regime (constant 28.5 °C vs. autumn cooling in situ), and pH regime (stable setpoint vs. the highly fluctuating exposure characteristic of vent systems). In addition the pH logger data are mentioned but never shown. The observed lab/field divergence is thus overdetermined: it is compatible with at least half a dozen explanations besides the trophic/metabolic override favoured by the authors. Critically, the interpretation invokes precisely the variables that were not measured in either system (δ¹³C of DIC, POM, and diet), as the authors themselves acknowledge (L308–311). A descriptive juxtaposition of the two datasets would be defensible; the mechanistic conclusions drawn in the Abstract and Discussion are not supported by this design.
In addition, three structural problems are not easily fixable in revision:
1) Absence of carbonate system characterization. No measurements of TA, DIC, pCO₂, or Ω are reported for either the laboratory treatments or the field sites; only nominal pH values are given, and the deployment logger data are not presented. For an ocean acidification study whose entire interpretive framework rests on DIC pool dynamics, this falls short of established best practices, and the field chemistry of the 2024 deployment cannot be reconstructed retroactively.
2) Taxonomic indeterminacy of Phorcus sp. The Mediterranean hosts four Phorcus, at least three of which (P. turbinatus, P. richardi, P. mutabilis) co-occur in the Tyrrhenian and segregate primarily by shore height. The sampling protocol ("from the lower intertidal to the upper subtidal") targets exactly the zone of maximum species overlap. Consequently: (a) the pronounced inter-individual variability interpreted as physiological plasticity and "distinct calcification strategies" (Section 4.2) is indistinguishable from interspecific variance; (b) since species composition may co-vary with the pH gradient (microhabitat and depth change along it), the treatment effect is confounded with a possible taxon effect; (c) even the reported scarcity and reduced size at the Low pH station (L137–139) could reflect species turnover rather than a demographic response. Shell morphology in this genus is notoriously plastic and vent-induced corrosion further degrades its diagnostic value. Unless voucher tissues or photographic documentation of live animals are available for identification or DNA barcoding, this confound cannot be resolved. I ask the authors to state explicitly whether such material exists.
3) Pseudoreplication. In the mussel laboratory experiment, three treatment and three control tanks held 20 individuals each, yet individuals are treated as independent replicates (df ≈ 70) with no tank effect in the models. For the Phorcus lab experiment, the number of tanks is never stated, and the Welch test df (7.35) imply a very small effective sample. A mixed-model reanalysis with tank as a random effect would be the minimum requirement, but the design itself (effectively n = 3 tanks per treatment, and possibly fewer for gastropods) remains weak.
Regarding (2), traceability: the manuscript reports two irreconcilable statistics for the same analysis (one-way ANOVA on lab mussels: F₁,₆₉ = 34.79 in the text at L227 vs. F = 169.7, df = 1, 73 in Table 1), inconsistent pH values between Methods and Discussion (8.2/7.7 vs. ~8.0/~7.5), and three mutually incompatible statements of experimental duration (August–October ≈ 60 days; "3 months"; "90 days"). Furthermore, the Abstract states that "both species exhibited a significant depletion" in the laboratory, whereas the Phorcus result is explicitly non-significant (p = 0.144, L265–268), and the "paradoxical enrichment at low-pH sites (pH ~7.4)" was statistically significant only at Vullatura for mussels, while Phorcus was never sampled at Low pH at all. The Abstract thus presents a substantially cleaner pattern than the data support. Individually, some of these could be typographical; collectively, they undermine confidence in the analysis pipeline. I encourage the editor and authors to make the raw data and R scripts available for verification, as offered in the Code and Data Availability statement.
In addition, San Pietro reference site (L107–109): The site is stated to "share comparable biogeochemical parameters with the Castello's ambient zones", but no comparative data are shown or cited to support this claim. No TA, DIC, pCO₂, Ω, temperature, or salinity measurements for San Pietro are reported anywhere in the manuscript, nor, critically for this study, any δ¹³C DIC characterization. The validity of San Pietro as an external control for the Phorcus dataset rests entirely on this undocumented assertion. Note also that the non-significant difference in shell δ¹³C between San Pietro and Castello Ambient (F₁,₁₂ = 0.45, p = 0.514; L245–246) cannot by itself validate the site as a control, since with n as low as 3 per site the test is severely underpowered, and shell δ¹³C is the response variable, not an independent verification of water chemistry. Please provide the environmental data supporting the comparability claim, or cite the specific source from which it derives. Relatedly, in Fig. 1 the westernmost marker (San Pietro) is shown in the map and legend but is never described in the caption, which mentions only Castello Aragonese and Vullatura; please describe San Pietro in the caption, clarify which sites pertain to which dataset (mussel transplants: Castello and Vullatura; Phorcus collections: Castello and San Pietro), and consider a distinct symbol/color for San Pietro, as it is an external reference site rather than an ambient station of the vent gradient.
Finally, The Castello Aragonese vent system falls within the 'Regno di Nettuno' Marine Protected Area, where the collection of marine fauna requires authorization from the managing body. The manuscript does not mention any collection or sampling permits for the field activities (mussel cage deployment at Castello Aragonese and Vullatura; Phorcus collection at Castello Aragonese and San Pietro). Please state the permits obtained and the issuing authority.
Specific comments
- L78: The study is described as "applying a stable isotope labelling approach". No isotopic labelling (enrichment) was performed; all measurements are natural-abundance. This terminology is misleading and must be corrected throughout.
- L96: Among the references cited for the biogeochemical characterization of the sites, Gambi et al. (2026) concerns Posidonia dwarf morphotypes and does not appear pertinent here.
- L133–139: See general comment on Phorcus taxonomy. Please state the identification criteria used, whether tissues/photographs are available, and consider reporting the data disaggregated by putative species if identification is possible. Please also report n per site explicitly; "3 up to 15 individuals" (L137) implies that some cells of the two-way ANOVA rest on n = 3.
- L145–162: The origin of the field-transplanted mussels is never stated; the lab mussels came from Chioggia (northern Adriatic). If transplanted mussels also originated from the Adriatic, the baseline δ¹³C_DIC of source and deployment waters differ; if not, the lab and field arms used different populations. Either way this is an uncontrolled confound that should have been addressed. Additionally, 28.5 °C maintained from August to October approaches thermal stress thresholds for M. galloprovincialis and is never discussed as a potential confounder; the actual experimental temperature over the 60–90 day period is not reported.
- L158–161: The lab mussels' diet (cultured Isochrysis, Nannochloropsis, Tetraselmis, Phaeodactylum) has an unmeasured and likely atypical δ¹³C signature, while lab gastropods were fed field-collected epiphytes. Since metabolic carbon incorporation is the paper's core interpretive mechanism, the absence of dietary δ¹³C characterization in both arms is a fundamental gap.
- L164: "collected at the same site" , which site? The mussels came from Chioggia; please clarify where the Phorcus used in the laboratory experiment were collected.
- L172–187: δ¹⁸O was measured (standards and analytical precision are reported) but no δ¹⁸O data appear anywhere in the manuscript. These data are directly relevant: they would help constrain temperature effects and assess isotopic equilibrium, i.e., precisely the vital-effect framework the Discussion builds on. Please report them.
- L190–209: Please clarify whether normality tests were run on raw data or model residuals. More importantly, see the general comment on pseudoreplication: tank must enter the models.
- L214–216 and Table 1: The Vullatura ambient baseline is already significantly lower than Castello (p = 0.036); the "enrichment" at Vullatura Low pH brings values approximately back to Castello levels. Without site-specific δ¹³C_DIC measurements, interpreting this as metabolic override rather than baseline heterogeneity between sites is not warranted.
- L227 vs. Table 1: F₁,₆₉ = 34.79 vs. F = 169.7 (df = 1, 73) for the same test. Please resolve and verify the full pipeline.
- L249–253 and Table 2: A post-hoc contrast within the treatment × stage interaction ("Medium:Old vs. Ambient:Old", p = 0.026) is reported although the interaction itself is non-significant (p = 0.279). Moreover, the post-hoc labels ("Intermediate", "Adult (Young)", "Juvenile (Old)") do not match the categories defined in the text and Fig. 4b (Juvenile/Sub-adult/Adult), and the signs of the differences (−0.62, −0.90) appear inverted relative to the increasing ontogenetic trend shown in Fig. 4b. This suggests output was transcribed from software with different reference levels; please verify.
- L295–321: The ontogenetic interpretation for vent-native Phorcus conflates ontogeny with seasonality and exposure history: the apex was deposited months to years before the aperture, under different seasonal δ¹³C_DIC (which the authors themselves report varies by 0.9–1.4‰; L296–297) and possibly before recruitment into the vent site. This confound deserves explicit discussion and substantially weakens the ontogenetic conclusions.
- L327–328: pH values (~8.0/~7.5) contradict the Methods (8.2/7.7 ± 0.05). Similarly, reconcile the exposure duration (L146–147: August–October; L161: "3 months"; L226: "90 days").
- Section 4.2: The heteroscedasticity in the treated Phorcus group (Levene p < 0.001), interpreted as physiological plasticity, is equally consistent with a taxonomically mixed sample (see general comment on Phorcus taxonomy), given that lab specimens were also identified only to genus.
- Throughout: The shell mineralogy sampled is never addressed. The growing margin of Mytilus comprises both calcite and aragonite with different fractionation factors; Phorcus is aragonitic. Please specify what was sampled and discuss implications.
Technical corrections
- Fig. 5 caption: an asterisk indicating significance is described, but the reported result is non-significant; also "Phocus sp." → "Phorcus sp."
- Table 1: p < 2e-16 is flagged "**" but should be "***" per the stated convention.
- Language requires thorough revision; examples: "To achieve this issue" (L284), "two difference acidified condition" (L193), "freezed" (L140), "acclimatation" (L150, L164), "carbonatic structures" (L48).
Concluding remark. Beyond the individual issues listed above, I believe this manuscript effectively contains two studies that are not mutually comparable: a field study at the Ischia vent systems (mussel transplants and native Phorcus populations, including the ontogenetic component) and a laboratory mesocosm study. As detailed above, the two arms differ simultaneously in source populations, life stage, exposure duration and history, diet, thermal regime, pH stability and range, and habitat, that is, in essentially every variable that is isotopically relevant. The lab–field 'divergence' that frames the title and Abstract is therefore not an interpretable result but an artefact of comparing systems that were never designed to be compared. I would encourage the authors to consider developing the two datasets as separate manuscripts: a descriptive field study contributing to the well-established Ischia vent literature (contingent on resolving the taxonomic identity of Phorcus and providing site-level carbonate chemistry and δ¹³C DIC data), and a laboratory study of isotopic responses to controlled acidification (contingent on a tank-level reanalysis and characterization of dietary and seawater δ¹³C). Each dataset has merit within its own scope; neither is served by the current framing, which asks the juxtaposition to support mechanistic conclusions that the design cannot deliver.
Citation: https://doi.org/10.5194/egusphere-2026-4016-RC1 -
AC1: 'Reply on RC1', Annalisa Capasso, 02 Sep 2026
We thank the referee #1 for the rigorous and constructive evaluation of our manuscript. The issues raised identify substantive concerns that we have taken very seriously. In response, we have carefully re-examined our raw data and analysis scripts, discussed the overall structure of the study with all co-authors, and reviewed and clarified the information concerning sampling permits for the Marine Protected Area.
We address the reviewer's comments individually below.
Structural problems in the experimental design and in the comparability of the field and laboratory arms:
We accept and agree with the reviewer's criticism regarding the differences between the laboratory and field arms of the study, such as source populations, life stage, exposure duration and history, and diet. We recognise these as important limitations of the original framing and we worked to address them throughout the revision. Rather than presenting the laboratory and field datasets as a direct, mechanistically interpretable comparison, we now present them as separate case studies, each examining whether and how shell δ¹³C in marine calcifiers tracks seawater carbonate chemistry under ocean acidification, under two different environmental contexts (controlled laboratory conditions and natural CO₂ vent systems). We believe this reframing more accurately reflects what the data can support. We have revised the Title, Abstract, Introduction, and Discussion accordingly.
We propose revising the title to: “Shell δ¹³C responses to ocean acidification across contrasting biological and ecological contexts”
R) Regarding the pH data, following the reviewer’s comment, we were able to retrieve the continuous pH records collected at the deployment stations throughout the 2024 field experiment. We have therefore incorporated these data into the revised manuscript and added a new Supplementary Table reporting the mean ± SD and range of pH recorded at each station over the deployment period. This information provides a more robust characterization of the actual pH conditions experienced by the organisms during the field exposure.
Regarding the absence of TA, DIC, pCO₂, or Ω measurements: the manuscript, as submitted, characterises the acidification gradient only in terms of pH classes and we rely on biogeochemical characterization available for each provided in the previous studies.
For Castello Aragonese vent system, we draw on as reference at the well-established biogeochemical characterization, derived from continuous in-situ monitoring reported in previous studies (Hall-Spencer et al., 2008 - dataset: PANGAEA, https://doi.org/10.1594/PANGAEA.728719); Hofmann et al., 2011 [https://doi.org/10.1371/journal.pone.0028983]; Kroeker et al., 2011 [https://doi.org/10.1073/pnas.1107789108]; Foo et al., 2018; Gambi et al., 2020 [Art7]; Teixidò et al., 2018, 2024 [https://doi.org/10.1111/gcb.17105]) some of these already cited in the manuscript. These confirm the pattern already assumed in the manuscript (L104–106): total alkalinity is essentially stable across the gradient, while pH decline is driven by DIC/pCO₂ increase.
For Vullatura, published characterization is more limited than for Castello Aragonese, but not entirely absent (Gambi et al., 2019/2020; Mecca et al., 2020 [https://doi.org/10.12681/mms.20795]; Lee et al., 2022 [https://doi.org/10.1111/gcb.16265]), and a three-zone classification consistent with our own (Ambient 8.2–8.0; Low 7.8 ± 0.3; Extremely low 7.2 ± 0.5 - Low and Extremely low categories correspond respectively to our own Medium and Low classes; Barruffo et al., 2021 [https://doi.org/10.3390/d13070311]). We have added these citations and values at the paragraph “2.1.1 Study Sites and Sample collection” and to the relevant supplementary table.
These are historical characterization data for the same vent systems, not measurements taken during our own 2024 deployment, and vent chemistry at these sites is known to fluctuate both seasonally and sub-daily. They therefore can’t substitute for site- and time-specific measurements, and we no longer imply that our nominal pH classes are equivalent to a full carbonate system characterization of the 2024 field experiment. We agree that full, deployment-specific carbonate system characterization, for both sites, especially for Vullatura, remains a priority for future work, and we now state this explicitly as a limitation in the Discussion: in the absence of a directly measured characterization of the surrounding carbonate chemistry, the interpretation of our results based on carbonate system dynamics should be considered accordingly constrained.
R) Regarding the taxonomic indeterminacy of Phorcus sp., we thank the reviewer for raising this important point. We agree that genus-level identification does not allow us to unequivocally assign all sampled individuals to a single Phorcus species.
Photographic documentation of the collected individuals is available; however, after re-examination this material, we concluded that species-level identification based exclusively on shell morphology would not be sufficiently reliable, particularly considering the morphological plasticity of the genus and the degree of shell erosion observed at the vent sites. Unfortunately, no soft tissue samples were preserved for subsequent molecular identification. We have therefore revised the manuscript to explicitly acknowledge this taxonomic uncertainty and to avoid species-specific interpretations of the observed patterns. In particular, the inter-individual variability in shell δ¹³C can no longer be attributed exclusively to individual physiological plasticity or distinct calcification strategies, as taxonomic heterogeneity may also contribute to this variability. We have revised Section 4.2 to explicitly flag this alternative explanation.
Similarly, we have revised our interpretation of the lower abundance and smaller size of individuals observed at the most acidified station. Rather than interpreting this pattern as evidence of a demographic response to acidification, we now acknowledge that differences in the occurrence and distribution of Phorcus taxa along the environmental gradient may also contribute to the observed pattern. Throughout the revised manuscript we therefore treat the field material conservatively as Phorcus sp. and restrict our conclusions to patterns observed at the genus-level assemblage represented by our samples. We do not attempt to attribute these patterns to the response of any particular Phorcus species. This taxonomic uncertainty is now explicitly acknowledged as a limitation when interpreting the biological mechanisms underlying the observed isotopic variability.
R) Regarding the pseudoreplication point: we have re-analyzed both the mussel and Phorcus laboratory experiment datasets accordingly, as detailed below.
Mussel laboratory experiment. We re-fitted the model as nested design, with Tank as a random effect (linear mixed-effects model: δ¹³C ~ Condition + (1|Tank), fitted via "lme4/lmerTest" packages), in which each tank belongs to a single condition and is therefore not crossed with it. We additionally cross-checked this against the simplest possible correction, aggregating each tank to its mean δ¹³C and analyzing the resulting n = 3 vs. n = 3 tank means as experimental unit, following the recommendation of Harrison et al. (2018) [https://doi.org/10.7717/peerj.4794] for designs with very few clusters, since mixed-model variance-component estimates can be unstable when the number of random-effect levels is this small. Both approaches confirm that the treatment effect remains robust once pseudoreplication is corrected: the mixed model yields F₁,₃.₉ = 160.3, p = 0.0002 (Satterthwaite-approximated degrees of freedom), and the tank-aggregated analysis yields F₁,₄ = 175.2, p = 0.0002 (control mean = −2.51‰, treatment mean = −4.11‰, both expressed as means of tank means), compared to the original pseudoreplicated result of F₁,₇₃ = 169.7. We have replaced the original one-way ANOVA in the Methods and Table 1 with this tank-corrected analysis.
Phorcus laboratory experiment. We have re-analyzed the Phorcus laboratory dataset accordingly, following the same tank-corrected approach applied to the mussel laboratory experiment (mixed-effects model with Tank as a random effect, δ¹³C ~ Condition + (1|Tank), cross-checked against a tank-aggregated t-test on the n = 3 vs. n = 3 tank means). Unlike the mussel dataset, where both corrected approaches converged on the same conclusion, here the two approaches disagree, and we report this transparently rather than selecting the more favourable result. The mixed model returns a non-significant effect (F₁,₁₅ = 3.01, p = 0.103) and converges to a singular fit, with the between-tank variance component estimated at zero; the tank-aggregated t-test, in contrast, returns a significant effect (t₄ = 7.42, p = 0.0018, control mean = −1.60‰, treatment mean = −2.14‰, both as means of tank means).
We investigated the source of this disagreement and traced it to the small and uneven number of individuals per tank in this experiment (2–4 individuals, compared to 20 in the mussel experiment), and in particular to one tank (Treatment replicate 1, n = 2) with very high internal dispersion between its two individuals. With only three tanks per condition, the mixed model cannot reliably distinguish genuine between-tank variability from the noise contributed by this single, sparsely sampled tank. The tank-aggregated test, by construction, does not attempt to estimate this variance component and is therefore more stable, but it is correspondingly more sensitive to the fact that tank means themselves are based on as few as two individuals. We have revised the manuscript to present the Phorcus laboratory result descriptively, reporting both corrected analyses and explicitly stating that the tank-level replication in this experiment is insufficient to draw a statistically robust conclusion about the treatment effect. We have added this as an explicit limitation in the Discussion.
R) We thank the reviewer for highlighting this point and we agree that the role of San Pietro in the study required clearer description. San Pietro has already been used as an external ambient-pH reference site in several previous studies conducted at the Castello Aragonese vent system, based on its comparable hydrodynamic conditions, temperature and salinity (Ricevuto et al., 2015a [http://dx.doi.org/10.1016/j.marenvres.2015.09.005]; Migliaccio et al., 2019 [https://doi.org/10.1016/j.scitotenv.2019.04.005]; Palombo et al., 2023 [https://doi.org/10.1016/j.marpolbul.2023.114700]; Signorini et al., 2025 [https://doi.org/10.1016/j.envres.2025.121874];). We have added these references at and revised the corresponding text to make clear that the environmental comparability of San Pietro is supported by previous studies rather than by the shell δ¹³C measurements obtained here. We nevertheless acknowledge the reviewer’s point that site-specific carbonate-system measurements, including δ¹³C_DIC, are not available for San Pietro. We have therefore revised the manuscript to avoid implying that San Pietro is independently validated as equivalent to the Castello Ambient station in terms of carbonate chemistry. Instead, we identify it explicitly as an external ambient-pH reference site for the Phorcus dataset and acknowledge the absence of contemporaneous carbonate-system and δ¹³C_DIC measurements as a limitation when interpreting the isotopic patterns.
Finally, regarding the statistical test cited in the manuscript (F₁,₁₂ = 0.45, p = 0.514, San Pietro vs. Castello Ambient shell δ¹³C): we agree that the absence of a significant difference in shell δ¹³C between San Pietro and Castello Ambient cannot independently demonstrate environmental equivalence between the two sites. We have therefore removed this interpretation from the revised manuscript and now present San Pietro strictly as an external site with prior-documented hydrodynamic/physical comparability.
San Pietro was used exclusively as an external reference site for the Phorcus field dataset and was not part of the mussel transplantation experiment, which instead included the Castello Aragonese and Vullatura vent systems. We have clarified which sites pertain to which dataset in the Methods, and revised Fig. 1 and its caption to describe San Pietro explicitly and to assign it a distinct symbol, differentiating it visually from the ambient stations of the vent gradient. The revised figure is included in the Supplementary Material.
R) Regarding the shell mineralogy: we acknowledge that it was not directly characterised in this study. Shells of M. galloprovincialis and Phorcus sp. were consistently collected from the growing shell margin and the shell aperture, respectively, following the same powder-drilling procedure described in the Methods. The resulting powders therefore represent an integrated signal from the mineral phases present in these shell regions (a mixture of calcite and aragonite for the bimineralic growing margin of M. galloprovincialis, and aragonite alone for the exclusively nacreous shell aperture of Phorcus sp.). The standardised sampling protocol should have minimised differences in mineralogical composition among samples; if the relative proportions of the two mineral phases remained constant, their different carbon-isotope fractionation factors would primarily produce a systematic offset rather than affect the observed δ¹³C trends.
-
RC2: 'Comment on egusphere-2026-4016', Anonymous Referee #2, 14 Sep 2026
General assessment
The manuscript addresses a potentially interesting topic, namely the effects of ocean acidification on Mytilus galloprovincialis and Phorcus sp., using the carbon isotopic composition of shells combining two approaches: from natural CO₂ vent systems and from laboratory experiments. However the manuscript presents substantial methodological, experimental, and statistical limitations that compromise the robustness of the discussion and conclusion. In particular, the comparison between organisms belonging to inadequately characterized taxonomic groups, different exposure approaches, different size classes, non-comparable exposure times, and different environmental conditions introduces numerous sources of variability that are not adequately controlled. These issues are further compounded by important gaps in the description of the methods, potential pseudoreplication in the laboratory experiments, and an interpretation of the results that, in several instances, goes substantially beyond what the data can demonstrate. I also consider the lack of species-level identification of the Phorcus specimens and the use of individuals from natural populations without excluding differences in population composition among sites to be particularly problematic. Indeed, at least 3-4 species belonging to the Phorcus genus are present in the Mediterraenan Sea. This issue is especially relevant considering that the authors themselves emphasize in the Introduction the importance of species-specific responses to acidification. Overall, the issues identified do not appear to be resolvable through a simple revision of the manuscript or the Discussion. Some of them would require new experimental information, improved characterization of the organisms and study sites, and a substantial revision of the statistical approach. For these reasons, I recommend rejection of the manuscript.
Main concerns
Materials and Methods – Natural acidified conditions
The comparison between Mytilus galloprovincialis and Phorcus sp. does not allow any robust biological conclusions. The authors compare a single species, M. galloprovincialis, with a group of organisms belonging to the genus Phorcus, without providing species-level taxonomic identification of the Phorcus specimens collected in the field. This approach is problematic because it prevents the assessment of species-specific responses to acidification. Considering that many Phorcus species occur in the Mediterranean, the presence of different species at different sites could introduce a substantial source of biological variability and represent an important bias in the comparison. It is possible that the difficulty of taxonomically identifying Phorcus specimens motivated this choice, whereas M. galloprovincialis is relatively easy to identify. However, this methodological difficulty cannot be considered negligible, as it fundamentally compromises the comparability of the organisms used in the study.
The Mussel Watch field experiment is also insufficiently described. The manuscript does not clearly state how many cages were used at each site, how many animals were present in each cage, or whether mortality was monitored throughout the experiment. Furthermore, considering that mussels are filter-feeding organisms, the characteristics of the cages and their degree of permeability to water should be described, particularly to assess whether individuals located in the inner parts of the cages had comparable access to water flow. Moreover, the depth at which sampling was conducted at San Pietro is also not reported, and it is unclear whether it was comparable to the depths at Vullatura and Castello Aragonese. Although the authors state that the San Pietro site has biogeochemical parameters comparable to those of the ambient areas at Castello and is unaffected by volcanic inputs, this statement is not supported by any biogeochemical characterization presented in the manuscript. The authors describe pH, DIC, and TA gradients, but the actual values measured at different sites are not reported.
Another important gap concerns the origin of the M. galloprovincialis individuals used in the transplant experiment. The authors emphasize that the two species have different ecological histories; however, the origin of the mussels used in the field experiment is not adequately described. It is essential to specify which population the mussels originated from and whether they were collected from the wild, particularly if they were obtained from a site not shown on the map of the study area. This information is important because the previous environmental history of the transplanted individuals may influence their physiological and isotopic responses during the experiment. There is also insufficient information regarding the initial size of the animals and their growth during the 60-day exposure period. Mussel size is particularly important because it is directly related to growth and therefore to the amount of carbonate deposited during the experiment. It would therefore be important to know whether animal growth was measured and to what extent the environmental conditions allowed normal growth.
The statement that the Phorcus individuals represent populations chronically exposed to acidified conditions “from the larval stage through the juvenile, sub-adult, and adult phases” is particularly problematic. This statement cannot be demonstrated based on the data presented. Given the geographical proximity of the different sites, larval dispersal or recruitment from other areas cannot be excluded. No evidence is provided to demonstrate that the individuals were born and spent their entire life cycle at the site where they were collected. Furthermore, only three individuals were collected at some sites. Such a low sample size appears insufficient to represent the natural variability of a population, particularly when the authors collected up to 15 individuals at other sites. I understand the difficulty of obtaining large sample sizes from natural populations; however, with such a low “n”, it is impossible to exclude the possibility that the sampled individuals were transient rather than representative of a resident population. This issue is further aggravated by the absence of species-level taxonomic identification. Finally, the two species were not sampled at the same sites: M. galloprovincialis was used across the different sites of the gradient, whereas Phorcus was absent from Vullatura. This further complicates the overall comparison between the two organisms.
Materials and Methods – Laboratory experiments
The comparison between the natural and laboratory experiments also presents important limitations. The animals used in the two approaches differ in size, exposure duration, environmental conditions, and probably origin (lacking information). In the case of M. galloprovincialis, the laboratory animals were juvenile specimens (mean length of approximately 1.5 cm) collected from the northern Adriatic coast near Chioggia, whereas the organisms used in the field experiment were probably adults (again no info available). Size therefore represents another potentially important variable that could explain differences between the two experiments. Moreover, the two environmental seawaters used can be highly different, one from Nort Italy from a lagoon and the other from South Italy near an island, performed also in different years. The manuscript does not report that the water parameters used in the laboratory were comparable to those present in the natural environment at Ischia. Differences in water chemistry, salinity, temperature, nutrient availability, and other environmental parameters could contribute to the differences observed. The exposure durations also differ: approximately 60 days in the field experiment and 90 days in the laboratory experiment for M. galloprovincialis, whereas the laboratory experiment with Phorcus lasted eight weeks. These differences make impossible a rigorous comparison between the different experiments. The laboratory experiments also lack sufficient information on the actual amount of food available, and, in the case of mussels, the concentration of microalgae supplied ( cells/mL). Similarly, the actual pH values measured in the different treatments are not adequately reported. It is therefore unclear whether the nominal values set by the system actually corresponded to the conditions to which the animals were exposed throughout the experiment. For Phorcus, the presence of several species within the genus makes taxonomic identification of individuals from the different populations particularly important. Comparing populations from southern and northern Italy without verifying their species identity may introduce additional variability into the dataset and makes it difficult to attribute the observed differences specifically to acidification.
Isotopic analyses
The description of the shell sampling procedures is insufficient to ensure study reproducibility. It is unclear how many individuals were selected from each cage, how they were selected, and which specific portion of the shell was analyzed. Although the authors generally identify the growing shell margin for M. galloprovincialis and the aperture for Phorcus, they do not provide sufficiently detailed information on the size and precise location of the analyzed shell portion. For example, it is unclear how the newly deposited shell portion was identified, whether it represented a defined percentage of the total shell length, and whether an initial shell size measurement was available before the experiment. Did the authors use image analysis to quantify the newly deposited shell portion? Similar information is missing for Phorcus. These details are particularly important for the laboratory experiments, where the objective is to interpret the isotopic signal as a consequence of the experimental exposure. For the field experiment, it is also difficult to understand how the shell portion corresponding to growth during the 60-day exposure period was identified in a standardized manner. The statement “All samples consisted of multiple shell specimens” is particularly problematic. It is unclear whether “sample” refers to a single isotopic sample obtained by pooling several shells or to multiple shells analyzed separately. This distinction is fundamental because the two approaches result in completely different experimental units and sample sizes. If several shells were pooled into a single sample, inter-individual variability cannot be assessed. If the shells were analyzed individually, the number of individuals analyzed and how these measurements were subsequently used in the statistical analysis must be clearly reported. As currently written, this statement makes it impossible to determine the actual n used in the analyses and the true experimental units. This is an important issue both for study reproducibility and for the correct interpretation of the statistical results.
Statistical analysis
The main statistical concern is pseudoreplication. In the laboratory experiments, multiple individuals were maintained within the same tank, yet the statistical analyses appear to treat individual animals as independent observations. However, the experimental unit is the tank rather than the individual animal. Individuals maintained within the same tank share the same experimental conditions and therefore cannot be considered fully independent replicates of the treatment. In the mussel experiment, 20 individuals were maintained in each tank, with three replicate tanks per treatment. Therefore, the effective experimental sample size is n = 3 tanks per treatment, rather than the total number of individuals analyzed. The reported ANOVA degrees of freedom (df = 1,73), however, suggest that individual mussels were treated as independent observations. This may substantially inflate the degrees of freedom and result in artificially high statistical significance. The same concern applies to the Phorcus laboratory experiment. The analysis should therefore account for the hierarchical structure of the experimental design, either by considering the tank as the experimental unit or by using an appropriate mixed-effects model with tank included as a random effect. Treating individuals within the same tank as independent replicates is not appropriate for this experimental design.
Discussion – Natural acidified conditions
Section 4.1 presents a potentially interesting result, namely the enrichment in δ13C observed in Mytilus galloprovincialis shells at low-pH sites. However, I believe that the interpretation proposed by the authors goes substantially beyond what the data can demonstrate. In particular, the conclusion that biological modulation is capable of “overriding” the ambient geochemical signal is not adequately supported by the dataset. The main issue is the absence of δ13C-DIC measurements along the environmental gradient and of measurements of the main organic carbon sources. Without these data, it is not possible to distinguish the contribution of ambient DIC from that of metabolic carbon or to quantify the relative importance of the different carbon sources. Consequently, the statement that respiratory CO₂ represents the dominant carbon source for calcification remains purely speculative. Similarly, the hypothesis that acidification induces prolonged valve closure, isolation of the extrapallial fluid, and consequently an increased metabolic contribution has not been experimentally tested. No behavioral, physiological, respiratory, or calcification measurements are available to verify this mechanism. The section concerning primary productivity is also particularly problematic. The statement that photosynthesis produces biomass enriched in 13C that is subsequently assimilated by the organisms is not adequately supported. Again, measurements of the δ13C of organic matter and its sources are lacking, making it impossible to substantiate the proposed mechanism. Similar caution is required for the ontogenetic interpretation of Phorcus sp., where the δ13C pattern is attributed to a progressive reduction in vital effects without independent measurements of growth or calcification rates. Several alternative explanations remain possible, including changes in diet, metabolism, microhabitat, and seasonality. Finally, I do not consider the statement that seasonal variability in δ13C-DIC cannot explain the observed pattern to be sufficiently justified, since this variable was not directly measured at the study sites. Comparison with values reported in the literature is not sufficient to exclude this possibility.
Overall, an isotopic observation is used to construct a highly specific physiological and biogeochemical explanation without measuring the variables required to test the proposed mechanisms.
Discussion – Laboratory experiments
The interpretation of the laboratory experiment is also more speculative than the data allow. The authors observe a decrease in δ13C in Mytilus galloprovincialis under reduced pH and interpret this result as a direct geochemical response to the reduction in CO₃²⁻ availability. However, the experiment does not allow the authors to identify CO₃²⁻ reduction as the specific cause of the observed isotopic change. Lowering pH by adding CO₂ affects several components of the carbonate system simultaneously, and the study does not provide evidence that the observed δ13C shift can be specifically attributed to changes in CO₃²⁻. The authors further conclude that, under controlled laboratory conditions, the shell acts as a relatively passive recorder of environmental chemistry. However, controlled conditions do not eliminate physiological effects. The mussels were still actively feeding, growing, respiring, and calcifying, and no measurements of growth, calcification, respiration, or metabolic activity were provided. Therefore, the study cannot determine whether the observed isotopic change was driven primarily by environmental carbonate chemistry, physiological responses, or a combination of both. The data support an association between reduced pH and changes in shell δ13C, but they do not demonstrate the specific mechanism proposed by the authors. The interpretation of Phorcus sp. is even weaker. A non-significant trend is used to support the existence of high physiological plasticity, different calcification strategies, changes in energy allocation, and stochastic calcification events. None of these hypotheses is directly supported by measurements of growth, metabolism, respiration, or calcification. Inter-individual variability alone does not demonstrate physiological plasticity or different calcification strategies. The laboratory result for Phorcus is in fact not statistically significant (p = 0.144), and therefore cannot be used to support an isotopic response to acidification. Likewise, the significant treatment effect observed in natural Phorcus populations cannot simply be interpreted as an effect of acidification because the individuals originate from different natural populations, the species were not taxonomically resolved, sample sizes can be variable, and several environmental factors may be confounded with pH. The strong ontogenetic effect observed in the Phorcus data (F₂,₃₉ = 29.67; p < 0.001) further demonstrates that δ¹³C is strongly influenced by growth stage. However, the criteria used to assign individuals to the different ontogenetic stages (juvenile, sub-adult, and adult) are not described. It is unclear whether these categories were based on shell size, morphological characteristics, or another criterion. This information is particularly important because shell size and growth rate may directly influence the isotopic signal. Moreover, without independent measurements of growth or calcification rates, the subsequent interpretation in terms of changes in vital effects remains speculative.
Overall, also in this section the authors transform descriptive isotopic patterns into untested mechanistic explanations. The experiments demonstrate, at most, an association between acidification and changes in shell carbonate δ13C; they do not demonstrate either predominantly geochemical control of the signal in Mytilus or metabolic control in Phorcus.
Conclusions
The methodological and statistical issues described above, considered together, substantially compromise the robustness of the conclusions of the manuscript. The problems are not limited to the overinterpretation of some results but also concern the difficulty of reliably identifying the experimental units, assessing their independence, determining the taxonomic composition of the Phorcus populations, verifying the actual exposure conditions, and identifying the sources of carbon incorporated into the shells. These issues cannot be resolved through a simple revision of the text or a reformulation of the Discussion. Additional experimental information and, for some of the main issues, new data would be required to adequately address these concerns. Furthermore, although the authors compare natural and laboratory approaches to study ocean acidification, the two experiments differ in several additional factors, including species, animal size, origin, and exposure duration. Therefore, differences in isotopic responses cannot be attributed specifically to the experimental approach (natural versus laboratory), as they may instead result from these uncontrolled differences. Overall, I consider these limitations to be fundamental to the experimental design and interpretation of the results. Therefore, I recommend rejection of the manuscript in its present form.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 224 | 86 | 36 | 346 | 34 | 28 |
- HTML: 224
- PDF: 86
- XML: 36
- Total: 346
- BibTeX: 34
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
This manuscript addresses a relevant and interesting question: whether shell δ¹³C in marine calcifiers tracks seawater carbonate chemistry under acidification, and whether laboratory and natural CO₂ vent systems yield consistent isotopic responses. The paired vents/mesocosm comparison on two taxa with contrasting life histories, complemented by an ontogenetic component in Phorcus, is in principle a valuable approach, and the underlying dataset has genuine documentary value for the Ischia vent system. I therefore consider the scientific significance of the topic to be good.
However, I regret to conclude that the manuscript in its current form suffers from major deficiencies in scientific quality that cannot be remedied through revision. My concerns fall into two categories: 1: structural problems in the experimental design and in the comparability of the field and laboratory arms, and 2: serious issues of internal consistency and traceability of the reported results.
Regarding (1), the central claim of the paper, that biological/trophic processes override the geochemical signal in complex natural environments, requires that the field and laboratory arms differ only in ecological complexity. Instead, the two approaches differ simultaneously in: source populations (Adriatic vs. Tyrrhenian; the origin of the transplanted mussels is never stated), life stage (juvenile mussels in the lab vs. adults in the field; lifetime-exposed native Phorcus vs. 8-week exposed lab adults), exposure duration and history, diet (cultured microalgae of unmeasured δ¹³C for lab mussels vs. natural seston; field-collected epiphytes for lab gastropods), thermal regime (constant 28.5 °C vs. autumn cooling in situ), and pH regime (stable setpoint vs. the highly fluctuating exposure characteristic of vent systems). In addition the pH logger data are mentioned but never shown. The observed lab/field divergence is thus overdetermined: it is compatible with at least half a dozen explanations besides the trophic/metabolic override favoured by the authors. Critically, the interpretation invokes precisely the variables that were not measured in either system (δ¹³C of DIC, POM, and diet), as the authors themselves acknowledge (L308–311). A descriptive juxtaposition of the two datasets would be defensible; the mechanistic conclusions drawn in the Abstract and Discussion are not supported by this design.
In addition, three structural problems are not easily fixable in revision:
1) Absence of carbonate system characterization. No measurements of TA, DIC, pCO₂, or Ω are reported for either the laboratory treatments or the field sites; only nominal pH values are given, and the deployment logger data are not presented. For an ocean acidification study whose entire interpretive framework rests on DIC pool dynamics, this falls short of established best practices, and the field chemistry of the 2024 deployment cannot be reconstructed retroactively.
2) Taxonomic indeterminacy of Phorcus sp. The Mediterranean hosts four Phorcus, at least three of which (P. turbinatus, P. richardi, P. mutabilis) co-occur in the Tyrrhenian and segregate primarily by shore height. The sampling protocol ("from the lower intertidal to the upper subtidal") targets exactly the zone of maximum species overlap. Consequently: (a) the pronounced inter-individual variability interpreted as physiological plasticity and "distinct calcification strategies" (Section 4.2) is indistinguishable from interspecific variance; (b) since species composition may co-vary with the pH gradient (microhabitat and depth change along it), the treatment effect is confounded with a possible taxon effect; (c) even the reported scarcity and reduced size at the Low pH station (L137–139) could reflect species turnover rather than a demographic response. Shell morphology in this genus is notoriously plastic and vent-induced corrosion further degrades its diagnostic value. Unless voucher tissues or photographic documentation of live animals are available for identification or DNA barcoding, this confound cannot be resolved. I ask the authors to state explicitly whether such material exists.
3) Pseudoreplication. In the mussel laboratory experiment, three treatment and three control tanks held 20 individuals each, yet individuals are treated as independent replicates (df ≈ 70) with no tank effect in the models. For the Phorcus lab experiment, the number of tanks is never stated, and the Welch test df (7.35) imply a very small effective sample. A mixed-model reanalysis with tank as a random effect would be the minimum requirement, but the design itself (effectively n = 3 tanks per treatment, and possibly fewer for gastropods) remains weak.
Regarding (2), traceability: the manuscript reports two irreconcilable statistics for the same analysis (one-way ANOVA on lab mussels: F₁,₆₉ = 34.79 in the text at L227 vs. F = 169.7, df = 1, 73 in Table 1), inconsistent pH values between Methods and Discussion (8.2/7.7 vs. ~8.0/~7.5), and three mutually incompatible statements of experimental duration (August–October ≈ 60 days; "3 months"; "90 days"). Furthermore, the Abstract states that "both species exhibited a significant depletion" in the laboratory, whereas the Phorcus result is explicitly non-significant (p = 0.144, L265–268), and the "paradoxical enrichment at low-pH sites (pH ~7.4)" was statistically significant only at Vullatura for mussels, while Phorcus was never sampled at Low pH at all. The Abstract thus presents a substantially cleaner pattern than the data support. Individually, some of these could be typographical; collectively, they undermine confidence in the analysis pipeline. I encourage the editor and authors to make the raw data and R scripts available for verification, as offered in the Code and Data Availability statement.
In addition, San Pietro reference site (L107–109): The site is stated to "share comparable biogeochemical parameters with the Castello's ambient zones", but no comparative data are shown or cited to support this claim. No TA, DIC, pCO₂, Ω, temperature, or salinity measurements for San Pietro are reported anywhere in the manuscript, nor, critically for this study, any δ¹³C DIC characterization. The validity of San Pietro as an external control for the Phorcus dataset rests entirely on this undocumented assertion. Note also that the non-significant difference in shell δ¹³C between San Pietro and Castello Ambient (F₁,₁₂ = 0.45, p = 0.514; L245–246) cannot by itself validate the site as a control, since with n as low as 3 per site the test is severely underpowered, and shell δ¹³C is the response variable, not an independent verification of water chemistry. Please provide the environmental data supporting the comparability claim, or cite the specific source from which it derives. Relatedly, in Fig. 1 the westernmost marker (San Pietro) is shown in the map and legend but is never described in the caption, which mentions only Castello Aragonese and Vullatura; please describe San Pietro in the caption, clarify which sites pertain to which dataset (mussel transplants: Castello and Vullatura; Phorcus collections: Castello and San Pietro), and consider a distinct symbol/color for San Pietro, as it is an external reference site rather than an ambient station of the vent gradient.
Finally, The Castello Aragonese vent system falls within the 'Regno di Nettuno' Marine Protected Area, where the collection of marine fauna requires authorization from the managing body. The manuscript does not mention any collection or sampling permits for the field activities (mussel cage deployment at Castello Aragonese and Vullatura; Phorcus collection at Castello Aragonese and San Pietro). Please state the permits obtained and the issuing authority.
Specific comments
Technical corrections
Concluding remark. Beyond the individual issues listed above, I believe this manuscript effectively contains two studies that are not mutually comparable: a field study at the Ischia vent systems (mussel transplants and native Phorcus populations, including the ontogenetic component) and a laboratory mesocosm study. As detailed above, the two arms differ simultaneously in source populations, life stage, exposure duration and history, diet, thermal regime, pH stability and range, and habitat, that is, in essentially every variable that is isotopically relevant. The lab–field 'divergence' that frames the title and Abstract is therefore not an interpretable result but an artefact of comparing systems that were never designed to be compared. I would encourage the authors to consider developing the two datasets as separate manuscripts: a descriptive field study contributing to the well-established Ischia vent literature (contingent on resolving the taxonomic identity of Phorcus and providing site-level carbonate chemistry and δ¹³C DIC data), and a laboratory study of isotopic responses to controlled acidification (contingent on a tank-level reanalysis and characterization of dietary and seawater δ¹³C). Each dataset has merit within its own scope; neither is served by the current framing, which asks the juxtaposition to support mechanistic conclusions that the design cannot deliver.