Quantifying soil accumulation of atmospheric mercury using fallout radionuclide chronometry

Soils are a principal global reservoir of mercury (Hg), a neurotoxic pollutant that is accumulating through anthropogenic emissions to the atmosphere and subsequent deposition to terrestrial ecosystems. The fate of Hg in global soils remains uncertain, however, particularly to what degree Hg is re-emitted back to the atmosphere as gaseous elemental mercury (GEM). Here we use fallout radionuclide (FRN) chronometry to directly measure Hg accumulation rates in soils. By comparing these rates with measured atmospheric fluxes in a mass balance approach, we show that representative Arctic, boreal, temperate, and tropical soils are quantitatively efficient at retaining anthropogenic Hg. Potential for significant GEM re-emission appears limited to a minority of coniferous soils, calling into question global models that assume strong re-emission of legacy Hg from soils. FRN chronometry poses a powerful tool to reconstruct terrestrial Hg accumulation across larger spatial scales than previously possible, while offering insights into the susceptibility of Hg mobilization from different soil environments.


Specific comments on text:
-In the Introducfion, the authors refer to the 228Th/228Ra chronometer, and the reader might assume that this was applied in the present study.However, these isotopes do not appear to have been measured as part of this study (no data for 228Th or 228Ra appear in Table S1).For clarity up front, the authors should indicate which measurements were made in this study (e.g.somewhere in the paragraph beginning at line 97).
-Table S1 includes measurements of the 137Cs inventory in the soil profiles, but these data are not discussed in the manuscript.It would be worth menfioning 137Cs in the Introducfion (paragraph beginning at line 97), and also worth including a brief sentence or two discussion about what is learned from the 137Cs inventory data somewhere later in the text or in the SI; if not, perhaps these data should be deleted from Table S1?
-The capfion to Figure 3 needs some edifing.The two panels are not labeled "a" and "b", even though there is reference to (b) in the capfion.Also, in the right side of panel (b), there are different colored circles that indicate Students' t-test results of each pair, but it is unclear what the colors indicate or which pairs are being compared.A bit more explanafion of this in the capfion would be helpful for the reader to understand this part of the figure .-In the last sentence of the Figure 4 capfion, "throughfall (LF)" should be corrected to "throughfall (TF)".
Reviewer #2 (Remarks to the Author): The manuscript enfitled, "Quanfifying soil accumulafion of atmospheric mercury using fallout radionuclide chronometry" by Landis et al. uses a novel soil chronometer to esfimate soil accumulafion rates for soil of varying age.The complexity of soil processes and development have made elucidafing terrestrial mercury cycling difficult to tease apart, a major problem given the importance of soil within the broader mercury cycle.Thus, the new chronometer holds great promise for the soil community broadly and the mercury community, specifically.The study specifically looks at how mercury accumulafion rates very across lafitude, ecosystem type, soil horizon, and soil age.Several interesfing pafterns are documented through this approach, exemplifying the ufility of the chronometer.Perhaps the two most salient findings are the dafing of peak mercury accumulafion to the 1950-1990 period and the implicafions from the mass-balance, which suggest relafively liftle soil mercury is lost to the atmosphere.It is the lafter point which gives me pause, considering some of the possible limitafions of the approach used here (expanded on below).Overall, I found the manuscript well wriften, fimely, impacfful, and generally quite interesfing.Major Comments: 1.The LRC model produces an esfimate of average SOC age and the authors have provided convincing evidence of its ability to do so.Although this approach has advantages over other methods for soil dafing (e.g., the authors point out limitafions with 14C), I would think there are sfill many of the same problems with connecfing soil mercury processes with SOC age.In fact, the authors do an excellent job of highlighfing some of these in their accompanying paper.A) any Hg process independent of SOC could bias results.For instance, if the mineralizafion of older SOM was accompanied by a release of mercury and that mercury was subsequently bound by other (perhaps newer) SOM, the age model would no longer accurately reflect the deposifion age of mercury.B) the esfimated age of the soil is really the average of a mixture of both young and new carbon and thus could be young and new mercury (which the authors point out in their companion paper, See Figure 2).C) A number of physical processes may affect the movement of soil SOC and thus affect reservoir masses (and therefore accumulafion rates).The accompanying paper presents several examples (lateral flow, verfical flow, etc.).I think that addressing some of these limitafions is necessary.Alternafively, highlighfing that the other paper discusses them at length would be helpful.While I do not think they are parficularly problemafic for most of the data interpretafion, I think they become an issue for the mass balance approach.The above three points could all result in biases of the actual accumulafion at any given point in fime.In fact, the isotopic mass balance in the accompanying paper makes this point quite clearly and, notably, that is with a single pulse (i.e., no previous deposifion).Here, the mass balance reflects mercury deposifion since soil formafion.
In summary, I think the uncertainty in the mass balance should at the very least be commented on, given the importance of the finding.Perhaps you can report the ELA mass balance findings and some of the possible sources of variability.I think this point is especially important as readers may not be working through both documents concurrently.2. In the boftom row of figure 4, the peak accumulafion rates all roughly occur at the Oa/mineral interface, but the peak year gets younger as lafitude decreases (~1950 for arcfic, ~1970 for temperate, and ~2000 for tropical).It seems like these pafterns could also be explained by the different turnover fimes of SOM, with the largest amount of mercury associated with the largest, slowest horizon (i.e., the Oa).Do the authors have thoughts on this?General Comments: I think the abstract art looks very nice.I menfion this under figure 1, but are the concentrafion values presented in text and in Figure 1 aluminum normalized/atm?59 -61: It might be nice to add specific citafions to the high and low esfimates, for the convenience of readers.127: Consider adding measurements of variability to horizonal averages and in the reporfing of subsequent values.191: Addifionally, could the lower accumulafion rate in lifter reflect the fact that not all deposifion is being taken up by the Oi horizon?If direct deposifion or throughfall passed largely through the Oi to, say, the Oe, then there wouldn't be missing mercury.328: Assuming that the DOF peak in the deep mineral horizon was due to increased verfical migrafion of mercury, wouldn't the carbon signature associated with it sfill be younger than the observed ~100 years?
The "missing mercury" was deposited in the last 20 years, no?I may have misunderstood.Figures Figure 1 Comments: Is mercury accumulafion on a per cm basis or are you scaling up the inventory by horizon depth prior the accumulafion calculafion?Are the mercury values in panel B total values or do they represent atmospheric values (i.e., Al normalized)?If it is the lafter, I think it would be good to somehow denote that in the figure.

Figure 3:
This plot has a "(b)" but no "(a)" and no demarcafion on the plots.Does the left-most (a) plot consist of both organic and mineral soils?Could these be disfinguished somehow?I also would be interested in seeing a plot (a) with the concentrafion values.Perhaps as an SI inclusion.If my interpretafion is correct, the right-most plot uses red for two different purposes: highlighfing the "increasing accumulafion" zone and for the different comparison circles.I would suggest using two different color schemes to improve clarity.Line edits: 16: This sentence is worded somewhat awkwardly, and I would suggest reworking it."Accumulated" alone reads odd.Addifionally, this sentence neglects on-going emissions of mercury and direct releases to soil.17: I might specify that "typical soils" refers here to "typical forested and arcfic soils", unless you feel that the forested soils here are representafive of other soil types (e.g., grasslands).The majority of US soils are not forested or arcfic soils.28: This sentence has four clauses, which make it a bit cumbersome to read and may lead to confusion.Consider reworking.65: Consider cufting "...to soils".183: Should this read "...1 to 5 years post soil collecfion" instead?470: "every" instead of "each" Reviewer #3 (Remarks to the Author):

Key Results
This study ufilizes novel FRN chronometry to compare Hg accumulafion in upland soil horizons at different lafitudes encompassing 3 disfinct ecosystems: arfic, temperate, and tropical.
The study reports the highest Hg accumulafion rates in tropical ecosystems, consistent with higher atmospheric deposifion in the tropics.Considering the small number of tropical and arcfic sites, these results need further verificafion, but they are indeed the first of their kind reported.
The study also reports increasing Hg accumulafion with soil depth, in general.The authors note that foliage flux is lower than total ecosystem flux (EF), but that Hg accumulafion in deeper soil layers exceeds EF, except in a few temperate soils dominated by conifers.Temperate soils cores make up the majority of the samples and it is noteworthy that excepfions arise here.Arcfic and tropical have n = 3-4, which is concerning.Further studies could confirm or deny these results.

Validity
The authors do an excellent job describing how their data fits with previous work in lake sediments, atmospheric deposifion, recycling of GEM, implicafions for atmospheric Hg mass balances, etc.I have some concerns about migrafion of Pb and other isotopes and how that impacts dafing of the cores that may need to be addressed.

Significance
The work described in this manuscript is highly significant.Following up on this work will lead to befter constraints on Hg mass balances in the atmosphere and a befter understanding of how ecosystems may respond to decreases in Hg deposifion.The work is certainly novel.Relying on 3 or 4 cores from the arcfic and tropical ecosystems is the primary troubling factor.However, even without the comparisons across ecosystems the results for the temperate soils are significant in the field of Hg research.

Data and methodology
The authors do an excellent job framing the findings and implicafions of the soil cores, parficularly relafive to reported atmospheric fluxes of Hg.Quesfions that I had after examining the figures were addressed clearly and impressively in the arficle.The examinafion of EF relafive to the accumulafion rates in the soil cores was quanfitafive and thorough.
The finding of some cores being low-accumulafing is difficult to explain and fies into concerns regarding the dafing of these soils.The authors offer plausible explanafions.However, if there is significant downward transport of Hg via dissolved organic mafter (DOM), which is offered as potenfially explaining higher Hg accumulafion in deeper soil horizons, how can the dafing sfill be valid?Would not the 210Pb also be transported downward and impact the dafing?Also, if the Pb and Hg (and C) did NOT behave the same with colloidal transport, wouldn't this invalidate the dafing of the Hg accumulafion?
And furthermore, if there is downward movement of Hg there would also be lateral movement.Is this lateral movement insignificant, i.e. how much dissolved organic mafter is transported out of each watershed?If DOM is being transported out there will be Hg associated with it.Is this transport negligible at the sites, especially those low-accumulafing sites?

Analyfical Approach
The study is very strong in this regard.More cores would have been preferred, but it is not very pracfical to get a large value of n for this type of field work.

Suggested improvements
The authors should address the limitafions of the dafing methods and the implicafions of DOM transport of Pb and other isotopes on the dafing methods.

Clarity and context
This manuscript was excepfionally well-wriften and clearly states the implicafions of using this new method of dafing soil cores.I really enjoyed reading this arficle and am excited by the results presented.

References Appropriate
My experfise Lake and peat coring are areas of experfise for me.I have always used 210Pb dafing and know some shortcomings of that method.The newer methods described here are not as familiar to me.

Response to Reviewers, NCOMMS-24-07443
The Reviewer comments are reproduced in green text, and our responses are in black.Unless specified otherwise, line references refer to the revised manuscript.
With respect to the Editors comments requiring aftenfion to limitafions and uncertainfies in the mass balance model, these are addressed throughout the revised manuscript as detailed below in responses to Reviewer comments.

Reviewer #1 (Remarks to the Author):
This manuscript demonstrates a novel means by which to characterize the history of atmospheric deposifion of Hg to soil through the applicafion of independent fallout radionuclide chronometers.The results obtained by this study are significant to the field of Hg geochemistry in parficular, and to the general issue of using soils as an archive of recent environmental change in the annual to decadal scale.The chronometry of Hg deposifional fluxes obtained by this method are consistent with other published records, and the manuscript cites numerous published studies to validate this consistency.The work supports the claim that the fallout radionuclide approach can be used widely to invesfigate Hg deposifion history across the globe.Although soils are complex and heterogeneous, the authors have taken a viable approach in sampling and analysis that deals well with the heterogeneity, and they used a concordance criterion (cumulafive 241Am/210Pb) to weed out disturbed soil profiles.The analyfical methodology used by the authors is sounds and sufficiently described to allow reproducfion of the results by others.The first author is one of the leading experts in the applicafion of fallout radionuclides to invesfigafions of soil accumulafion rates and the physical and chemical behavior of soils.
Overall the manuscript is well-wriften and deals well with uncertainfies in the data.The presentafion and stafisfical treatment of the data is of high quality.In this reviewer's opinion, this is an important manuscript that can be published nearly in its present form, with only minor revision in response to the points noted below.It is certainly fimely, of broad interest, and suitable for publicafion in NATURE COMMUNICATIONS.

Specific comments on text:
-In the Introducfion, the authors refer to the 228Th/228Ra chronometer, and the reader might assume that this was applied in the present study.However, these isotopes do not appear to have been measured as part of this study (no data for 228Th or 228Ra appear in Table S1).For clarity up front, the authors should indicate which measurements were made in this study (e.g.somewhere in the paragraph beginning at line 97).
Apologies for misleading.The 228 Th: 228 Ra data were reported in a previous publicafion, but for many of the same samples now reported here for Hg (the "FLH" code we use in Table S1).We feel it is important to highlight the 228 Th: 228 Ra data because they give extra confidence in the Hg fluxes we report, since the dafing was done by mulfiple concordant methods.We have added text to clarify this at line 97 and referenced at line 509.
-Table S1 includes measurements of the 137Cs inventory in the soil profiles, but these data are not discussed in the manuscript.It would be worth menfioning 137Cs in the Introducfion (paragraph beginning at line 97), and also worth including a brief sentence or two discussion about what is learned from the 137Cs inventory data somewhere later in the text or in the SI; if not, perhaps these data should be deleted from Table S1?
We believe the 137 Cs data are important to include because readers familiar with fallout radionuclide dafing in other systems (peat, sediment) will expect this corroborafing bomb-pulse marker.In soils we use 241 Am because Cs is strongly biologically cycled and thus unreliable.We have included a new reference to this effect in line 510 (Methods) where have introduced discussion of 137 Cs (Kaste et al. 2021 in our reference list).We chose not to discuss at line 97 in order to maintain readability of the Introducfion, since there we are describing more broadly the advances in FRN chronometry that have made possible the present study of Hg.
-The capfion to Figure 3 needs some edifing.The two panels are not labeled "a" and "b", even though there is reference to (b) in the capfion.Also, in the right side of panel (b), there are different colored circles that indicate Students' t-test results of each pair, but it is unclear what the colors indicate or which pairs are being compared.A bit more explanafion of this in the capfion would be helpful for the reader to understand this part of the figure.
We have labeled panels.We have changed colors of the circles for befter clarity, and now include a descripfion in the capfion.

Reviewer #2 (Remarks to the Author):
The manuscript enfitled, "Quanfifying soil accumulafion of atmospheric mercury using fallout radionuclide chronometry" by Landis et al. uses a novel soil chronometer to esfimate soil accumulafion rates for soil of varying age.The complexity of soil processes and development have made elucidafing terrestrial mercury cycling difficult to tease apart, a major problem given the importance of soil within the broader mercury cycle.Thus, the new chronometer holds great promise for the soil community broadly and the mercury community, specifically.The study specifically looks at how mercury accumulafion rates very across lafitude, ecosystem type, soil horizon, and soil age.Several interesfing pafterns are documented through this approach, exemplifying the ufility of the chronometer.Perhaps the two most salient findings are the dafing of peak mercury accumulafion to the 1950-1990 period and the implicafions from the mass-balance, which suggest relafively liftle soil mercury is lost to the atmosphere.It is the lafter point which gives me pause, considering some of the possible limitafions of the approach used here (expanded on below).Overall, I found the manuscript well wriften, fimely, impacfful, and generally quite interesfing.

Major Comments:
1.The LRC model produces an esfimate of average SOC age and the authors have provided convincing evidence of its ability to do so.Although this approach has advantages over other methods for soil dafing (e.g., the authors point out limitafions with 14C), I would think there are sfill many of the same problems with connecfing soil mercury processes with SOC age.In fact, the authors do an excellent job of highlighfing some of these in their accompanying paper.We clarified (line 118) that a crifical difference is that radionuclides accurately date exposure age of SOM whereas 14 C is heavily biased to include carbon recycling fime.We agree that there is sfill a conceptual problem of dafing SOM versus dafing the Hg itself (more below).
A) any Hg process independent of SOC could bias results.For instance, if the mineralizafion of older SOM was accompanied by a release of mercury and that mercury was subsequently bound by other (perhaps newer) SOM, the age model would no longer accurately reflect the deposifion age of mercury.
We agree and clarified (lines 487-493, 512-514, 590-598) that it is likely that Hg is somefimes decoupled and convoluted with SOM in ways not fully understood.However, while we are dafing organic mafter, the age model itself is based on (primarily) 210 Pb, and the central thesis of the manuscript becomes, whether Hg and Pb follow the same pathways during SOC decomposifion.Our central hypothesis (lines 111-115) is that the FRNs ( 210 Pb) provide a normalizing term that isolates Hg flux from various biogeochemical transformafions, and that Hg and FRNs should follow the same pathways during decomposifion, absent the influence of Hg reducfion to GEM and possible subsequent mobilizafion.Where Hg quanfitafively follows Pb, we can truly date it (this appears to be the case in most soils studied given their converging accumulafion histories).Where it does not, the model provides a way of idenfifying rates of losses (Hg mobilizafion).B) the esfimated age of the soil is really the average of a mixture of both young and new carbon and thus could be young and new mercury (which the authors point out in their companion paper, See Figure 2).We agree, this is certainly true.The FRN ages represent averages as well (same as in sediment/peat records).To the extent that Pb traces Hg behavior, the LRC age represents average age of the deposited Hg.It is only when Hg behavior diverges from the tracers that the Hg "age" is substanfially different than that of SOC to which it is bound.We added a short statement (line 650) to clarify this.
We also note that we referred to Hg deviafion from tracers due to GEM "mobilizafion", "translocafion", or "percolafion" throughout the original manuscript at lines 97, 204, 263, 280, 306, 348, 362, 369, and 385.We believe these are calling out the very processes that Reviewer has cited, although perhaps not as explicitly as possible describing how these processes relate to the age model and mass balance.We have added text to emphasize that Hg can deviate from expectafion based on FRNs, at line 417.We have also taken this opportunity to menfion that Hg has ligand preferences that may also contribute to its fracfionafion from FRNs (now cifing Jiskra et al. 2014, Chen et al. 2016, Zhang et al. 2018).
C) A number of physical processes may affect the movement of soil SOC and thus affect reservoir masses (and therefore accumulafion rates).The accompanying paper presents several examples (lateral flow, verfical flow, etc.).
I think that addressing some of these limitafions is necessary.Alternafively, highlighfing that the other paper discusses them at length would be helpful.While I do not think they are parficularly problemafic for most of the data interpretafion, I think they become an issue for the mass balance approach.The above three points could all result in biases of the actual accumulafion at any given point in fime.In fact, the isotopic mass balance in the accompanying paper makes this point quite clearly and, notably, that is with a single pulse (i.e., no previous deposifion).Here, the mass balance reflects mercury deposifion since soil formafion.
The Reviewer has done an excellent job of summarizing the conceptual challenge to interprefing soil chronometry to Hg.They key to the method is appreciafing that the LRC model, by way of dafing SOC, provides a hypothesis for behavior of Hg.This is our key statement, at line 93.We have restructured this paragraph to emphasize this as it should guide the reader through the paper.Unfortunately, our companion paper is sfill in Review and we believe best to publish the present manuscript as quickly as possible, but note that the changes we have made, listed above, will make this clear to the reader.
In summary, I think the uncertainty in the mass balance should at the very least be commented on, given the importance of the finding.Perhaps you can report the ELA mass balance findings and some of the possible sources of variability.I think this point is especially important as readers may not be working through both documents concurrently.
We agree, per comments above.We also cite references to METAALICUS at line 460.
2. In the boftom row of figure 4, the peak accumulafion rates all roughly occur at the Oa/mineral interface, but the peak year gets younger as lafitude decreases (~1950 for arcfic, ~1970 for temperate, and ~2000 for tropical).It seems like these pafterns could also be explained by the different turnover fimes of SOM, with the largest amount of mercury associated with the largest, slowest horizon (i.e., the Oa).Do the authors have thoughts on this?
It is a good observafion.We had already suggested that the SOM decay might regulate the extent to which Oi/Oe horizon retain Hg, with higher rates of mobilizafion in tropical soils.However, we added a sentence to refer to this possibility (line 422), i.e., that if the shift in apparent peak accumulafion across biomes is related to SOM dynamics and preference for Oa humic mafter, this would require that Hg migrate more quickly downward than Pb in the arcfic soils, but more slowly in the tropics.It is unclear why this should be.We also emphasize that the tropical soils all show the same trend, despite LeF3 being 80% organic mafter and the others 50% oxyhydroxide, about as different as soils can get!It is unclear why Hg would behave idenfically in these disparate tropical soils, and yet differ from temperate or arcfic soils.Substanfially more work will be required to resolve this, beyond present scope.
For the arcfic soils we suggested that the late peak is very likely due to uncertainfies in both age model and atmospheric Hg correcfion (lines 420-425) since it is very difficult to resolve both atmospheric 210 Pb and Hg from their geogenic counterparts.So, these combined, we believe more likely that Hg and Pb are sfill following similar behaviors across biomes, but that soils of each biome soil have specific challenges to the age dafing itself that requires further work, and the histories in both Arcfic and tropics require further study.
General Comments: I think the abstract art looks very nice.Thank you!I menfion this under figure 1, but are the concentrafion values presented in text and in Figure 1 aluminum normalized/atm?Good catch.They are not, which we believe is important for comparability with other studies.In our data repository we include total Hg and total atmospheric Hg.
59 -61: It might be nice to add specific citafions to the high and low esfimates, for the convenience of readers.These were added at the sentence following the esfimates but we have moved forward for clarity (Smith-Downey et al. 2020, Zhu et al. 206, Zhou et al. 2020, Zhang et al. 2023).
127: Consider adding measurements of variability to horizonal averages and in the reporfing of subsequent values.
We have omifted them since the averages are informafional to provide a frame of reference, but variance would reflect cross-ecosystem range which is less straighfforward to interpret.These are now added in a new Supporfing Table .191: Addifionally, could the lower accumulafion rate in lifter reflect the fact that not all deposifion is being taken up by the Oi horizon?If direct deposifion or throughfall passed largely through the Oi to, say, the Oe, then there wouldn't be missing mercury.Yes, we think it just so, and we clarified this in lines 430-436.This is the reason we discuss percolafion as a dominant process for translocafing Hg (line 308 and elsewhere) and described the depth distribufion of 7 Be (Supporfing Informafion, Figure S4) since this short-lived isotope (half-life 54 days) records primarily percolafion and does not survive long enough to record subsequent advecfion/diffusion.This is surprising for Hg, however, since most Hg enters soil as GEM either with lifterfall or directly to forest floor, rather than by rainfall as for 7 Be.In this sense the Hg is missing from surface Organic horizons where we should expect it based on both foliar and non-foliar GEM deposifion, but found deeper in the soil profile than we expect if GEM were to remain bound to this bulk organic mafter.328: Assuming that the DOF peak in the deep mineral horizon was due to increased verfical migrafion of mercury, wouldn't the carbon signature associated with it sfill be younger than the observed ~100 years?The "missing mercury" was deposited in the last 20 years, no?I may have misunderstood.We would not expect the carbon age to be different than expectafion, since in this case we expect that the Hg has moved deeper relafive to either the FRN tracers or bulk carbon itself that the FRNs are dafing.We believe more likely that Hg is mobilized by GEM or instead some highly mobile Hg-DOC complex, but in this case, yes, the apparent age of the Hg would be much older than reality.We have clarified at line 341 that the process would require preferenfial mobilizafion of Hg.We also caufion over-interpretafion of individual soil profiles since their local histories are difficult to constrain with certainty.

Figures
Figure 1 Comments: Is mercury accumulafion on a per cm basis or are you scaling up the inventory by horizon depth prior the accumulafion calculafion?
The figure shows Hg inventories on a cm -1 basis for panel C.For the accumulafion calculafion, the inventories [ug m -2 ] are instead normalized by the fime durafion of each cm interval, which is actually one of the most powerful uses of the chronometry since depth is an arbitrary axis with respect to atmospheric fluxes.This is what allows use to reconstruct high-resolufion histories, but it does inflate the variance of flux values we observe within an individual soil horizon due to changing fluxes through depth/fime.We have added clarificafion to the figure capfion.
Are the mercury values in panel B total values or do they represent atmospheric values (i.e., Al normalized)?If it is the lafter, I think it would be good to somehow denote that in the figure.These are not normalized to befter compare with literature data.This is now clarified in the figure and legend.

Figure 3:
This plot has a "(b)" but no "(a)" and no demarcafion on the plots.Does the left-most (a) plot consist of both organic and mineral soils?Could these be disfinguished somehow?We now use the same shape and color scheme as previous figures to show both ecotype and soil horizon.
I also would be interested in seeing a plot (a) with the concentrafion values.Perhaps as an SI inclusion.We have added this as requested.It is reassuring to see the same trends in both concentrafions and fluxes, this is due to the preponderance of atmospheric Hg residing in organic soils with low geogenic contribufions, and liftle in mineral soil where bulk density is much higher.
If my interpretafion is correct, the right-most plot uses red for two different purposes: highlighfing the "increasing accumulafion" zone and for the different comparison circles.I would suggest using two different color schemes to improve clarity.We now use the same shape and color scheme as previous figures and show both ecotype and soil horizon.
Line edits: 16: This sentence is worded somewhat awkwardly, and I would suggest reworking it."Accumulated" alone reads odd.Addifionally, this sentence neglects on-going emissions of mercury and direct releases to soil.We have reworded, thank you. 17: I might specify that "typical soils" refers here to "typical forested and arcfic soils", unless you feel that the forested soils here are representafive of other soil types (e.g., grasslands).The majority of US soils are not forested or arcfic soils.We have changed to 'forest and tundra soils'.Grasslands should be a priority for future work.
28: This sentence has four clauses, which make it a bit cumbersome to read and may lead to confusion.Consider reworking.We changed this accordingly.

Key Results
This study ufilizes novel FRN chronometry to compare Hg accumulafion in upland soil horizons at different lafitudes encompassing 3 disfinct ecosystems: arfic, temperate, and tropical.
The study reports the highest Hg accumulafion rates in tropical ecosystems, consistent with higher atmospheric deposifion in the tropics.Considering the small number of tropical and arcfic sites, these results need further verificafion, but they are indeed the first of their kind reported.
We agree that applying our method to addifional sites should be a high priority.The excellent consistency of the results at the two arcfic sites (Alaska and Greenland), and within the one tropical site (Puerto Rico) gives good confidence that the method is reliable, but we need broader spafial coverage to understand global Hg dynamics.
The study also reports increasing Hg accumulafion with soil depth, in general.The authors note that foliage flux is lower than total ecosystem flux (EF), but that Hg accumulafion in deeper soil layers exceeds EF, except in a few temperate soils dominated by conifers.Temperate soils cores make up the majority of the samples and it is noteworthy that excepfions arise here.Arcfic and tropical have n = 3-4, which is concerning.Further studies could confirm or deny these results.
We have added a comment about the low replicafion and pointed out the need for more site (lines 285-286).However, we point out that the soils within the Arcfic (both Greenland and Alaska) and tropics (only Puerto Rico) are very consistent internally.This provides a very strong lafitudinal gradient that is of great interest and should be explored further.

Validity
The authors do an excellent job describing how their data fits with previous work in lake sediments, atmospheric deposifion, recycling of GEM, implicafions for atmospheric Hg mass balances, etc.I have some concerns about migrafion of Pb and other isotopes and how that impacts dafing of the cores that may need to be addressed.The corroborafion of mulfiple age models including 5 different metals as well as 14 C gives us confidence that, to the first order and to such an extent to be an extremely useful method, we are measuring fundamental soil processes that very similarly impact a range of metals and carbon.That Hg is an outlier does in fact show that, in some fashion, it is preferenfially mobilized, but that is of course the crux of our paper... that we can use the age models to reconstruct Hg fluxes in some cases, and in others to highlight very important aspects of Hg environmental chemistry.We have added clarity to this point in responses to Reviewer #2 above.

Significance
The work described in this manuscript is highly significant.Following up on this work will lead to befter constraints on Hg mass balances in the atmosphere and a befter understanding of how ecosystems may respond to decreases in Hg deposifion.The work is certainly novel.Relying on 3 or 4 cores from the arcfic and tropical ecosystems is the primary troubling factor.However, even without the comparisons across ecosystems the results for the temperate soils are significant in the field of Hg research.
We agree that temperate soils are the focus, and the arcfic and tropical data, while limited, provide a strong gradient and important global perspecfive that at once is consistent with available deposifion data for these regions, but also highlights the need for new work in these locafions.

Data and methodology
The authors do an excellent job framing the findings and implicafions of the soil cores, parficularly relafive to reported atmospheric fluxes of Hg.Quesfions that I had after examining the figures were addressed clearly and impressively in the arficle.The examinafion of EF relafive to the accumulafion rates in the soil cores was quanfitafive and thorough.
The finding of some cores being low-accumulafing is difficult to explain and fies into concerns regarding the dafing of these soils.The authors offer plausible explanafions.However, if there is significant downward transport of Hg via dissolved organic mafter (DOM), which is offered as potenfially explaining higher Hg accumulafion in deeper soil horizons, how can the dafing sfill be valid?Would not the 210Pb also be transported downward and impact the dafing?Also, if the Pb and Hg (and C) did NOT behave the same with colloidal transport, wouldn't this invalidate the dafing of the Hg accumulafion?
The fundamental challenge to using FRN for soil chronometer is understanding what it is that we are dafing.Soils are generally not aggradafional in the sense of sediment or peat, except perhaps in deep O horizons of cold climates, so the advecfion that we measure must be interpreted differently.The model demonstrates that 210 Pb migrates through soil at rates of a one to few mm y -1 , which is an order of magnitude higher than soil producfion, so must represent some DOM or colloidal process (described more in Landis et al. 2016).However, the concordance of mulfiple chronometers including 14 C that we discuss in lines 80-90 is crifical for appreciafing that this migrafion process incorporates both parficle-reacfive metals and carbon itself.The correlafion between 210 Pb and 14 C first described by Dorr and Munich (1989) and now in a coming manuscript of ours for which data was provided in our Cover Lefter, confirm that many metals and carbon are migrafing together.Thus, the migrafion of 210 Pb does not invalidate the method, it is in fact the basis of the method, insofar as 210 Pb provides a tracer of this fundamental soil process(es).
The real quesfion is, we believe, whether Hg and Pb will behave the same way.This is our crifical hypothesis, as posed in lines 93-96: we hypothesize that Hg and Pb will behave the same, since C, Pb, Be, Ra, and Th tracers appear to behave the same insofar as their age models are concordant.If, however, there is evidence Hg does not, this becomes the point of acute interest and the secondary value of our method (the first being flux reconstrucfion).Our ability to see that in some coniferous forests Hg appears to be preferenfially mobilized is a key finding that points to a crifical need for more work in these environments.We address these same points following the recommendafions of Reviewer 2 above.
And furthermore, if there is downward movement of Hg there would also be lateral movement.Is this lateral movement insignificant, i.e. how much dissolved organic mafter is transported out of each watershed?If DOM is being transported out there will be Hg associated with it.Is this transport negligible at the sites, especially those low-accumulafing sites?
We agree that this is true.Our soils are collected in suitable reference sites where to every extent possible we try to isolate atmospheric deposifion and minimize potenfial for lateral transport by sampling at hilltops and ridgelines in locally flat locafions (now clarified in line 532 and more in Supporfing Informafion).
Nonetheless, we cannot rule it out, and this is now specified at lines 342-348 and shown in Figure S2.We also cite rates of Hg export that may be expected in associafion with DOM and how these compare with our Hg esfimates at lines 356-363.
Lateral subsurface flow is most likely the cause of high subsurface concentrafions in soils ELA302-2 (bedrock depression, confined) and HoF (irregular fill and boulder confined).But we do have some tools such as concordance of the LRC and 241 Am models and Am/Pb inventories and rafios (Figure S1), and now, expected Hg deposifion inventories, to help rule out these situafions.Truly it is extraordinary the degree to which these age models agree in dafing the bomb-pulse horizon (befter than 10%!).
Rather than a weakness of the dafing, as we have argued to Reviewer 2, this really is a key importance of it.In localifies where Hg inventories and fluxes do not conform to expectafion based on histories of Hg deposifion and regional soil inventories (smaller or larger), we can gain insight into processes that may be responsible.This is the approach that highlights the need for befter understanding of Hg mobilizafion in coniferous forests since they are likely at greater risk of Hg accumulafion in food webs (e.g., lines 355-358 and lines 367-369, 426-428).

Analyfical Approach
The study is very strong in this regard.More cores would have been preferred, but it is not very pracfical to get a large value of n for this type of field work.The Reviewer's understanding is much appreciated.The analyfical demands for soil dafing are extreme and there simply is not enough instrumentafion available (anywhere).We have, to our knowledge, the largest academic facility in the US in terms of the number of appropriate state-of-the-art gamma detectors (large, Broad-Energy style), and we sfill must count each sample for 4 days to provide reasonable uncertainfies for the FRNs.A single soil pit requires over one month of 24/7 lab fime!We are trying very hard to double our capacity, but this requires about $1.5M.Hopefully this publicafion will help sway the funding agencies!Suggested improvements