Review of the manuscript “Carbonatite-induced petit-spot melts squeezed upward from the asthenosphere beneath the Jurassic Pacific Plate” by Mikuni, K. et al.

General comments

The manuscript presents new geochemical and geochronological (40Ar/39Ar) data on lavas from petit-spot volcanism in western Pacific Ocean. By using trace element and Sr-Nd-Pb isotopes, the authors model the mantle source from which basaltic melts were generated as a garnet lherzolite fluxed by small amounts of carbonatite melts, and speculate about recycling of crustal components into the mantle. According to their model, carbonatite melt and recycled oceanic crust components induced partial melting at the lithosphere-asthenosphere boundary (LAB).

The case study is extremely interesting, making the data worth of being published. On the other hand, a poor data filtering, coupled to a general lack of rigour and structure through the modelling and discussion, make the manuscript hard to follow, thus making a major revision necessary.

The authors should focus more in detail on the presented model, maybe cutting some side information (e.g., mineral chemistry and treatment of mantle xenoliths in the supplementary data, or all other informations in the main text that were already published elsewhere) to leave space to the main message of their study. Some results (e.g., negative Ar/Ar ages) must be also better supported to be included in the manuscript.

English language should be also improved.

Specific comments

• Title: the title is totally un-representative of the content of the paper. A better title could be “Crustal recycled and carbonate components required for the genesis of petit-spot alkaline basalts”, or something similar.
• Abstract: the data in the manuscript do not contribute to define the origin nor the structure of the LAB. The model provided is only petrogenetic. The opening of the abstract must be modified.
• Lines 74-82: The list of alkali basalts from random worldwide localities is not very meaningful, as readers might be not aware of each specific geodynamic context. Better re-shape this section, discriminating between tectonic settings and inferred geodynamic context, i.e. plume vs. not plume-related origin, continental vs. oceanic, etc….as it is done for petit-spot volcanoes shortly below. Intracontinental alkali basalts are not occurring only in China and North America; intraoceanic plume-related alkali basalts are not occurring only in Hawaii
• Lines 87-88: references are needed here! Falloon & Green 1989 EPSL; Falloon & Green 1990 Geology; Foley et al. 2009 Lithos; Ghosh et al. 2009 ChemGeo
• Lines 124-131: In this (long) sentence, the concepts of LAB nature and regional petit-spot volcanism are put together, but I feel some explanation is missing. Petit-spot volcanism provides info about asthenospheric melting as many other types of volcanism do. Why should the reader find this type of magmatism particularly useful? This section is the key of the scientific background, as it links the “big picture” to the case study!
• Section 4.4 (geochemical modelling) should be removed from the methodology and included later in the modelling/discussion section, at line 679. At the moment is out of context, and forces the reader to go back to 4.4 section while being in the middle of the discussion about the nature of the mantle source and the partial melting processes.
• Geochronology: 40Ar/39Ar dating on these samples is for sure problematic, due to the intense state of alteration. However, at the moment the presentation of the data is not convincing enough to be acceptable for publication: no discussion about K/Ca as a function of alteration is provided; the negative age (-0.11 Ma) obtained for the sample R01 is simply meaningless, and obtained with only 49% of released 39Ar; why for sample 1466R6-001 was the inverse isochron age chosen? More details are needed. On the other hand, the age ranges obtained using Fe-Mn crust or palagonite are interesting and useful.
• Geochemistry: along the whole results and discussion sections, sample 1466-R6 and R7 are treated as “anomalous” Si-undersaturated rocks, more vesicular, ascribable to a genesis from a different magma suite and thus mantle source (lines 564). This is a crucial issue, because authors use this discrepancy to speculate about different mantle sources for the magmatism. However, from Table 2 it is clear that samples 1466-R6 and R7 have the highest LOI (up to 6 wt%), as reported by the authors themselves later on (lines 594-596; lines 604 “…alteration was not extensive…excluding sample 1466R7-001, 1466R7-003”). In Supplementary Fig. S1 we see that sample 1466-R7 contains the most Fe-rich olivine phenocrysts, but at the same time also altered peridotite xenoliths or xenocrysts are included in this sample (Fig. S4). The question is: is it possible that the whole-rock analyses are just effect of alteration, or have just been “doped” because xenoliths/xenocrysts fragments were simply mixed up with the host rock during sample preparation? This would also explain why Sr-Nd-Pb isotopes are not showing any difference with respect to the other samples (lines 575-576), not requiring any differential contribution of carbonatite flux to the source (lines 584-586). As large portions of fresh glass are analysable in these samples (Fig. 4a, BSE image), is it possible to compare the whole-rock analysis of sample 1466-R7 with the composition of its glass (as done for the other samples) and thus verify the difference between the two magmatic suites?
• Chapter 6.2, containing the petrographic description of the samples and the evaluation of their alteration state and chemical representativeness, must be moved right at the beginning of the result section.
• Lines 670-673: Carbonatites are not depleted in U, Th and Nb. A typical feature is the high Nb/Ta, together with low Zr-Hf. Include reference, and/or plot an average carbonatite trace element pattern in Fig. 7 (using the same dataset as that for isotopes).
• Why for major elements and isotopes are author using the whole carbonatites GEOROC database, but for the partial melting model (section 6.4 and Fig. 11) only the “average carbonatite” of Bizimis et al. (2003) is used?
• Lines 698-707: I agree the partial melting of garnet lherzolite with small crustal component + low carbonatite flux is the most plausible among the model performed, but still this model is not able to fully reproduce the Ta, La, Sm-Eu-Ti and Y concentrations. Is there a reason? Did authors try to change the modal composition of the source, melting contribution percentages and/or amount of recycled oceanic crustal material to better fit the petit-spot pattern?
• Lines 732-747, Isotopic modelling: as authors in the previous section model the percentage of recycled crustal components, peridotitic fertile mantle and carbonatitic flux, it would be extremely interesting to verify if such percentages are also applicable to the isotopic mixing models. Day and Hilton (2011, EPSL; 2021, Geology) and Day et al. (2022, GCA) tried to perform such model for the Canary Island magmatism; the same was done by Markse et al. (2008, JPet) and Pietruszka et al. (2013, EPSL) for Hawaii.
• Chapter 6.4 is entirely speculative, does not add value nor constraints to the model. I suggest to remove it up to line 785. The content of lines 785-792, that is the hypotheses made to reconcile the model with the geodynamic context of the area, can be incorporated at the end of section 6.3.
• Lines 811-812: remove this sentence. In this paper the origin of LAB is not discussed at all.

Technical corrections

• Many analyses in the manuscript come from already published papers by the same authors. This information must be clarified through the text and the tables, as they cannot be presented as new analyses.
• Line 73-74: the opening sentence is too broad, as it is having a poor meaning. Maybe correct to: “the petrogenesis of alkali basalts and the nature/evolution of their source mantle…”. Which tectonic settings? Be more specific.
• Lines 104-105: Maybe better re-phrasing to: “…to understand the nature of the underlying lithosphere-asthenosphere system and model the geodynamic evolution of the region”
• Lines 109-112: Maybe the concept can be smoothened and made more concise: “In the last 20 years, the increasing knowledge of petit-spot volcanic settings has provided useful insights on the nature of the lithosphere-asthenosphere system, especially in the NW Pacific region (Hirano et al., 2006, Hirano & Machida 2022)”
• Lines 117-121: Sentence too long, can be cut to clarify the concept
• Lines 121-122: Where? Provide some names of localities, as this might be of broad interest to the reader!
• Line 129: what does “hybrid factor” mean?
• Line 276: EPMA analyses were not made on glasses only. Please correct and distinguish in this section the analytical conditions for glasses (defocussed beam?) from those used for minerals. How about matrix correction (Pouchou and Pichoir 1991)?
• Line 417 and figure 4a: alkaline and sub-alkaline; dividing line is from Irvine and Baragar (1971)
• Line 433: “…compared to the representative ocean island basalt (OIB)”
• Data from worldwide carbonatites presented in Fig. 9 are extremely scattered and not very useful, as they cover the entire diagram. Are they needed? If so, I recommend the authors to filter the dataset at least for the geodynamic settings or sample freshness/data quality.
• Line 556: “petrography and geochemistry”.
• Line 607-608: not entirely true. Some samples are indeed altered, and are specifically those used to argue for the heterogeneity of the mantle!
• Lines 617-619: “mass balance calculation of fractional phases….could not be performed because of inadequate phenocrysts” what does it mean?
• Lines 668-670: concept repeated.
• Line 729: Herzberg

Figures and Tables

• 1 can be re-organized in a 2x2 panel, as at the moment the panels are too small to be read properly. In the tomographic image to the right, maybe it would be useful to have also a more “zoomed” view of the upper 700-800 km.
• Line 144: Replace “although” with “Notwithstanding”
• 3: mineral labels should be added to all the optical microscope photos and BSE images. Why not splitting the figure in 4 different figures, or move some material to the supplementary information? As it is, it cannot be presented as a single figure composed by 4 multi-panels…
• 4: Panel b does not look very helpful, could be replaced with a K2O vs Na2O diagram to discriminate Na- from K-affinity. Also, plotting the literature data for kimberlites causes a “squeezing” of the OIB data down to K2O/Na2O <1.5, with all the data overlapping with each other. Is this really necessary?
• 5-6: I would recommend to plot MgO wt% in reverse order on the x-axis, so that to increase the degree of differentiation rightwards. This would also mark better the differences with respect to NW Pacific rocks, clearly visible on the CaO/Al2O3 vs. MgO plot.
• 7 caption: “Primitive-mantle- (PM, Sun & McDonough, 1989) normalized trace element…..”. Delete the last sentence.
• 9: carbonatite symbols in light green are poorly readable. The details about the mixing model in the caption can be moved to the supplementary material.
• 10b: provide the reference for the age of the detrital zircon in the Fig. caption.
• S9. Is it different from Fig. 11 in the main text? If not, why duplicate it?
• Table 2: Is this dataset including both glass analyses performed by EPMA and LA-ICP-MS and whole-rock analyses performed by XRF? If so, it must be clearly defined, and possibly the 2 samples set must be distinguishable from each other, not mixed. Values of 0.00 have no sense. Are they not analysed (n.a.) or below detection limit (b.d.l.)?
• Supplementary Table S1: mineral chemistry analyses must be filtered for data quality, at the moment they are totally un-presentable. Some olivine analyses sum up to <98 or >101 (sometimes >102!!!!), must be discarded. Some olivine analyses have Al2O3 >0.2 wt% (sometimes >1 wt%): must be deleted, are just mixed analyses. Filters must also be applied according to the quality of apfu calculations. Values 0.00 are not presentable: are they be not analysed (n.a.) or below detection limit (b.d.l.)? The same applies for cpx and plag, as many grains sum up to <98. Many spinel analyses have >0.8 wt% SiO2, must be deleted. One spinel analysis has 15 wt% SiO2, it is just a mixed analysis, delete.

General comments

This new manuscript from K. Mikuni and co-authors provides new petrological, geochemical and geochrological data on petit-spot lavas sampled in western Pacific. They also present a geochemical model, mainly based on trace elements, which along with isotopic analyses attempts to explain both source and formation of parental melts for such lavas. The authors speculate petit-spot basaltic melts originate from melting of a garnet lherzolite fluxed by small amounts of carbonatitic melts, with involvement of oceanic crustal components recycled in the mantle.

The study is worth of publishing, both data and modelling parts, bringing new light into the petit-spot volcanism, but suffers of many points, which should justify moderate to major revisions.

First, the geochemical model provided is unsufficiently explained. Although the model used refers to another study from Ozawa et al. (2001), some aspects regarding the different parameters used and how their variations could impact the final results would deserve to be more explained and explored. As an example, this is stated in the manuscript that flux of small amounts of carbonatitic melts are involved in forming petit-spot lavas, but the notion of small is relative. Could it be possible to have a more quantitative estimation of such amounts in %, or in terms of carbon content in the lherzolite source even approximative? Besides, could we consider drastic changes in the ultimate model results by varying the critical melt fraction of a certain amount? I suppose the latter could vary depending e.g. on the melt composition ('better' connectivity of a carbonatitic liquid compared to a basalt typically).

Regarding the origin of fluxing carbonatitic melts as discussed in l. 775-791 in section 6.4, some sources are discussed: “subducted carbonated pelite, pyroxenite/eclogite, or peridotite stored as diamond or metal carbide in the reduced lower portion of the upper mantle(….or) the existence of a carbonate-rich layer in the LAB owing to the horizontally spread carbonate from around the wedge mantle rather than upwelling from the deep mantle”. Carbonatites can be produced by melting of such lithologies, and I do not contest these possible origins, but since decades now, plethora of publications in experimental and modeling petrology have demonstrated upper parts of the asthenosphere, where the bottom layer of LAB is probably located, can be enough oxidized to have an ‘original’ source of carbon hosted as carbonates, and within the P-T conditions prevailing in this region, such carbonates can melt to form carbonatites (see e.g. Massuyeau et al., 2021). In this convective mantle, deep we get carbonatites and by ascending into the mantle, the melt composition gets close to a basalt with a kind of opposite trend between the enrichment in silica and the carbon depletion. Besides carbonatites can also be stabilized in the deepest and cold parts of the lithosphere. So, this scenario appears as most likely to explain formation of carbonatites in the LAB region than those explained above as source for carbonatites, and should at least deserve to be discussed too.

Ultimately, English language and most particularly the abusive use of past tense impedes to understand what is raised by this new study from what was already done in previous works already published.

Specific comments

• 44-46 : The two first sentences of the abstract should be reworked. As it, it can let imagine the manuscript will discuss about the origin of the geophysical discontinuities observed at LAB, while this is not the purpose of the manuscript. And the first sentence could be removed to my mind.
• 57-58 : “… partial melting of garnet lherzolite with a small degree of carbonatite melt flux with 58 crustal components.” à I understand the crustal component are brought by carbonatitic melts, which is probably not the case here, right?
• 59-60 : “…and provides an implication for the genesis of tectonic-induced volcanism with similar geochemical signatures to those of petit-spots.” à I do not understand this sentence.
• 68 : add a reference to the geochemical modeling done in the study
• 88-89 : It would be interesting to mention that with variations of P-T, as it can occur at LAB potentially, melting of carbonated peridotite can also produce a chemical continuum in the composition of produced melts, from carbonatite to basalt, with intermediate terms like kimberlites, nephelinite, melilitite, etc. It could possibly partly explain the variations obtained in terms of major elements between different series.
• 100 : Instead of “carbonatitic materials”, I would rather use “carbonated materials”.
• 101-105 : As the notion of LAB and its relationships with both mantle melting and geophysical properties have not been discussed yet (coming next in the Background section), I would replace "LAB" by "the uppermost part of the asthenosphere" or similar, something more neutral in the description.
• 119 : As you mention the LAB, it would be appropriate to have a short and general discussion/presentation about the latter and its relationships with mantle melting and geophysical signatures. Besides, once considered the notion of LAB, could we consider the bottom of a possibly metasomatized lithosphere as a possible source of petit-spot lavas too ?
• 143 : The authors mention a low-velocity zone, but no reference is mentioned. Same in line 144-145. Moreover, it would deserve more explanation , why do they need to mention. This is kind of information from nowhere.
• 451-452 : Here authors should also precise that the samples 1203 and 1206 are from another study (Hirano et al., 2019)
• 452-454 : Any kind of filtering done from the PetDB database ?
• 488 : “with Bizimis et al. (2003)” à it could be of interest to precise this point, e.g. as part of the Supplementary part
• 564 : “were suggested” à By who ? If this is the authors suggestion, they should use the present tense then, here readers could think they are mentioning another reference not cited.
• 595-596 : Could we simply imagine this LOI being a volatile (CO2-H2O) rest of the primary melt ? Does the Figure 8 completely rejects this hypothesis ? In terms of major elements (Fig. 5), it seems difficult to exclude completely this hypothesis relatively to other samples, except maybe the high FeO content. Does the latter could maybe favor your alteration scenario ?
• 655-660 : To reconsider in light of the previous comment regarding the nature of 1466R7.
• 741-744 : It seems to be in contradiction with what is said in previous sentences, no ?.. Maybe this sentence would need to be reworked a bit.
• 767-770 : I do not understand the point here, it seems both sentences repeat the same point with different locations for 'representative' carbonatites.
• 780-781 : “transition oxidation state” à it would deserve more explanation

Technical corrections

• 52 : remove “their”
• 65 : replace “moving” with “motion”
• 65 : replace “on” with “over”
• 92 : “to be explained”
• 101 : replace “LAB” with "Lithosphere-Asthenosphere Boundary (LAB)"
• 119-120 : replace “leading to an understanding” with “and further leading to a better understanding”
• 144 : replace “Although the” with “In spite of”
• 219 : “and comprised”
• 329 : replace “The” with “Both”
• 344 : replace “value was” with “values were”
• 377-378 : replace “In the model in this study” with “In this present study, the model uses a critical melt fraction”, and remove “was” in l. 379.
• 544 : replace “was” with “is”, the petit-spot volcanic field still exists I guess ;-)
• 566 : “K#1466R3” right ?
• 570 : replace “this” with “these”
• 590 : replace “might affect” with “might have affected”
• 733 : remove one “the”
• 787 : “Hammouda et al. 2021” ?