the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Jan 2026
Dating circulations of hydrothermal fluids in the crystalline basements of unconformity-related metallic deposits using in situ Rb/Sr geochronology: proof of concept
Abstract. The use of in situ Rb-Sr geochronology has boomed in recent years following its implementation using LA-ICP-QQQ-MS technology, which enables fast, in situ analyses at the micron scale on selected minerals. The Rb-Sr geochronometer applied to micas is now commonly used to date the crystallization or cooling of metamorphic and magmatic rocks, based on the assumptions of a closed isotopic system after passing the closure temperature and of a homogeneous Sr isotopic composition at the time of crystallization. In situ Rb-Sr geochronology applied to micas and related alteration products in geological contexts involving hydrothermal fluid circulation affecting micas after crystallization could provide a new way to decipher the timing and duration of fluid circulation in various settings such as mountain belts or sedimentary basins. The behavior and applicability of the Rb-Sr system in such contexts are, however, poorly understood, as the system may be partially reopened with differential redistribution of Rb and Sr at the grain scale. To test this hypothesis, we selected a case study related to unconformity-related U deposits from the Athabasca Basin (Canada), which formed through intense hydrothermal fluid circulation at the interface between crystalline basement and siliciclastic sedimentary rocks and represent archetypes of unconformity-related metallic deposits. Muscovite grains from metamorphic and magmatic rocks were targeted across a range of alteration states, from hydrothermally unaltered to strongly altered domains. We focused on a specific hydrothermal alteration linked to the formation of hydrothermal illite and sudoite at the expense of metamorphic or magmatic minerals. In unaltered zones, muscovite displayed variable but high Rb/Sr ratios, whereas the 87Sr/86Sr intercepts derived from Rb-Sr regressions were scattered and were not interpreted as meaningful initial isotopic compositions. The resulting ages ranged from ca. 1870 to ca. 1720 Ma and were consistent with the geological context. In distal-to-proximal alteration halos of U deposits, muscovite and related alteration products yielded lower 87Rb/86Sr ratios and highly variable regression intercepts. The mean age calculated across the different samples and investigated sites clustered around ~1640 Ma, a value previously obtained by Ar-Ar geochronology on illite and U-Pb geochronology on other hydrothermal phases and proposed to correspond to a major hydrothermal event linked to a geodynamic reorganization affecting the Canadian Shield at the circum-Laurentian scale. The ~1640 Ma age is geologically meaningful in the studied context and is interpreted as reflecting partial, micrometric-scale resetting of the Rb-Sr system in muscovite during this hydrothermal event. The wide range of regression intercept values commonly observed in disturbed Rb–Sr systems is interpreted as an apparent result of open-system behavior, reflecting partial system reopening and non-conservative redistribution of Rb and Sr at the grain scale, rather than as a physically meaningful initial isotopic composition. These results demonstrate that detailed analysis of Rb-Sr system perturbations in altered muscovite and related alteration products can constrain the timing of ancient hydrothermal activity and the spatial dynamics of fluid-rock interaction. This approach provides a valuable complement to conventional fluid-tracing methods and opens new perspectives for reconstructing paleo-hydrothermal systems in ancient basement terrains.
- Preprint
(4196 KB) - Metadata XML
-
Supplement
(314 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-6469', Anonymous Referee #1, 25 Feb 2026
-
AC1: 'Reply on RC1', Quentin Boulogne, 10 Apr 2026
Dear Referee,
We thank you for your careful evaluation of our manuscript and for your constructive comments. We have addressed all the technical points you raised. Please find below a point-by-point reply to your comments. Our responses are in bold.
Referee#1: Anonymous Referee
There is a lot to unpack in this contribution and I commend the authors on a comprehensive synthesis of this timely topic. The writing is excellent if somewhat dense in spots. I appreciate the attention to previous literature. The design of the project is thoughtful, and the data is collected using highly suitable methods. Incorporating textures and compositional data to support in situ Rb-Sr is essential and the authors do an overall excellent job of contextualizing the isotopic data. This is absolutely required in a setting like the Athabasca Basin where prolonged high-T to low-T hydrothermal activity has created a wide range of alteration assemblages during basin formation and readjustment. Integrating the in situ Rb-Sr data with models for U deposition is very novel indeed. Also particularly noteworthy is the suggestion, supported by the data, that partly altered white micas underwent preferential liberation of 87Sr* relative to non-radiogenic Sr. This could be a excellent case to follow-up using atom probe measurements to examine the distribution of Rb vs. 87Sr* and 86Sr in the mica lattice.
The only suggestion I can make is that the order of presentation of some important text could be improved. For example we read through all of the results and note that the initial Sr intercepts are impossibly low. I was questioning to myself if this was an analytical bias. It is only in the discussion that the robustness of La Posta secondary material is described which allays some fears of analytical bias. So perhaps it would be good to state more clearly the reproducibility of La Posta age AND initial Sr intercept at the end of the methods section.
We thank the reviewer for this constructive suggestion regarding the presentation of the La Posta secondary standard. In response, we have added a dedicated paragraph to the Methods section detailing the long-term reproducibility of both the isochron age and the initial 87Sr/86Sr intercept of La Posta. This section now includes a compilation of analytical sessions acquired over a three-year period. For each session, isochron ages and corresponding initial Sr isotopic ratios were calculated, and the resulting values were subsequently combined using weighted mean statistics, along with their associated uncertainties and dispersions. These results demonstrate robust long-term reproducibility of both age and initial Sr isotopic composition, and support the absence of instrumental drift, calibration bias, or matrix-dependent effects. We believe that this addition clarifies the reproducibility of La Posta prior to the Results section and addresses the reviewer’s concern.
Additionally, we provide the full results of this compilation as supplement attached to this response, in order to further document and illustrate the absence of analytical bias.
There are a few other minor things I’ve highlighted below. Notwithstanding these the manuscript is in very good shape and illuminates some very important processes affecting a world-class U camp.
Line 154: switching here from xxxx Ma to x.xx Ga.
We have revised Line 154 to report the ages in Ma rather than Ga, in order to maintain consistency with the other age values expressed in Ma throughout the manuscript.
Line 167: please consider italicizing P and T here and elsewhere
We have italicized P and T in Line 167 and throughout the manuscript for consistency with standard scientific notation.
Line 336-340: La Posta is listed as a secondary standard. But it is not obvious from the text how well it was reproduced based on NIST610/MicaMg-NP external calibration? I also note that no additional error was added to the final isochron ages and they are quoted as 1σ starting on line 456. This is a bit unconventional, with most studies using in situ Rb-Sr citing 2SE and including additional error reflecting the long-term reproducibility of secondary standards. Is the 1σ required to use the Gaussian Mixing Model? (this looks a lot like Isoplot ‘unmix age’ routine). What happens to this unmixing exercise if you use 2SE rather than 1σ? This gets back to La Posta results. What was the measured error on La Posta in this study? We are only given the Zack & Hogmalm 2016 reported value.
First, regarding the 1σ uncertainties reported for the isochron ages: in IsoplotR, the Gaussian Mixture Model (GMM) routine requires that each data point be associated with a 1σ uncertainty. These internal uncertainties are therefore used in the GMM analysis to weight contributions of individual analyses and to model the age distributions statistically. While some studies report 2SE to provide a conservative estimate of uncertainty, using 1σ is required for the GMM routine, and it accurately reflects the internal precision of the regression.
Second, to demonstrate long-term reproducibility, we compiled results from repeated analytical sessions of the La Posta secondary standard acquired over a three-year period. For each session, isochron ages and corresponding initial 87Sr/86Sr intercepts were calculated, and the resulting values were combined using weighted mean statistics, along with their dispersion. These results confirm stable and reproducible measurements. The total uncertainty on the reported isochrone ages is obtained by combining the internal regression error (1σ) with the external reproducibility derived from La Posta (~1.3%).
These additions are now described in a dedicated paragraph in the Methods section, making clear that the measured 87Sr/86Sr ratios and derived ages are robust, and that the reported 1σ uncertainties are both appropriate for the GMM analysis and consistent with long-term analytical reproducibility.
Line 499: Here and elsewhere in Results section I’m surprised to see some methodology details related to concentration calculations. Methods for calculating concentrations should be in an earlier section. If NIST610 was analyzed it would be a simple matter using the Trace Elements DRS in Iolite4.
We thank the reviewer for raising the point regarding the calculation of concentrations. We acknowledge that standard practice often employs the Trace Elements DRS routine in Iolite to derive concentrations automatically. However, in this study we calculated ⁸⁷Rb, ⁸⁶Sr, and ⁸⁷Sr concentrations using a rigorous, reproducible workflow based directly on measured isotopic ratios, regression outputs from Rb–Sr diagrams, and matrix corrections derived from the internal standard MicaMg.
To ensure transparency and clarity, we will add a dedicated paragraph in the Methods section detailing this calculation procedure, including the derivation of 87Rb from 85Rb, the calculation of 86Sr and 87Sr from measured ratios, and the application of the matrix correction factor. We believe that this addition will fully address the reviewer’s concern while demonstrating that our concentration calculations are scientifically robust and conform to standard practices in LA-ICP-MS geochronology.
Line 523: Is 1634.7 ± 1.7 Ma - an error of 0.1% - realistic? Again, how well as the secondary standard measured?
We thank the reviewer for raising this important point regarding the apparently small uncertainty associated with the isochron age.
The reported uncertainty of ±1.7 Ma corresponds to the 1σ internal regression error calculated by IsoplotR from 224 individual analyses defining the isochron. Although individual single-spot ages carry uncertainties on the order of ~20 Ma (1σ), the uncertainty on the isochron age reflects the statistical precision of the regression slope, which improves with the number of analyses. For a dataset of this size, the expected reduction in uncertainty follows approximately 1/√n. Given ~20 Ma single-spot uncertainties and n = 224 analyses, the theoretical reduction in uncertainty yields: 20 / √224 ≈ 1.3 Ma, which is consistent with the reported ±1.7 Ma. Therefore, the small reported internal uncertainty arises from the strong statistical constraint on the regression slope provided by the large number of analyses, rather than from unrealistically low analytical uncertainty on individual measurements.
We also note that the internal regression uncertainty does not account for long-term analytical variability. To provide a realistic estimate of the total uncertainty, the external reproducibility derived from the La Posta secondary standard (~1.3%, 1σ) is propagated. Combining the internal regression error with this external uncertainty yields a more representative total uncertainty on the isochron age, which in this case corresponds to approximately ±21 Ma (1σ). This ensures that reported ages reflect both the statistical precision of the regression and the long-term reproducibility of the analytical protocol.
Suggested statement for the manuscript:
“The Rb–Sr regression yields an age of 1634.7 ± 1.7 Ma (1σ internal regression error; the total uncertainty including external reproducibility is ~±21 Ma, 1σ).”
We have clarified in the revised manuscript that the quoted uncertainty represents the internal regression error (1σ) derived from the isochron fit.
Line 645: “…the mobility of radiogenic 87Sr* is significantly greater than that of Rb and non-radiogenic Sr…” The argument is that 87Sr* is preferentially located in interlayer site which is where alteration processes might start. Preferential removal of 87Sr* would, therefore, lower the 87Sr/86sr to values unsupported on earth. But the least-altered muscovite in basement gneiss also gives initial of 0.6709 ± 0.0115. This doesn’t make sense. The authors claim to have ruled out instrumental effects but is there some cryptic matrix-mismatch adversely affecting corrected 87Sr/86Sr? Some statement about the anomalously low initial in least-altered muscovite would be warranted (unless I missed it somewhere in the text)
Thank you for this comment. We believe there may be a misunderstanding, as the explanation for the anomalously low intercepts in the least-altered muscovite is already provided in the manuscript (Lines 686–705).
In that section, we explicitly state that the regression intercepts (e.g., 0.6704 ± 0.0075) fall below the minimum solar system 87Sr/86Sr value (~0.698) and therefore cannot represent physically meaningful initial isotopic compositions. We clearly indicate that these values are not interpreted as true initial 87Sr/86Sr ratios, but rather as apparent intercepts produced by disturbed Rb–Sr systematics.
We further explain that the well-correlated regressions, combined with low MSWD values (<1) and high p-values (p >> 0.05), indicate strong internal coherence despite the anomalous intercepts. We interpret this behavior as characteristic of rotated or disturbed isochrons formed under open-system conditions, where non-conservative redistribution of Rb and Sr (e.g., proportional Rb loss, selective Sr mobility, or partial isotopic resetting) modifies the intercept while preserving a geologically meaningful slope.
Line 697: I see here that the authors explain that age and initial intercept for La Posta were within accepted range and that anomalously low 87Sr/86Sr are not analytical artifacts. It would be nice to know this before launching into the results section where these anomalously low 87Sr/86Sr raise flags about this possibility.
Thank you for this constructive suggestion.
We agree that informing the reader earlier about the reproducibility of the La Posta secondary standard would improve clarity and help avoid potential concerns when encountering the anomalously low ⁸⁷Sr/⁸⁶Sr intercepts in the Results section.
Accordingly, we have added a dedicated paragraph to the Methods section explicitly detailing the analytical performance of the La Posta biotite secondary standard. This addition specifies both the reproducibility of the obtained age and the consistency of the initial 87Sr/86Sr intercept relative to accepted values. By presenting this information upfront, prior to the Results section, we aim to clearly demonstrate that the anomalously low intercepts observed in some samples are not attributable to analytical bias, calibration issues, or standardization problems.
We believe this modification strengthens the manuscript structure and improves the logical flow between the Methods and Results sections. Thank you again for highlighting this important point.
Line 723: typo in isochrone
The typo in “isochrone” (Line 723) has been corrected in the revised manuscript.
Line 728: back to x.xx Ga rather than xxxx Ma ages…
We have revised Line 728 to report the ages in Ma rather than Ga, in order to maintain consistency with the other age values expressed in Ma throughout the manuscript.
We believe that these revisions have significantly strengthened the manuscript, both scientifically and linguistically, and that it now meets the standards required for consideration in Geochronology. We sincerely thank you again for your insightful comments and for giving us the opportunity to resubmit an improved version of our work.
Yours sincerely,
Quentin Boulogne and co-authors
-
RC3: 'Reply on AC1', Anonymous Referee #1, 15 Apr 2026
Thanks to the authors for these clarifications. I hope to see the published version in due course.
Citation: https://doi.org/10.5194/egusphere-2025-6469-RC3 -
AC3: 'Reply on RC3', Quentin Boulogne, 27 Apr 2026
We thank the reviewer for their positive feedback and their time in evaluating our manuscript.
Citation: https://doi.org/10.5194/egusphere-2025-6469-AC3
-
AC3: 'Reply on RC3', Quentin Boulogne, 27 Apr 2026
-
RC3: 'Reply on AC1', Anonymous Referee #1, 15 Apr 2026
-
AC1: 'Reply on RC1', Quentin Boulogne, 10 Apr 2026
-
RC2: 'Comment on egusphere-2025-6469', Jarred Lloyd, 15 Apr 2026
The contribution, while somewhat dense in some sections, and needs refinement to the ordering of some content, is a comprehensive and timely contribution to the geochronology community. In general it is very well written, although I am unconvinced by some statements made by the authors; however, these concerns do not detract from the overall rigour of the paper nor the overall conclusions. I appreciate the attention to detail and prior literature. This assessment is based on the current version of the preprint, prior to the changes the authors have stated they've implemented in their reply to referee 1.
From my reading of the contribution, the aim is to investigate the utility and application of the Rb–Sr decay system geochronology in micas as a time record of hydrothermal events. This is done using samples from a Proterozoic basin in Canada that is economically important for uranium and is known to have been subject to long lived hydrothermal events.
On the use of 1s/sigma level - I note that the other referee had queried this as well, and the authors have replied to this. I would prefer to see these uncertainties referred to at 2s level so they represent the uncertainty parameter at the interval in which ~ 95 % of any random sample within the distribution μ ± 2s would fall, rather than ~ 68 %. I take the point that the GMM (and all statistical tests for that matter) require the uncertainty at 1s during computation; however, IsoplotR does this conversion so long as the appropriate input uncertainty level is noted. It can also be output as 2s or 95 % CI with the appropriate flag. Given you are also comparing Rb–Sr to existing geochronology from other studies that used different decay systems, it is appropriate to use 2s plus the decay constant uncertainty for comparison. You also do quote 2s uncertainties for other data in the paper.
On the 5 populations and BIC. IsoplotR's documentation notes that increasing sample size often causes the implemented algorithm to overestimate the number of populations as sample size increases, "This option should be used with caution, as the number of peaks steadily rises with sample size (n). If one is mainly interested in the youngest age component, then it is more productive to use an alternative parameterisation, in which all grains are assumed to come from one of two components, whereby the first component is a single discrete age peak (e^m, say) and the second component is a continuous distribution (as described by the central age model), but truncated at this discrete value."
The BIC itself probably shouldn't be used here as some of its assumptions break down in this particular type of modelling. A much more robust measure like the Bayes Factor (although computationally heavy) should probably be used here; however, this particular aspect is beyond the scope of your manuscript.
Instead, you should visually assess your distributions, taking note of the peakfit algorithm's comment quoted here. I'd interpret your data (notably in fig 8), as log-normal distributions of data, or as normal distributions with continuous tail, so at most 2 distributions, not 5.The uncertainty quartiles - why did you choose to plot these as histograms rather than density plots? Density plots aren't as sensitive to bin-width for visualisation (and you may as well fit distributions to them giving you mean and std). It looks like Q1 and Q2 have peaks (c. 1620 and 1645 Ma respectively) that correspond to the bimodal distribution of Q3, and then Q4 has a bimodal distribution with the lower peak on Q2 peak, and an additional peak at ~ 1800 Ma.
I'll note here, the "samples" data provided in the supplementary showed different single spot age ranges (overall 1353.6 to 1876.7 Ma) compared to fig 8(A) and in the text when I filtered by "Altered" and computed the quartiles based on this - did I miss a step in how you chose the data subset?The low Sr/Sr - Firstly, the explanation should come prior to the discussion of their values, I went through a lot of the paper thinking these values are exceptionally low and not feasible for well defined Sr concentrations.
I find it difficult to reconcile these values with well‑defined Sr concentrations, and cannot exclude the possibility of an analytical artefact (or artificially induced isochron rotation due to data processing), partly because the supplementary data I have does not contain the La Posta data, and I know (https://doi.org/10.5194/gchron-6-21-2024) using MicaMg increases the overall uncertainty of samples (unnecessarily) and incurs its own matrix offset from micas. I think there has been some misinterpretation of Dodson's paper in regard to the emphasis on radiogenic Sr - yes it controls the Rb–Sr geochronometer, but this is true of any radiogenic parent-child pair. My understanding of the statements in Dodson is they are more generally talking about the mobility of child nuclides (and those elements, e.g. Sr relative to Rb). Regardless of this, the observations the statements are made on (the Alps study) have since been shown to be a conflation of true volume diffusion (i.e. thermal closure) and hydrothermal alteration (e.g. https://doi.org/10.1046/j.1365-3121.1998.00156.x, https://doi.org/10.1144/jgs2021-098), and we still don't really have a good handle on Sr diffusivity in mica (very few studies, only one may not have conflated fluid-mediated influence with pure thermal diffusivity). In any case, IF the assumption that 87Sr is moved preferentially (I haven't got an issue with this, 87Sr that is a product of 87Rb decay will generally occupy the I site in a mica, and will be more readily moved out of that site) it still would not explain how you get to such low initial values. "Common" Sr is likely to be in the M-site (octahedral) rather than I-site. For Sr in the M-site, I see no reason for 87Sr to be preferentially removed in place of 86Sr or 88Sr, both of which should be more abundant (relative to "common" 87Sr) anyway. Removing Sr from the I-site (likely to be radiogenic 87Sr generated in the mica) should then bring the predicted initial 87/86Sr back to a feasible (i.e. chondritic or higher) value, unless the true initial Sr concentrations were close to zero (highly radiogenic, Sr poor micas (e.g. polylithionites) can show this effect). Most of your arguments suggests that alteration could slightly increase 87/86Sr in the presence of open system behaviour, and this is generally what is observed throughout the literature.
With that said, I still think your overall interpretations and conclusions are valid and supported by the data available, and your explanation of how you handle these anomalous values is reasonable. I am just hesitant on the absolute ages derived from your data. Fixing the intercept to 0.71 +/- 0.03 for your "preserved metamorphic micas" and recalculating the age takes it from ~ 1782 Ma down to 1737 Ma (using a Mahon type regression) and does not increase the reduced chi-squared value to unacceptable values.# Minor corrections
Terminology - as per Zack & Gilbert (2024) (https://doi.org/10.1016/B978-0-443-18803-9.00014-6), I strongly encourage the use of LA-ICP-MS/MS (yes the Agilent machine is named QQQ-MS, but it is generally configured as a QOQ-MS) to cover all varieties of ICP-MS with controlled mass shifting capabilities.Consistency of Ma/Ga - I'd understand switching to Ga use if you were generally talking about < 1000 Ma and then a much larger value above this, but Ga / Ma are used inconsistently throughout. I'd stick with Ma as the uncertainties on Rb–Sr dates are usually around 1–5% and ± 20 Ma is easier to read than ± 0.02 Ga.
Stylistic, but ranges or from-to should use an en-dash, not hyphen, e.g. Rb-Sr should be Rb–Sr, 2050-1860 Ma should be 2050–1860 Ma...
Units: while bar is a metric unit, Pa (MPa) is the SI unit and should be used preferentially. I understand bar may have been used for historic reasons, but in the SI system and other scientific fields its continued use is discouraged. All other units used are SI units anyway.
L530 - The first part of this sentence should be in the methods (I believe you may have made this change already) and "using and applied" is probably meant to be "using an applied". If this is a true "matrix correction factor" not just a mass bias correction, I'd be very careful about using MicaMg for micas in this way as MicaMg is not crystallographically similar to a mica, and as per previous, incurs its own offset. It would be much simpler to calculate these concentrations using NIST SRM 610 and an internal standard as per Iolite, LADR, literature etc.
L620 - "Some of these elements remain in and are ... whereas the rest leave ..." (suggested change for grammar)
# Wish list
As per the guidance on Geochronology's author guide, figures should endeavour to use colour schemes that are colour vision deficiency friendly. This should be best practice for authorship in the modern era, as free tools are readily available for the generation and use of CVD-safe colour schemes which also enhance the accuracy of perception/interpretation.Figure 1 this may be more difficult as TMI/gravity images sourced from government surveys tend to be rainbow coloured, but it isn't impossible. The use of red lines to annotate in panel A further decreases the accessibility of this figure for people with CVD. Also, I'm assuming the CRS is WGS84 - please put the correct CRS in the caption or on the maps - without a CRS, coordinates can be impossible to accurately reproduce.
Figure 3 - being true colour polarising microscopy images the concerns re. colour schemes are not applicable here.
Figures 5 & 6 - rainbow colour schemes really aren't appropriate here for count intensity. Firstly, they are not accessible, and secondly they actively hinder easy, intuitive and accurate interpretation by the reader; what is green, yellow etc - we perceive certain colour wavelengths better than others giving a false perception of greater range. Simple grey scale images would suffice here, but if you want to have them coloured use a single colour per image and map dark to low count, light to high counts.
Citation: https://doi.org/10.5194/egusphere-2025-6469-RC2 -
AC2: 'Reply on RC2', Quentin Boulogne, 27 Apr 2026
Dear Referee,
We thank you for your careful evaluation of our manuscript and for your constructive comments. We have addressed all the technical points you raised. Please find below a point-by-point reply to your comments. Our responses are in bold.
Referee#2: Jarred Lloyd
The contribution, while somewhat dense in some sections, and needs refinement to the ordering of some content, is a comprehensive and timely contribution to the geochronology community. In general it is very well written, although I am unconvinced by some statements made by the authors; however, these concerns do not detract from the overall rigour of the paper nor the overall conclusions. I appreciate the attention to detail and prior literature. This assessment is based on the current version of the preprint, prior to the changes the authors have stated they've implemented in their reply to referee 1.
From my reading of the contribution, the aim is to investigate the utility and application of the Rb–Sr decay system geochronology in micas as a time record of hydrothermal events. This is done using samples from a Proterozoic basin in Canada that is economically important for uranium and is known to have been subject to long lived hydrothermal events.
On the use of 1s/sigma level - I note that the other referee had queried this as well, and the authors have replied to this. I would prefer to see these uncertainties referred to at 2s level so they represent the uncertainty parameter at the interval in which ~ 95 % of any random sample within the distribution μ ± 2s would fall, rather than ~ 68 %. I take the point that the GMM (and all statistical tests for that matter) require the uncertainty at 1s during computation; however, IsoplotR does this conversion so long as the appropriate input uncertainty level is noted. It can also be output as 2s or 95 % CI with the appropriate flag. Given you are also comparing Rb–Sr to existing geochronology from other studies that used different decay systems, it is appropriate to use 2s plus the decay constant uncertainty for comparison. You also do quote 2s uncertainties for other data in the paper.
Thank you for this comment. We agree that reporting uncertainties at the 2σ level is more appropriate for comparison with geochronological data from other studies and more informative in terms of confidence intervals. We have revised the manuscript accordingly and converted all uncertainties to 2σ throughout the text. This conversion was performed directly from IsoplotR outputs using the appropriate 2σ flag and does not affect the underlying statistical treatment of the data, which was performed at the 1σ level as required by the GMM algorithm.
On the 5 populations and BIC. IsoplotR's documentation notes that increasing sample size often causes the implemented algorithm to overestimate the number of populations as sample size increases, "This option should be used with caution, as the number of peaks steadily rises with sample size (n). If one is mainly interested in the youngest age component, then it is more productive to use an alternative parameterisation, in which all grains are assumed to come from one of two components, whereby the first component is a single discrete age peak (e^m, say) and the second component is a continuous distribution (as described by the central age model), but truncated at this discrete value."
The BIC itself probably shouldn't be used here as some of its assumptions break down in this particular type of modelling. A much more robust measure like the Bayes Factor (although computationally heavy) should probably be used here; however, this particular aspect is beyond the scope of your manuscript.
Instead, you should visually assess your distributions, taking note of the peakfit algorithm's comment quoted here. I'd interpret your data (notably in fig 8), as log-normal distributions of data, or as normal distributions with continuous tail, so at most 2 distributions, not 5.Thank you for this constructive comment.
We fully acknowledge the limitations of the BIC-based approach for selecting the number of Gaussian components in this context, and we agree that more robust criteria such as the Bayes Factor would be preferable in principle, although this is beyond the scope of the present manuscript.
We also acknowledge the warning in IsoplotR's documentation regarding the tendency of the peakfit algorithm to overestimate the number of populations with increasing sample size. This is a valid concern that we have taken seriously.
However, we wish to clarify several important points:
First, the five-component GMM was not used as the sole basis for identifying discrete geological events. As explicitly stated in the manuscript: 'the components should be regarded primarily as probabilistic centers, which we attempt to compare with published datasets from the literature.' We therefore do not claim that five discrete hydrothermal episodes are unambiguously resolved by the data alone.
Second, the five GMM components and radial plot populations are not interpreted independently, their geological significance is assessed through their correspondence with independently dated tectono-hydrothermal events documented in the regional literature (Fig. 10). Each age population is discussed in the context of existing geochronological constraints from U–Pb, K–Ar, and Ar–Ar chronometers applied to minerals from the same geological setting. It is this multi-system convergence, rather than the statistical decomposition alone, that underpins our geological interpretations.
Third, the KDE distributions shown in the revised Figure 8A are consistent with log-normal or asymmetric normal distributions with continuous tails. The dominant population at ca. 1640 Ma is robust and clearly identifiable regardless of the number of components assumed. The subordinate populations at ca. 1680, 1600, 1550, and 1467 Ma are acknowledged to carry greater uncertainty and are explicitly discussed as potentially reflecting either discrete hydrothermal pulses or artificial segmentation of an otherwise continuous distribution.
In response to this comment, we have added the following clarification to the manuscript: “It should be noted that BIC-based GMM algorithms tend to overestimate the number of populations with increasing sample size (Vermeesch, 2018), and that the present data are equally consistent with a smaller number of broader, partially overlapping distributions. The age components identified here are therefore not interpreted as discrete geological events but as probabilistic centers whose significance is evaluated through their correspondence with independently dated tectono-hydrothermal events documented in the regional record.”
The uncertainty quartiles - why did you choose to plot these as histograms rather than density plots? Density plots aren't as sensitive to bin-width for visualisation (and you may as well fit distributions to them giving you mean and std). It looks like Q1 and Q2 have peaks (c. 1620 and 1645 Ma respectively) that correspond to the bimodal distribution of Q3, and then Q4 has a bimodal distribution with the lower peak on Q2 peak, and an additional peak at ~ 1800 Ma.
Thank you for this suggestion. We agree that kernel density estimates (KDE) are more appropriate than histograms for this type of visualization, as they are not sensitive to bin-width choice and provide a smoother representation of the underlying distributions. We have replaced the histograms in Figure 8A with KDE curves. The KDE plots by uncertainty quartile (revised Figure 8A) show that all four quartiles yield near-identical unimodal distributions centered at ca. 1630–1637 Ma, with no significant difference in peak position or shape between low- and high-uncertainty spot analyses. This indicates that analytical uncertainty does not introduce a systematic age bias in our dataset. We note that the bimodal structure and the ~1800 Ma peak previously observed by the reviewer in the supplementary data were artefacts of the data inconsistency identified and corrected in the following point.
I'll note here, the "samples" data provided in the supplementary showed different single spot age ranges (overall 1353.6 to 1876.7 Ma) compared to fig 8(A) and in the text when I filtered by "Altered" and computed the quartiles based on this - did I miss a step in how you chose the data subset?
Thank you for catching this discrepancy. Upon careful review, we identified an inconsistency between the age values reported in the supplementary material and those used to generate Figure 8(A), which arose during the preparation of the supplementary file. The supplementary data have now been corrected to accurately reflect the dataset subset used in the analysis, and all reported age ranges are consistent throughout the manuscript. We apologize for any confusion this may have caused.
The low Sr/Sr - Firstly, the explanation should come prior to the discussion of their values, I went through a lot of the paper thinking these values are exceptionally low and not feasible for well defined Sr concentrations.
I find it difficult to reconcile these values with well‑defined Sr concentrations, and cannot exclude the possibility of an analytical artefact (or artificially induced isochron rotation due to data processing), partly because the supplementary data I have does not contain the La Posta data, and I know (https://doi.org/10.5194/gchron-6-21-2024) using MicaMg increases the overall uncertainty of samples (unnecessarily) and incurs its own matrix offset from micas. I think there has been some misinterpretation of Dodson's paper in regard to the emphasis on radiogenic Sr - yes it controls the Rb–Sr geochronometer, but this is true of any radiogenic parent-child pair. My understanding of the statements in Dodson is they are more generally talking about the mobility of child nuclides (and those elements, e.g. Sr relative to Rb). Regardless of this, the observations the statements are made on (the Alps study) have since been shown to be a conflation of true volume diffusion (i.e. thermal closure) and hydrothermal alteration (e.g. https://doi.org/10.1046/j.1365-3121.1998.00156.x, https://doi.org/10.1144/jgs2021-098), and we still don't really have a good handle on Sr diffusivity in mica (very few studies, only one may not have conflated fluid-mediated influence with pure thermal diffusivity). In any case, IF the assumption that 87Sr is moved preferentially (I haven't got an issue with this, 87Sr that is a product of 87Rb decay will generally occupy the I site in a mica, and will be more readily moved out of that site) it still would not explain how you get to such low initial values. "Common" Sr is likely to be in the M-site (octahedral) rather than I-site. For Sr in the M-site, I see no reason for 87Sr to be preferentially removed in place of 86Sr or 88Sr, both of which should be more abundant (relative to "common" 87Sr) anyway. Removing Sr from the I-site (likely to be radiogenic 87Sr generated in the mica) should then bring the predicted initial 87/86Sr back to a feasible (i.e. chondritic or higher) value, unless the true initial Sr concentrations were close to zero (highly radiogenic, Sr poor micas (e.g. polylithionites) can show this effect). Most of your arguments suggests that alteration could slightly increase 87/86Sr in the presence of open system behaviour, and this is generally what is observed throughout the literature.We thank you for these detailed and constructive comments, which we address point by point below.
- Anomalously low intercept values
We agree that the anomalously low 87Sr/86Sr intercept values (~0.67) obtained from preserved muscovite regressions are physically impossible as true initial isotopic compositions, as they fall below the minimum value of the solar system (~0.698; Papanastassiou and Wasserburg, 1969). As stated in the manuscript, these intercepts are explicitly not interpreted as meaningful initial 87Sr/86Sr ratios, but rather as mathematical artefacts arising from open-system Rb–Sr behavior and isochron rotation under non-conservative redistribution of Rb and Sr, consistent with the framework discussed by Brooks et al. (1976), Villa (1998), and Faure and Mensing (2013). We acknowledge that the explanation of this interpretation should appear earlier in the manuscript, prior to the discussion of the intercept values themselves, and we have restructured the relevant section accordingly.
- Use of MicaMg as primary reference material
We acknowledge the limitations associated with the use of MicaMg as a primary reference material, including the matrix offset relative to natural micas and the additional uncertainty it introduces, as documented by the study cited. However, we wish to emphasize that MicaMg currently represents the only widely available and internationally distributed nano-powder reference material specifically designed and validated for in situ Rb–Sr dating by LA-ICP-MS/MS. In the absence of a commercially available mica reference material with better matrix-matching properties and similarly well-constrained Rb–Sr isotopic composition, its use remains unavoidable. We have added a statement in the methods section acknowledging this limitation explicitly.
- Interpretation of Dodson (1973)
We thank you for this important clarification. Upon re-reading Dodson (1973), we agree that his statements on the control of the geochronometer by radiogenic nuclides are made in a general sense applicable to any parent-daughter isotopic system, and are not specifically directed at the mobility of 87Sr relative to other Sr isotopes. We have revised the relevant sentence in the manuscript to avoid any misinterpretation of Dodson's original work.
- Crystallochemical argument regarding site occupancy
We thank you for this particularly insightful comment, which we find scientifically compelling. You are correct that radiogenic 87Sr produced in situ by decay of 87Rb occupies the interlayer (I-site) in mica, substituting for monovalent cations such as K+ and Rb+, whereas common Sr (86Sr, 88Sr) is more likely to be hosted in the octahedral (M-site), where divalent cations such as Ca2+ are stable. As you rightly point out, preferential removal of radiogenic 87Sr from the I-site should, in principle, drive the regression intercept back toward chondritic or higher values rather than toward the sub-chondritic values we observe. To reach intercepts as low as ~0.67, the muscovites would need to have had near-zero common Sr concentrations from the outset, which is not the case here.
We therefore agree that the sub-chondritic intercepts are better explained as a consequence of isochron rotation under open-system conditions involving non-conservative, grain-scale redistribution of both Rb and Sr, a process well-documented in the literature (Brooks et al., 1976; Villa, 1998; Eberlei et al., 2015; Glodny and Grauert, 2009) — rather than as a direct result of selective 87Sr loss from the I-site alone. We have revised the discussion accordingly to reflect this more nuanced interpretation, removing the over-reliance on the Dodson framework and replacing it with a clearer discussion of open-system isochron rotation as the primary mechanism responsible for the anomalous intercept values.
With that said, I still think your overall interpretations and conclusions are valid and supported by the data available, and your explanation of how you handle these anomalous values is reasonable. I am just hesitant on the absolute ages derived from your data. Fixing the intercept to 0.71 +/- 0.03 for your "preserved metamorphic micas" and recalculating the age takes it from ~ 1782 Ma down to 1737 Ma (using a Mahon type regression) and does not increase the reduced chi-squared value to unacceptable values.
We thank you for this thoughtful comment and for performing an independent age calculation. We would like to address several methodological and geological points that we believe are critical for interpreting this result.
- Regression method
Our isochron ages were calculated using the York regression as implemented in IsoplotR (Vermeesch, 2018), which is the current community standard for Rb–Sr geochronology and accounts for uncertainties in both 87Rb/86Sr and 87Sr/86Sr ratios as well as their error correlations. The Mahon-type regression differs methodologically from the York regression, and direct comparison of ages derived from these two approaches requires caution, as they may yield systematically different results depending on the data structure and the weighting of individual data points.
- Anchoring the intercept
We respectfully argue that anchoring the regression intercept to a fixed value of 0.71 ± 0.03 is not appropriate in our geological context. As clearly demonstrated by Rösel and Zack (2022), fixing the initial 87Sr/86Sr ratio converts an isochron regression into an anchored single-spot-type calculation. This is fundamentally different from a free isochron, where both the age and the intercept are simultaneously and independently constrained by the data themselves. Rösel and Zack (2022) explicitly show that anchoring the intercept introduces an age bias that scales with the fraction of common 87Sr and with the deviation between the assumed and the true initial 87Sr/86Sr ratio. The larger this deviation, the larger the age bias introduced.
- The choice of initial 87Sr/86Sr in a Paleoproterozoic polymetamorphic basement
This point is of particular importance in our study. The rocks of the Wollaston-Mudjatik Transition Zone are Archean to Paleoproterozoic crustal lithologies that have experienced multiple tectono-metamorphic events between ca. 1840 and 1720 Ma, as well as subsequent hydrothermal fluid circulations. In such a geological context, the initial 87Sr/86Sr ratio at the time of muscovite crystallization (ca. 1780 Ma) cannot be assumed a priori to be 0.71. Crustal rocks with long and complex pre-histories of Rb/Sr fractionation commonly develop elevated initial 87Sr/86Sr ratios well above mantle or primitive crustal values. As discussed by Rösel and Zack (2022), for crustal rocks the appropriate model range for the initial 87Sr/86Sr is 0.730 ± 0.030, reflecting the inherently large variability in evolved crustal sources. Imposing a fixed value of 0.71 without independent constraint from a co-genetic low-Rb phase would therefore likely underestimate the true initial 87Sr/86Sr ratio, resulting in a systematic underestimation of the calculated age, which is precisely what the independent recalculation appears to demonstrate.
- Geological consistency of the obtained age
Our isochron age of 1780.8 ± 32 Ma (2σ, combined internal and external uncertainty) is fully consistent with independent geochronological constraints from U–Pb dating of monazite and zircon from the same basement terranes, which constrain the M2–D2 metamorphic event to ca. 1813–1770 Ma (Annesley et al., 1992, 1997, 1999; Jeanneret et al., 2017; Toma et al., 2024), thereby confirming the geological significance of our Rb–Sr age.
For all of the above reasons, we maintain that the free isochron approach using the York regression in IsoplotR, without anchoring the intercept to an assumed initial 87Sr/86Sr composition, is the most appropriate and unbiased method for our dataset, and that our reported age of 1780.8 ± 32 Ma (2σ, combined internal and external uncertainty) is robust and geologically meaningful.
# Minor corrections
Terminology - as per Zack & Gilbert (2024) (https://doi.org/10.1016/B978-0-443-18803-9.00014-6), I strongly encourage the use of LA-ICP-MS/MS (yes the Agilent machine is named QQQ-MS, but it is generally configured as a QOQ-MS) to cover all varieties of ICP-MS with controlled mass shifting capabilities.
Thank you for this comment. We have updated the terminology throughout the manuscript from QQQ-MS to LA-ICP-MS/MS, in line with Zack & Gilbert (2024).
Consistency of Ma/Ga - I'd understand switching to Ga use if you were generally talking about < 1000 Ma and then a much larger value above this, but Ga / Ma are used inconsistently throughout. I'd stick with Ma as the uncertainties on Rb–Sr dates are usually around 1–5% and ± 20 Ma is easier to read than ± 0.02 Ga.
Thank you for this comment. We have revised the manuscript accordingly and converted all age values to Ma throughout, ensuring full consistency in the use of time units.
Stylistic, but ranges or from-to should use an en-dash, not hyphen, e.g. Rb–Sr should be Rb–Sr, 2050-1860 Ma should be 2050–1860 Ma...
Thank you for this comment. We have carefully revised the manuscript and replaced hyphens with en-dashes throughout for all ranges and intervals (temperatures, pressures, ages, isotopic systems, etc.), in line with standard typographic conventions.
Units: while bar is a metric unit, Pa (MPa) is the SI unit and should be used preferentially. I understand bar may have been used for historic reasons, but in the SI system and other scientific fields its continued use is discouraged. All other units used are SI units anyway.
Thank you for this comment. We have updated all pressure units throughout the manuscript to SI units: bar and kbar values have been converted to Pa and GPa respectively, in line with SI conventions.
L530 - The first part of this sentence should be in the methods (I believe you may have made this change already) and "using and applied" is probably meant to be "using an applied". If this is a true "matrix correction factor" not just a mass bias correction, I'd be very careful about using MicaMg for micas in this way as MicaMg is not crystallographically similar to a mica, and as per previous, incurs its own offset. It would be much simpler to calculate these concentrations using NIST SRM 610 and an internal standard as per Iolite, LADR, literature etc.
Thank you for this comment.
The first part of this sentence has been moved to the methods section (section 3.2.4), as suggested.
The typographic error 'using and applied' has been corrected to 'using an applied' in the revised manuscript.
Regarding the matrix correction factor, we acknowledge that the use of MicaMg for calculating elemental concentrations introduces an approximation, as MicaMg is a nano-powder reference material and is not crystallographically identical to natural muscovite. However, this correction factor was applied consistently across all samples and reference materials, and its influence on the calculated concentrations is proportional and reproducible. The long-term reproducibility of La Posta biotite over three years of analytical sessions, yielding ages and intercepts in excellent agreement with published ID-TIMS values, independently confirms that no significant systematic bias is introduced by this approach. We note that the concentration data are used exclusively for illustrative purposes in this study and do not form the basis of any geochronological interpretation, which relies solely on the measured isotopic ratios.
L620 - "Some of these elements remain in and are ... whereas the rest leave ..." (suggested change for grammar)
L650-652 Thank you for this suggestion. The sentence has been corrected accordingly: "Some of these elements remain in the system and are in situ transferred to the newly-formed clays, whereas the rest leave the system."
# Wish list
As per the guidance on Geochronology's author guide, figures should endeavour to use colour schemes that are colour vision deficiency friendly. This should be best practice for authorship in the modern era, as free tools are readily available for the generation and use of CVD-safe colour schemes which also enhance the accuracy of perception/interpretation.Figure 1 this may be more difficult as TMI/gravity images sourced from government surveys tend to be rainbow coloured, but it isn't impossible. The use of red lines to annotate in panel A further decreases the accessibility of this figure for people with CVD. Also, I'm assuming the CRS is WGS84 - please put the correct CRS in the caption or on the maps - without a CRS, coordinates can be impossible to accurately reproduce.
Thank you for this helpful comment. We agree that the original use of red lines and the lack of coordinate reference information reduced the accessibility and reproducibility of the figure. We have revised Figure 1 accordingly: the graphitic conductors are now displayed in magenta to improve accessibility for readers with color vision deficiency, and the figure caption now explicitly states that the map is projected in WGS 84. In addition, we have updated the color palette of the RMI images to a Viridis-based scheme to improve overall accessibility and ensure better perceptual uniformity.
Figure 3 - being true colour polarising microscopy images the concerns re. colour schemes are not applicable here.
Figures 5 & 6 - rainbow colour schemes really aren't appropriate here for count intensity. Firstly, they are not accessible, and secondly they actively hinder easy, intuitive and accurate interpretation by the reader; what is green, yellow etc - we perceive certain colour wavelengths better than others giving a false perception of greater range. Simple grey scale images would suffice here, but if you want to have them coloured use a single colour per image and map dark to low count, light to high counts.
Thank you for this helpful and important comment regarding the color schemes used in Figures 5 and 6.
We agree that the original rainbow color scale was not optimal in terms of accessibility and could hinder accurate visual interpretation of count intensity. In response, we have revised both figures accordingly. The updated versions now use a more appropriate color mapping that improves readability and ensures a clearer, more intuitive representation of the data.
Specifically, we have replaced the previous rainbow scheme with a perceptually consistent scale, mapping lower counts to darker tones and higher counts to lighter tones, as suggested. This change enhances both accessibility and interpretability for the reader.
We appreciate your suggestion, which has significantly improved the quality of the figures.
We believe that these revisions have significantly strengthened the manuscript, both scientifically and linguistically, and that it now meets the standards required for consideration in Geochronology. We sincerely thank you again for your insightful comments and for giving us the opportunity to resubmit an improved version of our work.
Yours sincerely,
Quentin Boulogne and co-authors
Citation: https://doi.org/10.5194/egusphere-2025-6469-AC2
-
AC2: 'Reply on RC2', Quentin Boulogne, 27 Apr 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 351 | 311 | 41 | 703 | 39 | 29 | 42 |
- HTML: 351
- PDF: 311
- XML: 41
- Total: 703
- Supplement: 39
- BibTeX: 29
- EndNote: 42
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
There is a lot to unpack in this contribution and I commend the authors on a comprehensive synthesis of this timely topic. The writing is excellent if somewhat dense in spots. I appreciate the attention to previous literature. The design of the project is thoughtful, and the data is collected using highly suitable methods. Incorporating textures and compositional data to support in situ Rb-Sr is essential and the authors do an overall excellent job of contextualizing the isotopic data. This is absolutely required in a setting like the Athabasca Basin where prolonged high-T to low-T hydrothermal activity has created a wide range of alteration assemblages during basin formation and readjustment. Integrating the in situ Rb-Sr data with models for U deposition is very novel indeed. Also particularly noteworthy is the suggestion, supported by the data, that partly altered white micas underwent preferential liberation of 87Sr* relative to non-radiogenic Sr. This could be a excellent case to follow-up using atom probe measurements to examine the distribution of Rb vs. 87Sr* and 86Sr in the mica lattice.
The only suggestion I can make is that the order of presentation of some important text could be improved. For example we read through all of the results and note that the initial Sr intercepts are impossibly low. I was questioning to myself if this was an analytical bias. It is only in the discussion that the robustness of La Posta secondary material is described which allays some fears of analytical bias. So perhaps it would be good to state more clearly the reproducibility of La Posta age AND initial Sr intercept at the end of the methods section.
There are a few other minor things I’ve highlighted below. Notwithstanding these the manuscript is in very good shape and illuminates some very important processes affecting a world-class U camp.
Line 154: switching here from xxxx Ma to x.xx Ga.
Line 167: please consider italicizing P and T here and elsewhere
Line 336-340: La Posta is listed as a secondary standard. But it is not obvious from the text how well it was reproduced based on NIST610/MicaMg-NP external calibration? I also note that no additional error was added to the final isochron ages and they are quoted as 1σ starting on line 456. This is a bit unconventional, with most studies using in situ Rb-Sr citing 2SE and including additional error reflecting the long-term reproducibility of secondary standards. Is the 1σ required to use the Gaussian Mixing Model? (this looks a lot like Isoplot ‘unmix age’ routine). What happens to this unmixing exercise if you use 2SE rather than 1σ? This gets back to La Posta results. What was the measured error on La Posta in this study? We are only given the Zack & Hogmalm 2016 reported value.
Line 499: Here and elsewhere in Results section I’m surprised to see some methodology details related to concentration calculations. Methods for calculating concentrations should be in an earlier section. If NIST610 was analyzed it would be a simple matter using the Trace Elements DRS in Iolite4.
Line 523: Is 1634.7 ± 1.7 Ma - an error of 0.1% - realistic? Again, how well as the secondary standard measured?
Line 645: “…the mobility of radiogenic 87Sr* is significantly greater than that of Rb and non-radiogenic Sr…” The argument is that 87Sr* is preferentially located in interlayer site which is where alteration processes might start. Preferential removal of 87Sr* would, therefore, lower the 87Sr/86sr to values unsupported on earth. But the least-altered muscovite in basement gneiss also gives initial of 0.6709 ± 0.0115. This doesn’t make sense. The authors claim to have ruled out instrumental effects but is there some cryptic matrix-mismatch adversely affecting corrected 87Sr/86Sr? Some statement about the anomalously low initial in least-altered muscovite would be warranted (unless I missed it somewhere in the text)
Line 697: I see here that the authors explain that age and initial intercept for La Posta were within accepted range and that anomalously low 87Sr/86Sr are not analytical artifacts. It would be nice to know this before launching into the results section where these anomalously low 87Sr/86Sr raise flags about this possibility.
Line 723: typo in isochrone
Line 728: back to x.xx Ga rather than xxxx Ma ages…