the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Jul 2023
A step towards measuring connectivity in the deep-sea: elemental fingerprints of mollusk larval shells discriminate hydrothermal vent sites
Christophe Pecheyran
Fanny Claverie
Cécile Cathalot
Marjolaine Matabos
Yoan Germain
Olivier Rouxel
Didier Jollivet
Thomas Broquet
Thierry Comtet
Abstract. Deep-sea hydrothermal-vent systems are under investigation for base and precious metal exploitations. The impact of mining will depend critically on the ability of larval dispersal to connect and replenish endemic populations. However, assessing connectivity is extremely challenging, especially in the deep sea. Here, we investigate the potential of elemental fingerprinting of mollusc larval shells to discriminate larval origins between multiple hydrothermal sites in the Southwest Pacific Ocean. The gastropod Shinkailepas tollmanni represents a suitable candidate as it uses capsules to hold larvae before dispersal, which facilitates sampling. Multielemental microchemistry was performed using cutting-edge femtosecond laser ablation Inductively Coupled Plasma Mass Spectrometry analysis to obtain individual measurements on 600 encapsulated larval shells. We used classification methods to discriminate the origin of individuals from 14 hydrothermal sites spanning over 3,500 km, with an overall success rate of 70 %. When considering less sites within more restricted areas, reflecting dispersal distances reported by genetic and modelling approaches, the success rate increased up to 86 %. We conclude that individual larval shells register site-specific elemental signatures that can be used to assess their origin. These results open new perspectives to get direct estimates on population connectivity from the geochemistry of pre-dispersal shell of recently settled juveniles.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(0 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Journal article(s) based on this preprint
Vincent Mouchi et al.
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1324', Lisa Levin, 12 Aug 2023
Review of Mouchi et al. A step towards measuring connectivity in the deep sea.General comments:
This work represents an important first step in the application of trace-elemental fingerprinting to deep-sea hydrothermal vent connectivity. The authors have selected a suitable test species and studied an impressive array of sites (14). I think this effort is a precursor to many important applications.
Although the authors present deep seabed mining as the key motivator , I would argue that there are many additional reasons for wanting to know about connectivity of vents – from basic science, to biodiversity conservation (30x30 goals), to addressing consequences of climate change on connectivity (e.g., Mitarai’s work in Levin et al. 2020 DOI: 10.1111/gcb.15223), vent roles in the carbon cycle etc. I would encourage at least mention of these other motivators. Within the mining realm, in addition to understanding consequences of mining disturbance, the authors should point out the importance of connectivity data to the designation of no-mining protected areas (APEIs) and also reference zone PRZ and IRZ (preservation and impact) designation.
Please consider adding to the end of the introduction a paragraph that clearly lays out the goals or objectives of the research. This could be in the form of questions, hypotheses or other… but should frame the science around the data presented in the paper. E.g., does the water chemistry of ‘habitat water’ differ among vent sites where S. tollmanni egg capsules occur? What elements are key to distinguishing sites? Does the trace element signatures of S. tollmanni larval shell reflect the habitat water chemistry? Are there specific sites or scales of connectivity where the application of trace elemental fingerprinting to the vent systems is likely to most reliable? What elements are key to this distinction? Some of these questions show up as the headings in the results section – but should be presented earlier.
Tell us a little bit more about the study species Shinkailepas tollmanni – its distribution, and its host Iffremeria distribution. Is the relationship obligate? Is anything known about depth ranges, longevity, development time, feeding mode, planktonic duration? How does its life history affect inferences about connectivity? Will information about mollusc/gastropod connectivity be relevant to other vent taxa – what will or won’t?
Consider discussion which of the study sites are targeted for mining, and which might serve as source populations. IUCN has red listed some species like the scaly foot snail based on their limited occurrence primarily in areas targeted for mining. I realize you don’t have any source or sink data generated yet but it might be useful to explain how this precursor work can lead to analyses that inform identification of vulnerable species.
Specific comments.
Title – please only hyphenate deep sea when it is used as a double adjective. Here it is a noun and should not be hyphenated.
Abstract: Line 19. The presence of capsules not only facilitates sampling, it means the embryos develop in a fixed location – and form carbonates whose trace elemental signature could reflect that location.
Line 30 – Even vent sites on mid ocean ridges far from coastlines are of interest to commercial mining (ISA contracts on Mid Atlantic Ridge and Southwest Indian Ridge).
Line 65 – replace ‘minute’ with a more accurate indication of size…. 100 micron?
Line 72 – replace ‘which carbonate shells ‘with ‘whose carbonate shells’
Line 263 Simmonds et al. should come out of the parentheses.
Note that Levin 2006 (doi:10.1093/icb/icj024) in discussing future directions for larval dispersal studies in a larval dispersal review wrote How much larval exchange occurs within and among reducing ecosystems such as vents, seeps, and whale falls? Analysis of short-term larval exchange among seep or hydrothermal vent ecosystemsmight be tractable if these impart distinct trace element signatures to larval shells.
I think this paper is really the first to tackle this problem. (17 years later).
Fig. 4 and 5 caption. Indicate what the lighter and darker shades of blue and red mean.
The supplemental information needs a table of contents as it is extensive.
Table S4 should replace the ‘,’ with decimal points.
The comparison of habitat water to larval shell chemistry should be included in the main paper.
Overall this is a significant contribution in need of relatively minor revision.
Citation: https://doi.org/10.5194/egusphere-2023-1324-RC1 -
AC2: 'Reply on RC1', Vincent Mouchi, 11 Nov 2023
Final reply to the comments from RC1:
Review of Mouchi et al. A step towards measuring connectivity in the deep sea.
Reviewer’s comments are in italics. Author responses are in bold.
General comments:
This work represents an important first step in the application of trace-elemental fingerprinting to deep-sea hydrothermal vent connectivity. The authors have selected a suitable test species and studied an impressive array of sites (14). I think this effort is a precursor to many important applications.We thank you for your interest in our study.
Although the authors present deep seabed mining as the key motivator, I would argue that there are many additional reasons for wanting to know about connectivity of vents – from basic science, to biodiversity conservation (30x30 goals), to addressing consequences of climate change on connectivity (e.g., Mitarai’s work in Levin et al. 2020 DOI: 10.1111/gcb.15223), vent roles in the carbon cycle etc. I would encourage at least mention of these other motivators. Within the mining realm, in addition to understanding consequences of mining disturbance, the authors should point out the importance of connectivity data to the designation of no-mining protected areas (APEIs) and also reference zone PRZ and IRZ (preservation and impact) designation.
We agree that the listed motivators (in addition to deep-sea mining) are also important. We added some references to include these in the introduction: “As a basic process of population dynamics, connectivity is key to understand the persistence of vent communities as well as the potential of vent species to colonize new areas and habitats (e.g., Adams et al. 2012; Levin et al. 2016). As such, it is of major interest for biodiversity conservation and the design of deep-sea marine protected areas (Danovaro et al. 2020; Combes et al., 2021), for example in the framework of the 30x30 initiative, which aims to protect at least 30% of the world’s oceans by 2030 (O’Leary et al. 2019). In the particular context of deep-sea mining, connectivity has been identified as a key scientific knowledge gap (Gollner et al., 2017; Miller et al., 2018; Smith et al., 2020; Amon et al., 2022) that should be tackled taking into account its interactions with climate change (Levin et al., 2020)”.
Also, we mention protected areas when we give details on how we can use elemental fingerprinting (see last general comment below).Please consider adding to the end of the introduction a paragraph that clearly lays out the goals or objectives of the research. This could be in the form of questions, hypotheses or other… but should frame the science around the data presented in the paper. E.g., does the water chemistry of ‘habitat water’ differ among vent sites where S. tollmanni egg capsules occur? What elements are key to distinguishing sites? Does the trace element signatures of S. tollmanni larval shell reflect the habitat water chemistry? Are there specific sites or scales of connectivity where the application of trace elemental fingerprinting to the vent systems is likely to most reliable? What elements are key to this distinction? Some of these questions show up as the headings in the results section – but should be presented earlier.
We added a paragraph at the end of the introduction to list the main questions the manuscript aims at answering: “To explore the use of elemental fingerprints of larval shells in this species, several questions were investigated. Are there enough chemical contrasts between hydrothermal vent sites for elemental fingerprinting? Do elemental fingerprints of larval shells reflect their natal place? What are the main elements responsible for the elemental fingerprint of larval shells? At which spatial scale is elemental fingerprint more accurate? Does the elemental fingerprint of larval shells correspond to that of habitat water, in the sense that the habitat water fingerprint could be used as a reference for the shell fingerprint?”.
Tell us a little bit more about the study species Shinkailepas tollmanni – its distribution, and its host Iffremeria distribution. Is the relationship obligate? Is anything known about depth ranges, longevity, development time, feeding mode, planktonic duration? How does its life history affect inferences about connectivity? Will information about mollusc/gastropod connectivity be relevant to other vent taxa – what will or won’t?
We added available information on the biology of the studied species as requested by RC1 in a new first sub-section of the Materials and Methods. Most questions asked by RC1 about its life history traits were addressed, except the question of its longevity, for which no report exists, to our knowledge. At the end of this new sub-section, we briefly expose how larval history traits were used to interpret observations on genetic connectivity: “Based on morphological characteristics (e.g. size of encapsulated veligers vs size of postlarvae), larvae of S. tollmanni are considered planktotrophic, and are assumed to stay pelagic for up to one year (Yahagi et al., 2020). Although no direct evidence exists, larvae of S. tollmanni are assumed to disperse in surface waters as shown for other Phenacolepadidae (Yahagi et al., 2020). Such dispersal abilities could explain the genetic homogeneity observed among populations of S. tollmanni at the scale of the study area (Yahagi et al., 2020; Poitrimol et al., 2022)”.
Unfortunately, for now, we cannot reliably give information on how connectivity interpretation inferred from elemental fingerprinting assignment (we don’t have the necessary data yet; see reply below) would be ‘relevant to other vent taxa’. In particular, we don’t have the data to discuss the possibility of using elemental fingerprint references from S. tollmanni to assign the origin of individuals from other taxa occurring on the same vent sites.Consider discussion which of the study sites are targeted for mining, and which might serve as source populations. IUCN has red listed some species like the scaly foot snail based on their limited occurrence primarily in areas targeted for mining. I realize you don’t have any source or sink data generated yet but it might be useful to explain how this precursor work can lead to analyses that inform identification of vulnerable species.
The identification of source and/or sink populations is currently impossible, as we have no assignment of juvenile individuals from our method. We however added in the introduction specifics to how the assignment of juveniles from elemental fingerprints of their larval shells can indeed identify these populations, and help determine which of these sites should be particularly critical for the survival of the studied species: “Assigning the origin of juveniles from elemental fingerprints will provide rare evidence of single-generation connectivity ranges and identify which sites correspond to sources and sinks for populations over a wide area. This type of information can subsequently be used to identify sites of higher priority for preservation and protection from disturbance to ensure the resilience of populations (e.g., no-mining areas known as Areas of Particular Environmental Interest, or APEIs, International Seabed Authority, 2011; Dunn et al. 2018; Preservation Reference Zones and Impact Reference Zones; International Seabed Authority, 2019). Moreover, identifying source and sink population and migration rates for many species would therefore help identify the species vulnerable to disturbance or even destruction of a specific habitat.”.
Specific comments:
All the specific comments by RC1 have been answered in the corrected manuscript. In particular, as suggested, we transferred the details on the correlations between habitat water and larval shells from Supplementary Information to the main text, in section 3.3.
Adams, D.K., Arellano, S.M., and Govenar, B.: Larval dispersal: vent life in the water column, Oceanography, 25, 256–268, http://dx.doi.org/10.5670/oceanog.2012.24, 2012.
Amon, D. J., Gollner, S., Morato, T., Smith, C. R., Chen, C., Christiansen, S., Currie, B., Drazen, J. C., Fukushima, T., Gianni, M., Gjerde, K. M., Gooday, A. J., Grillo, G. G., Haeckel, M., Joyini, T., Ju, S.-J., Levin, L. A., Metaxas, A., Mianowicz, K., Molodtsova, T. N., Narberhaus, I., Orcutt, B. N., Swaddling, A., Tuhumwire, J., Palacio, P. U., Walker, M., Weaver, P., Xu, X.-W., Mulalap, C. Y., Edwards, P. E. T., and Pickens, C.: Assessment of scientific gaps related to the effective environmental management of deep-seabed mining, Marine Policy, 138, 105006, https://doi.org/10.1016/j.marpol.2022.105006, 2022.
Combes, M., Vaz, S., Grehan, A., Morato, T., Arnaud-Haond, S., Dominguez-Carrió, C., Fox, A., González-Irusta, J. M., Johnson, D., Callery, O., Davies, A., Fauconnet, L., Kenchington, E., Orejas, C., Roberts, J. M., Taranto, G., and Menot, L.: Systematic Conservation Planning at an Ocean Basin Scale: Identifying a Viable Network of Deep-Sea Protected Areas in the North Atlantic and the Mediterranean, Frontiers in Marine Science, 8, 2021.
Danovaro, R., Fanelli, E., Aguzzi, J., Billett, D., Carugati, L., Corinaldesi, C., Dell’Anno, A., Gjerde, K., Jamieson, A. J., Kark, S., McClain, C., Levin, L., Levin, N., Ramirez-Llodra, E., Ruhl, H., Smith, C. R., Snelgrove, P. V. R., Thomsen, L., Van Dover, C. L., and Yasuhara, M.: Ecological variables for developing a global deep-ocean monitoring and conservation strategy, Nat Ecol Evol, 4, 181–192, https://doi.org/10.1038/s41559-019-1091-z, 2020.
Dunn, D. C., Van Dover, C. L., Etter, R. J., Smith, C. R., Levin, L. A., Morato, T., Colaço, A., Dale, A. C., Gebruk, A. V., Gjerde, K. M., Halpin, P. N., Howell, K. L., Johnson, D., Perez, J. A. A., Ribeiro, M. C., Stuckas, H., Weaver, P., and SEMPIA Workshop Participants: A strategy for the conservation of biodiversity on mid-ocean ridges from deep-sea mining, Sci. Adv., 4, eaar4313, https://doi.org/10.1126/sciadv.aar4313, 2018.
Gollner, S., Kaiser, S., Menzel, L., Jones, D. O. B., Brown, A., Mestre, N. C., van Oevelen, D., Menot, L., Colaço, A., Canals, M., Cuvelier, D., Durden, J. M., Gebruk, A., Egho, G. A., Haeckel, M., Marcon, Y., Mevenkamp, L., Morato, T., Pham, C. K., Purser, A., Sanchez-Vidal, A., Vanreusel, A., Vink, A., and Martinez Arbizu, P.: Resilience of benthic deep-sea fauna to mining activities, Marine Environmental Research, 129, 76–101, https://doi.org/10.1016/j.marenvres.2017.04.010, 2017.
International Seabed Authority: Environmental management plan for the Clarion-Clipperton Zone. ISBA/17/LTC/7, https://www.isa.org.jm/documents/isba-17-ltc-7/, 2011.
International Seabed Authority: Report of ISA workshop on the design of “Impact Reference Zones” and “Preservation Reference Zones” in deep-sea mining contract areas, ISA Technical Study No 21, https://www.isa.org.jm/publications/technical-study-21-report-of-isa-workshop-on-the-design-of-impact-reference-zones-and-preservation-reference-zones-in-deep-sea-mining-contract-areas/, 2019.
Levin, L. A., Baco, A. R., Bowden, D. A., Colaco, A., Cordes, E. E., Cunha, M. R., Demopoulos, A. W. J., Gobin, J., Grupe, B. M., Le, J., Metaxas, A., Netburn, A. N., Rouse G. W., Thurber, A. R., Tunnicliffe, V., Van Dover, C. L., Vanreusel, A., and Watling, L.: Hydrothermal vents and methane seeps: rethinking the sphere of influence, Frontiers in Marine Science, 3, 72, https://doi.org/10.3389/fmars.2016.00072, 2016.
Levin, L. A., Wei, C.-L., Dunn, D. C., Amon, D. J., Ashford, O. S., Cheung, W. W. L., Colaço, A., Dominguez-Carrió, C., Escobar, E. G., Harden-Davies, H. R., Drazen, J. C., Ismail, K., Jones, D. O. B., Johnson, D. E., Le, J. T., Lejzerowicz, F., Mitarai, S., Morato, T., Mulsow, S., Snelgrove, P. V. R., Sweetman, A. K., and Yasuhara, M.: Climate change considerations are fundamental to management of deep-sea resource extraction, Global Change Biology, 26, 4664–4678, https://doi.org/10.1111/gcb.15223, 2020.
Miller, K. A., Thompson, K. F., Johnston, P., and Santillo, D.: An Overview of Seabed Mining Including the Current State of Development, Environmental Impacts, and Knowledge Gaps, Front. Mar. Sci., 4, 418, https://doi.org/10.3389/fmars.2017.00418, 2018.
O’Leary, B. C., Allen, H. L., Yates, K. L., Page, R. W., Tudhope, A. W., McClean, C., Rogers, A. D., Hawkins, J. P., and Roberts, C. M.: 30x30: A blueprint for ocean protection, Umweltstiftung Greenpeace, www.greenpeace.org/30x30, 2019.
Poitrimol, C., Thiébaut, É., Daguin-Thiébaut, C., Le Port, A.-S., Ballenghien, M., Tran Lu Y, A., Jollivet, D., Hourdez, S., and Matabos, M.: Contrasted phylogeographic patterns of hydrothermal vent gastropods along South West Pacific: Woodlark Basin, a possible contact zone and/or stepping-stone, PLoS ONE, 17, e0275638, https://doi.org/10.1371/journal.pone.0275638, 2022.
Smith, C. R., Tunnicliffe, V., Colaço, A., Drazen, J. C., Gollner, S., Levin, L. A., Mestre, N. C., Metaxas, A., Molodtsova, T. N., Morato, T., Sweetman, A. K., Washburn, T., and Amon, D. J.: Deep-Sea Misconceptions Cause Underestimation of Seabed-Mining Impacts, Trends in Ecology & Evolution, 35, 853–857, https://doi.org/10.1016/j.tree.2020.07.002, 2020.
Yahagi, T., Thaler, A. D., Van Dover, C. L., and Kano, Y.: Population connectivity of the hydrothermal-vent limpet Shinkailepas tollmanni in the Southwest Pacific (Gastropoda: Neritimorpha: Phenacolepadidae), PLoS ONE, 15, e0239784, https://doi.org/10.1371/journal.pone.0239784, 2020.Citation: https://doi.org/10.5194/egusphere-2023-1324-AC2
-
AC2: 'Reply on RC1', Vincent Mouchi, 11 Nov 2023
-
RC2: 'Comment on egusphere-2023-1324', Steffen Kiel, 28 Sep 2023
This is an interesting study exploring the use of the elemental fingerprint of gastropod larval shells to predict their origin at certain hydrothermal vent sites. Overall, the study is well executed and described, and is worth publication.
There is, however, one issue that the authors should check. In the last paragraph before the conclusions (lines 292-293), they write ‘our data clearly evidenced that chemical composition of habitat water is not a reliable predictor of shell chemistry, and thus cannot be used to assign the origin of individuals’. This raises the question of the overall role of water chemistry in larval shell chemistry, which in turn puts the whole approach into question: if larval shell chemistry is largely biologically controlled, it would have little geographic meaning.
What the authors could do is the following test: do the same set of tests for habitat water chemistry as they have done for larval shell chemistry, and check if the geographic structure found through both sets of tests match. It is entirely possible (and in my view likely) that water chemistry does indeed determine larval shell chemistry. However, there could be various, possibly biologically controlled, inter-dependencies among the elements that make it difficult to find a straightforward, one-to-one match between water and shell chemistry. By comparing the geographic structure of habitat water chemistry vs. larval shell chemistry, that problem could be avoided.
Citation: https://doi.org/10.5194/egusphere-2023-1324-RC2 -
AC1: 'Reply on RC2', Vincent Mouchi, 15 Oct 2023
We thank you for your assessment of our work and your interest.
Before going any further in modifying the manuscript, we would like to clear up any ambiguity, as we believe there could be a misunderstanding of our statement.
We perfectly agree that water chemistry is responsible, at some degree, of the shell elemental composition, as it is the core rationale behind the elemental fingerprint approach. We showed a weak but significant correlation between water and shell composition for some elements, which supports the role of habitat water on shell composition (Supplementary Information S3 of our initial manuscript, but will be incorporated in the main manuscript as requested by RC1). Because of this weak correlation, we then state that other factors (both biological and mineralogical) induce variations of this elemental composition and prevent using directly the exact composition of the water as a reference for shell elemental fingerprinting, requiring to determine the elemental fingerprint from the larval shells themselves, to be used as a reference.
Unfortunately, the classification of habitat water composition you request would be unreliable. Indeed, we won’t be able to compare ‘patterns’ of different datasets from the same method, due to the algorithms of the classification methods we use here. Contrary to methods such as Principal Component Analysis (whose results are unconclusive here; see Fig. 3 of the manuscript), most classification methods use artificial intelligence to cluster data, and representations of these clusters are not possible (explanations of some of these methods were given in Supplementary Information S4), much more so using multi-elemental datasets. Finally, the sample size of water measurements is substantially smaller than that of the larval shells (n=1 to 7 for sites where larvae were collected, mean sample size = 2.6), which prevents the classification at the site scale, and at larger scale the quality of the classification may not be as accurate, potentially overestimating the success rate. We nevertheless attempted the classification procedures for this comment reply (Fig. 1 attached), using Mg, Mn, Fe, Cu, Zn and Sr (the elements we used to compare habitat water with larval shells) from 15 and 17 habitat water measurements (for Western sites and Woodlark, and Eastern sites and Woodlark, respectively) corresponding to the larvae collection sites. We obtained 79.2% accuracy for Manus and Woodlark (using Cubic SVM), compared to 64.5% by random assignment, with a p-value of 0.075, which indicates a poor quality of the classification. For the Eastern areas and Woodlark (using Medium Neural Networks), we obtained a classification success rate of 71.3%, compared to 20.6% by random assignment (p=0.007, indicating significant difference between random and classification success). These figures can be compared to the classification of larval samples in Fig. 5a and 5c of the manuscript, respectively (although in Fig. 5a sites were considered, which cannot be done with water samples because of the small number of measurements). Water samples show discrimination of Eastern vent areas, but the possible confusions differ depending on the type of sample (i.e. water or larvae; present Fig. 1a and manuscript Fig 5c). However, the success rate for Western areas (present Fig. 1c) is more likely due to the proportion of measurements between the two areas (Manus and Woodlark basins), which is 80% Manus in origin (12 out of 15 measurements), reflected by the high p-value when comparing classification success rate with random assignment. So we really think that because of the number of water data available, it is better not to use these analyses.
To conclude, we consider that water partly controls the elemental composition of larval shells, and that biological factors (e.g. encapsulation, physiology…) also play an important role. This leads to the need to use the elemental composition of encapsulated shelled larvae as a reference, and not water chemistry directly.
If you agree with these clarifications, we will amend the manuscript to include this precision.
-
AC3: 'Final reply on RC2', Vincent Mouchi, 11 Nov 2023
Final reply to RC2 'Comment on egusphere-2023-1324', Steffen Kiel
Reviewer’s comments are in italics. Author responses are in bold.
“There is, however, one issue that the authors should check. In the last paragraph before the conclusions (lines 292-293), they write ‘our data clearly evidenced that chemical composition of habitat water is not a reliable predictor of shell chemistry, and thus cannot be used to assign the origin of individuals’. This raises the question of the overall role of water chemistry in larval shell chemistry, which in turn puts the whole approach into question: if larval shell chemistry is largely biologically controlled, it would have little geographic meaning.
What the authors could do is the following test: do the same set of tests for habitat water chemistry as they have done for larval shell chemistry, and check if the geographic structure found through both sets of tests match. It is entirely possible (and in my view likely) that water chemistry does indeed determine larval shell chemistry. However, there could be various, possibly biologically controlled, inter-dependencies among the elements that make it difficult to find a straightforward, one-to-one match between water and shell chemistry. By comparing the geographic structure of habitat water chemistry vs. larval shell chemistry, that problem could be avoided."We thank you for your interest and your assessment of our work.
As mentioned in the open discussion phase (AC1), we have now clarified that 1) habitat water composition does differ between hydrothermal sites, 2) it does influence the composition of larval shells, but 3) shell chemistry is also controlled by biological processes, meaning that habitat water alone, although largely responsible for differences in shell composition, is not sufficient to be used directly to assign the origin of larvae. Because of the biological processes that alter the link between water and shell chemistry, site-specific references must be built using the local composition of shells, not water. We have now clarified this point throughout.
We have also performed the analysis suggested in this comment (see AC1), but as there is too little power due to the lower number of measurements (habitat water was analyzed from a single sample per site, compared to shell chemistry analyses), we have chosen not to add it to the manuscript.Citation: https://doi.org/10.5194/egusphere-2023-1324-AC3
-
AC1: 'Reply on RC2', Vincent Mouchi, 15 Oct 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1324', Lisa Levin, 12 Aug 2023
Review of Mouchi et al. A step towards measuring connectivity in the deep sea.General comments:
This work represents an important first step in the application of trace-elemental fingerprinting to deep-sea hydrothermal vent connectivity. The authors have selected a suitable test species and studied an impressive array of sites (14). I think this effort is a precursor to many important applications.
Although the authors present deep seabed mining as the key motivator , I would argue that there are many additional reasons for wanting to know about connectivity of vents – from basic science, to biodiversity conservation (30x30 goals), to addressing consequences of climate change on connectivity (e.g., Mitarai’s work in Levin et al. 2020 DOI: 10.1111/gcb.15223), vent roles in the carbon cycle etc. I would encourage at least mention of these other motivators. Within the mining realm, in addition to understanding consequences of mining disturbance, the authors should point out the importance of connectivity data to the designation of no-mining protected areas (APEIs) and also reference zone PRZ and IRZ (preservation and impact) designation.
Please consider adding to the end of the introduction a paragraph that clearly lays out the goals or objectives of the research. This could be in the form of questions, hypotheses or other… but should frame the science around the data presented in the paper. E.g., does the water chemistry of ‘habitat water’ differ among vent sites where S. tollmanni egg capsules occur? What elements are key to distinguishing sites? Does the trace element signatures of S. tollmanni larval shell reflect the habitat water chemistry? Are there specific sites or scales of connectivity where the application of trace elemental fingerprinting to the vent systems is likely to most reliable? What elements are key to this distinction? Some of these questions show up as the headings in the results section – but should be presented earlier.
Tell us a little bit more about the study species Shinkailepas tollmanni – its distribution, and its host Iffremeria distribution. Is the relationship obligate? Is anything known about depth ranges, longevity, development time, feeding mode, planktonic duration? How does its life history affect inferences about connectivity? Will information about mollusc/gastropod connectivity be relevant to other vent taxa – what will or won’t?
Consider discussion which of the study sites are targeted for mining, and which might serve as source populations. IUCN has red listed some species like the scaly foot snail based on their limited occurrence primarily in areas targeted for mining. I realize you don’t have any source or sink data generated yet but it might be useful to explain how this precursor work can lead to analyses that inform identification of vulnerable species.
Specific comments.
Title – please only hyphenate deep sea when it is used as a double adjective. Here it is a noun and should not be hyphenated.
Abstract: Line 19. The presence of capsules not only facilitates sampling, it means the embryos develop in a fixed location – and form carbonates whose trace elemental signature could reflect that location.
Line 30 – Even vent sites on mid ocean ridges far from coastlines are of interest to commercial mining (ISA contracts on Mid Atlantic Ridge and Southwest Indian Ridge).
Line 65 – replace ‘minute’ with a more accurate indication of size…. 100 micron?
Line 72 – replace ‘which carbonate shells ‘with ‘whose carbonate shells’
Line 263 Simmonds et al. should come out of the parentheses.
Note that Levin 2006 (doi:10.1093/icb/icj024) in discussing future directions for larval dispersal studies in a larval dispersal review wrote How much larval exchange occurs within and among reducing ecosystems such as vents, seeps, and whale falls? Analysis of short-term larval exchange among seep or hydrothermal vent ecosystemsmight be tractable if these impart distinct trace element signatures to larval shells.
I think this paper is really the first to tackle this problem. (17 years later).
Fig. 4 and 5 caption. Indicate what the lighter and darker shades of blue and red mean.
The supplemental information needs a table of contents as it is extensive.
Table S4 should replace the ‘,’ with decimal points.
The comparison of habitat water to larval shell chemistry should be included in the main paper.
Overall this is a significant contribution in need of relatively minor revision.
Citation: https://doi.org/10.5194/egusphere-2023-1324-RC1 -
AC2: 'Reply on RC1', Vincent Mouchi, 11 Nov 2023
Final reply to the comments from RC1:
Review of Mouchi et al. A step towards measuring connectivity in the deep sea.
Reviewer’s comments are in italics. Author responses are in bold.
General comments:
This work represents an important first step in the application of trace-elemental fingerprinting to deep-sea hydrothermal vent connectivity. The authors have selected a suitable test species and studied an impressive array of sites (14). I think this effort is a precursor to many important applications.We thank you for your interest in our study.
Although the authors present deep seabed mining as the key motivator, I would argue that there are many additional reasons for wanting to know about connectivity of vents – from basic science, to biodiversity conservation (30x30 goals), to addressing consequences of climate change on connectivity (e.g., Mitarai’s work in Levin et al. 2020 DOI: 10.1111/gcb.15223), vent roles in the carbon cycle etc. I would encourage at least mention of these other motivators. Within the mining realm, in addition to understanding consequences of mining disturbance, the authors should point out the importance of connectivity data to the designation of no-mining protected areas (APEIs) and also reference zone PRZ and IRZ (preservation and impact) designation.
We agree that the listed motivators (in addition to deep-sea mining) are also important. We added some references to include these in the introduction: “As a basic process of population dynamics, connectivity is key to understand the persistence of vent communities as well as the potential of vent species to colonize new areas and habitats (e.g., Adams et al. 2012; Levin et al. 2016). As such, it is of major interest for biodiversity conservation and the design of deep-sea marine protected areas (Danovaro et al. 2020; Combes et al., 2021), for example in the framework of the 30x30 initiative, which aims to protect at least 30% of the world’s oceans by 2030 (O’Leary et al. 2019). In the particular context of deep-sea mining, connectivity has been identified as a key scientific knowledge gap (Gollner et al., 2017; Miller et al., 2018; Smith et al., 2020; Amon et al., 2022) that should be tackled taking into account its interactions with climate change (Levin et al., 2020)”.
Also, we mention protected areas when we give details on how we can use elemental fingerprinting (see last general comment below).Please consider adding to the end of the introduction a paragraph that clearly lays out the goals or objectives of the research. This could be in the form of questions, hypotheses or other… but should frame the science around the data presented in the paper. E.g., does the water chemistry of ‘habitat water’ differ among vent sites where S. tollmanni egg capsules occur? What elements are key to distinguishing sites? Does the trace element signatures of S. tollmanni larval shell reflect the habitat water chemistry? Are there specific sites or scales of connectivity where the application of trace elemental fingerprinting to the vent systems is likely to most reliable? What elements are key to this distinction? Some of these questions show up as the headings in the results section – but should be presented earlier.
We added a paragraph at the end of the introduction to list the main questions the manuscript aims at answering: “To explore the use of elemental fingerprints of larval shells in this species, several questions were investigated. Are there enough chemical contrasts between hydrothermal vent sites for elemental fingerprinting? Do elemental fingerprints of larval shells reflect their natal place? What are the main elements responsible for the elemental fingerprint of larval shells? At which spatial scale is elemental fingerprint more accurate? Does the elemental fingerprint of larval shells correspond to that of habitat water, in the sense that the habitat water fingerprint could be used as a reference for the shell fingerprint?”.
Tell us a little bit more about the study species Shinkailepas tollmanni – its distribution, and its host Iffremeria distribution. Is the relationship obligate? Is anything known about depth ranges, longevity, development time, feeding mode, planktonic duration? How does its life history affect inferences about connectivity? Will information about mollusc/gastropod connectivity be relevant to other vent taxa – what will or won’t?
We added available information on the biology of the studied species as requested by RC1 in a new first sub-section of the Materials and Methods. Most questions asked by RC1 about its life history traits were addressed, except the question of its longevity, for which no report exists, to our knowledge. At the end of this new sub-section, we briefly expose how larval history traits were used to interpret observations on genetic connectivity: “Based on morphological characteristics (e.g. size of encapsulated veligers vs size of postlarvae), larvae of S. tollmanni are considered planktotrophic, and are assumed to stay pelagic for up to one year (Yahagi et al., 2020). Although no direct evidence exists, larvae of S. tollmanni are assumed to disperse in surface waters as shown for other Phenacolepadidae (Yahagi et al., 2020). Such dispersal abilities could explain the genetic homogeneity observed among populations of S. tollmanni at the scale of the study area (Yahagi et al., 2020; Poitrimol et al., 2022)”.
Unfortunately, for now, we cannot reliably give information on how connectivity interpretation inferred from elemental fingerprinting assignment (we don’t have the necessary data yet; see reply below) would be ‘relevant to other vent taxa’. In particular, we don’t have the data to discuss the possibility of using elemental fingerprint references from S. tollmanni to assign the origin of individuals from other taxa occurring on the same vent sites.Consider discussion which of the study sites are targeted for mining, and which might serve as source populations. IUCN has red listed some species like the scaly foot snail based on their limited occurrence primarily in areas targeted for mining. I realize you don’t have any source or sink data generated yet but it might be useful to explain how this precursor work can lead to analyses that inform identification of vulnerable species.
The identification of source and/or sink populations is currently impossible, as we have no assignment of juvenile individuals from our method. We however added in the introduction specifics to how the assignment of juveniles from elemental fingerprints of their larval shells can indeed identify these populations, and help determine which of these sites should be particularly critical for the survival of the studied species: “Assigning the origin of juveniles from elemental fingerprints will provide rare evidence of single-generation connectivity ranges and identify which sites correspond to sources and sinks for populations over a wide area. This type of information can subsequently be used to identify sites of higher priority for preservation and protection from disturbance to ensure the resilience of populations (e.g., no-mining areas known as Areas of Particular Environmental Interest, or APEIs, International Seabed Authority, 2011; Dunn et al. 2018; Preservation Reference Zones and Impact Reference Zones; International Seabed Authority, 2019). Moreover, identifying source and sink population and migration rates for many species would therefore help identify the species vulnerable to disturbance or even destruction of a specific habitat.”.
Specific comments:
All the specific comments by RC1 have been answered in the corrected manuscript. In particular, as suggested, we transferred the details on the correlations between habitat water and larval shells from Supplementary Information to the main text, in section 3.3.
Adams, D.K., Arellano, S.M., and Govenar, B.: Larval dispersal: vent life in the water column, Oceanography, 25, 256–268, http://dx.doi.org/10.5670/oceanog.2012.24, 2012.
Amon, D. J., Gollner, S., Morato, T., Smith, C. R., Chen, C., Christiansen, S., Currie, B., Drazen, J. C., Fukushima, T., Gianni, M., Gjerde, K. M., Gooday, A. J., Grillo, G. G., Haeckel, M., Joyini, T., Ju, S.-J., Levin, L. A., Metaxas, A., Mianowicz, K., Molodtsova, T. N., Narberhaus, I., Orcutt, B. N., Swaddling, A., Tuhumwire, J., Palacio, P. U., Walker, M., Weaver, P., Xu, X.-W., Mulalap, C. Y., Edwards, P. E. T., and Pickens, C.: Assessment of scientific gaps related to the effective environmental management of deep-seabed mining, Marine Policy, 138, 105006, https://doi.org/10.1016/j.marpol.2022.105006, 2022.
Combes, M., Vaz, S., Grehan, A., Morato, T., Arnaud-Haond, S., Dominguez-Carrió, C., Fox, A., González-Irusta, J. M., Johnson, D., Callery, O., Davies, A., Fauconnet, L., Kenchington, E., Orejas, C., Roberts, J. M., Taranto, G., and Menot, L.: Systematic Conservation Planning at an Ocean Basin Scale: Identifying a Viable Network of Deep-Sea Protected Areas in the North Atlantic and the Mediterranean, Frontiers in Marine Science, 8, 2021.
Danovaro, R., Fanelli, E., Aguzzi, J., Billett, D., Carugati, L., Corinaldesi, C., Dell’Anno, A., Gjerde, K., Jamieson, A. J., Kark, S., McClain, C., Levin, L., Levin, N., Ramirez-Llodra, E., Ruhl, H., Smith, C. R., Snelgrove, P. V. R., Thomsen, L., Van Dover, C. L., and Yasuhara, M.: Ecological variables for developing a global deep-ocean monitoring and conservation strategy, Nat Ecol Evol, 4, 181–192, https://doi.org/10.1038/s41559-019-1091-z, 2020.
Dunn, D. C., Van Dover, C. L., Etter, R. J., Smith, C. R., Levin, L. A., Morato, T., Colaço, A., Dale, A. C., Gebruk, A. V., Gjerde, K. M., Halpin, P. N., Howell, K. L., Johnson, D., Perez, J. A. A., Ribeiro, M. C., Stuckas, H., Weaver, P., and SEMPIA Workshop Participants: A strategy for the conservation of biodiversity on mid-ocean ridges from deep-sea mining, Sci. Adv., 4, eaar4313, https://doi.org/10.1126/sciadv.aar4313, 2018.
Gollner, S., Kaiser, S., Menzel, L., Jones, D. O. B., Brown, A., Mestre, N. C., van Oevelen, D., Menot, L., Colaço, A., Canals, M., Cuvelier, D., Durden, J. M., Gebruk, A., Egho, G. A., Haeckel, M., Marcon, Y., Mevenkamp, L., Morato, T., Pham, C. K., Purser, A., Sanchez-Vidal, A., Vanreusel, A., Vink, A., and Martinez Arbizu, P.: Resilience of benthic deep-sea fauna to mining activities, Marine Environmental Research, 129, 76–101, https://doi.org/10.1016/j.marenvres.2017.04.010, 2017.
International Seabed Authority: Environmental management plan for the Clarion-Clipperton Zone. ISBA/17/LTC/7, https://www.isa.org.jm/documents/isba-17-ltc-7/, 2011.
International Seabed Authority: Report of ISA workshop on the design of “Impact Reference Zones” and “Preservation Reference Zones” in deep-sea mining contract areas, ISA Technical Study No 21, https://www.isa.org.jm/publications/technical-study-21-report-of-isa-workshop-on-the-design-of-impact-reference-zones-and-preservation-reference-zones-in-deep-sea-mining-contract-areas/, 2019.
Levin, L. A., Baco, A. R., Bowden, D. A., Colaco, A., Cordes, E. E., Cunha, M. R., Demopoulos, A. W. J., Gobin, J., Grupe, B. M., Le, J., Metaxas, A., Netburn, A. N., Rouse G. W., Thurber, A. R., Tunnicliffe, V., Van Dover, C. L., Vanreusel, A., and Watling, L.: Hydrothermal vents and methane seeps: rethinking the sphere of influence, Frontiers in Marine Science, 3, 72, https://doi.org/10.3389/fmars.2016.00072, 2016.
Levin, L. A., Wei, C.-L., Dunn, D. C., Amon, D. J., Ashford, O. S., Cheung, W. W. L., Colaço, A., Dominguez-Carrió, C., Escobar, E. G., Harden-Davies, H. R., Drazen, J. C., Ismail, K., Jones, D. O. B., Johnson, D. E., Le, J. T., Lejzerowicz, F., Mitarai, S., Morato, T., Mulsow, S., Snelgrove, P. V. R., Sweetman, A. K., and Yasuhara, M.: Climate change considerations are fundamental to management of deep-sea resource extraction, Global Change Biology, 26, 4664–4678, https://doi.org/10.1111/gcb.15223, 2020.
Miller, K. A., Thompson, K. F., Johnston, P., and Santillo, D.: An Overview of Seabed Mining Including the Current State of Development, Environmental Impacts, and Knowledge Gaps, Front. Mar. Sci., 4, 418, https://doi.org/10.3389/fmars.2017.00418, 2018.
O’Leary, B. C., Allen, H. L., Yates, K. L., Page, R. W., Tudhope, A. W., McClean, C., Rogers, A. D., Hawkins, J. P., and Roberts, C. M.: 30x30: A blueprint for ocean protection, Umweltstiftung Greenpeace, www.greenpeace.org/30x30, 2019.
Poitrimol, C., Thiébaut, É., Daguin-Thiébaut, C., Le Port, A.-S., Ballenghien, M., Tran Lu Y, A., Jollivet, D., Hourdez, S., and Matabos, M.: Contrasted phylogeographic patterns of hydrothermal vent gastropods along South West Pacific: Woodlark Basin, a possible contact zone and/or stepping-stone, PLoS ONE, 17, e0275638, https://doi.org/10.1371/journal.pone.0275638, 2022.
Smith, C. R., Tunnicliffe, V., Colaço, A., Drazen, J. C., Gollner, S., Levin, L. A., Mestre, N. C., Metaxas, A., Molodtsova, T. N., Morato, T., Sweetman, A. K., Washburn, T., and Amon, D. J.: Deep-Sea Misconceptions Cause Underestimation of Seabed-Mining Impacts, Trends in Ecology & Evolution, 35, 853–857, https://doi.org/10.1016/j.tree.2020.07.002, 2020.
Yahagi, T., Thaler, A. D., Van Dover, C. L., and Kano, Y.: Population connectivity of the hydrothermal-vent limpet Shinkailepas tollmanni in the Southwest Pacific (Gastropoda: Neritimorpha: Phenacolepadidae), PLoS ONE, 15, e0239784, https://doi.org/10.1371/journal.pone.0239784, 2020.Citation: https://doi.org/10.5194/egusphere-2023-1324-AC2
-
AC2: 'Reply on RC1', Vincent Mouchi, 11 Nov 2023
-
RC2: 'Comment on egusphere-2023-1324', Steffen Kiel, 28 Sep 2023
This is an interesting study exploring the use of the elemental fingerprint of gastropod larval shells to predict their origin at certain hydrothermal vent sites. Overall, the study is well executed and described, and is worth publication.
There is, however, one issue that the authors should check. In the last paragraph before the conclusions (lines 292-293), they write ‘our data clearly evidenced that chemical composition of habitat water is not a reliable predictor of shell chemistry, and thus cannot be used to assign the origin of individuals’. This raises the question of the overall role of water chemistry in larval shell chemistry, which in turn puts the whole approach into question: if larval shell chemistry is largely biologically controlled, it would have little geographic meaning.
What the authors could do is the following test: do the same set of tests for habitat water chemistry as they have done for larval shell chemistry, and check if the geographic structure found through both sets of tests match. It is entirely possible (and in my view likely) that water chemistry does indeed determine larval shell chemistry. However, there could be various, possibly biologically controlled, inter-dependencies among the elements that make it difficult to find a straightforward, one-to-one match between water and shell chemistry. By comparing the geographic structure of habitat water chemistry vs. larval shell chemistry, that problem could be avoided.
Citation: https://doi.org/10.5194/egusphere-2023-1324-RC2 -
AC1: 'Reply on RC2', Vincent Mouchi, 15 Oct 2023
We thank you for your assessment of our work and your interest.
Before going any further in modifying the manuscript, we would like to clear up any ambiguity, as we believe there could be a misunderstanding of our statement.
We perfectly agree that water chemistry is responsible, at some degree, of the shell elemental composition, as it is the core rationale behind the elemental fingerprint approach. We showed a weak but significant correlation between water and shell composition for some elements, which supports the role of habitat water on shell composition (Supplementary Information S3 of our initial manuscript, but will be incorporated in the main manuscript as requested by RC1). Because of this weak correlation, we then state that other factors (both biological and mineralogical) induce variations of this elemental composition and prevent using directly the exact composition of the water as a reference for shell elemental fingerprinting, requiring to determine the elemental fingerprint from the larval shells themselves, to be used as a reference.
Unfortunately, the classification of habitat water composition you request would be unreliable. Indeed, we won’t be able to compare ‘patterns’ of different datasets from the same method, due to the algorithms of the classification methods we use here. Contrary to methods such as Principal Component Analysis (whose results are unconclusive here; see Fig. 3 of the manuscript), most classification methods use artificial intelligence to cluster data, and representations of these clusters are not possible (explanations of some of these methods were given in Supplementary Information S4), much more so using multi-elemental datasets. Finally, the sample size of water measurements is substantially smaller than that of the larval shells (n=1 to 7 for sites where larvae were collected, mean sample size = 2.6), which prevents the classification at the site scale, and at larger scale the quality of the classification may not be as accurate, potentially overestimating the success rate. We nevertheless attempted the classification procedures for this comment reply (Fig. 1 attached), using Mg, Mn, Fe, Cu, Zn and Sr (the elements we used to compare habitat water with larval shells) from 15 and 17 habitat water measurements (for Western sites and Woodlark, and Eastern sites and Woodlark, respectively) corresponding to the larvae collection sites. We obtained 79.2% accuracy for Manus and Woodlark (using Cubic SVM), compared to 64.5% by random assignment, with a p-value of 0.075, which indicates a poor quality of the classification. For the Eastern areas and Woodlark (using Medium Neural Networks), we obtained a classification success rate of 71.3%, compared to 20.6% by random assignment (p=0.007, indicating significant difference between random and classification success). These figures can be compared to the classification of larval samples in Fig. 5a and 5c of the manuscript, respectively (although in Fig. 5a sites were considered, which cannot be done with water samples because of the small number of measurements). Water samples show discrimination of Eastern vent areas, but the possible confusions differ depending on the type of sample (i.e. water or larvae; present Fig. 1a and manuscript Fig 5c). However, the success rate for Western areas (present Fig. 1c) is more likely due to the proportion of measurements between the two areas (Manus and Woodlark basins), which is 80% Manus in origin (12 out of 15 measurements), reflected by the high p-value when comparing classification success rate with random assignment. So we really think that because of the number of water data available, it is better not to use these analyses.
To conclude, we consider that water partly controls the elemental composition of larval shells, and that biological factors (e.g. encapsulation, physiology…) also play an important role. This leads to the need to use the elemental composition of encapsulated shelled larvae as a reference, and not water chemistry directly.
If you agree with these clarifications, we will amend the manuscript to include this precision.
-
AC3: 'Final reply on RC2', Vincent Mouchi, 11 Nov 2023
Final reply to RC2 'Comment on egusphere-2023-1324', Steffen Kiel
Reviewer’s comments are in italics. Author responses are in bold.
“There is, however, one issue that the authors should check. In the last paragraph before the conclusions (lines 292-293), they write ‘our data clearly evidenced that chemical composition of habitat water is not a reliable predictor of shell chemistry, and thus cannot be used to assign the origin of individuals’. This raises the question of the overall role of water chemistry in larval shell chemistry, which in turn puts the whole approach into question: if larval shell chemistry is largely biologically controlled, it would have little geographic meaning.
What the authors could do is the following test: do the same set of tests for habitat water chemistry as they have done for larval shell chemistry, and check if the geographic structure found through both sets of tests match. It is entirely possible (and in my view likely) that water chemistry does indeed determine larval shell chemistry. However, there could be various, possibly biologically controlled, inter-dependencies among the elements that make it difficult to find a straightforward, one-to-one match between water and shell chemistry. By comparing the geographic structure of habitat water chemistry vs. larval shell chemistry, that problem could be avoided."We thank you for your interest and your assessment of our work.
As mentioned in the open discussion phase (AC1), we have now clarified that 1) habitat water composition does differ between hydrothermal sites, 2) it does influence the composition of larval shells, but 3) shell chemistry is also controlled by biological processes, meaning that habitat water alone, although largely responsible for differences in shell composition, is not sufficient to be used directly to assign the origin of larvae. Because of the biological processes that alter the link between water and shell chemistry, site-specific references must be built using the local composition of shells, not water. We have now clarified this point throughout.
We have also performed the analysis suggested in this comment (see AC1), but as there is too little power due to the lower number of measurements (habitat water was analyzed from a single sample per site, compared to shell chemistry analyses), we have chosen not to add it to the manuscript.Citation: https://doi.org/10.5194/egusphere-2023-1324-AC3
-
AC1: 'Reply on RC2', Vincent Mouchi, 15 Oct 2023
Peer review completion
Journal article(s) based on this preprint
Vincent Mouchi et al.
Vincent Mouchi et al.
Viewed
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 220 | 0 | 2 | 222 | 4 | 4 |
- HTML: 220
- PDF: 0
- XML: 2
- Total: 222
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.