Soil minerals mediate climatic control of soil C cycling on annual to centennial timescales
Abstract. Climate and parent material both affect soil C persistence, yet the relative importance of climatic versus mineralogical controls on soil C dynamics remains unclear. To test this, we collected soil samples in 2001, 2009, and 2019 along a combined gradient of parent material (andesite, basalt, granite) and climate (mean annual temperature (MAT): 6.5 °C “cold”, 8.6 °C “cool”, 12.0 °C “warm”). We measured the radiocarbon of heterotrophically respired CO2 (∆14Crespired) and bulk soil C (∆14Cbulk) as proxies for transient and persistent soil C, and characterized mineral assemblages using selective dissolution. Using linear regression, we observed that MAT was not a significant predictor of either ∆14Cbulk or ∆14Crespired, yet climate was highly significant as a categorical variable. Climate explained more variance in ∆14Cbulk and ∆14Crespired over 0–0.1 m, but parent material explained more from 0.1–0.3 m. Cool site soil C was more persistent (lower ∆14Cbulk) than cold or warm climate sites, and also more persistent on andesitic soils, followed by basaltic and then granitic soils. Poorly crystalline metal oxides (PCMs) (but not crystalline metal oxides) were significantly (p < 0.1) correlated with ∆14Cbulk, ∆14Crespired, and ∆14Crespired - ∆14Cbulk, indicating their importance for soil C cycling on both short and long timescales. The change in ∆14Crespired observed over the study period was linearly related to MAT for the granite soils with the lowest PCM content, but not in the andesitic and basaltic soils with higher PCM content. This link between PCM abundance and the decoupling of MAT and soil C cycling rates suggests PCMs may attenuate the temperature sensitivity of decomposition.
Jeffrey Prescott Beem-Miller et al.
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2022-1083', Anonymous Referee #1, 19 Jan 2023
- RC2: 'Comment on egusphere-2022-1083', Anonymous Referee #2, 28 Jan 2023
Jeffrey Prescott Beem-Miller et al.
Jeffrey Prescott Beem-Miller et al.
Viewed (geographical distribution)
Data interpretation and integration:
There seems to be some missed opportunities with regards to the data currently in the manuscript and the data available in the literature.
I was very excited to read this manuscript, anticipating the novel use and interpretation of the time series data. Having three sampling points over a twenty-year time period across different parent materials and vegetation types *with* statistical replication (3 pits per site) is quite a unique and powerful dataset! I was disappointed to see no results from the time series featured in the abstract, and no simple modeling exercises to parse the bulk C into pools with associated mean residence times (or system ages, or transit times, etc.). The paper cites the Baisden et al. (2013) paper but does not utilize the simple spreadsheet model available in the supplementary materials of that work. It seems that at least in the surface soils, the respiration and bulk radiocarbon data could be combined to yield some insight into partitioning and turnover of C in these soils. I understand that more complex modeling exercises are beyond the scope of what the authors are trying to convey. However, given that the paper is focused on the influence of mineralogy and climate on decadally cycling C, it seems appropriate to try to explore how the size and turnover of that pool varies across the sites.
There appears to be a much larger dataset from previously published work that could be used to inform the interpretations of the current dataset. There is density fraction data from Rasmussen et al. (2018). I’m extremely curious to know how the radiocarbon value of the respired CO2 compares to the radiocarbon value of the free/light fraction. I’m also curious to know how the respiration rates of the bulk soils compared to the respiration rates of the soils with the litter manipulations (Rasmussen et al., 2008).
Figure S2 caption states that the soils were incubated from 4-40 days. Why would the soils be incubated for different time periods when the data presented is a flux measurement (CO2 respired per day)? Varying the length of the incubation would have a strong influence on the total amount of C respired and the associated radiocarbon measurement. I supposed some soils were incubated for a longer time period in order to collect enough CO2 for a radiocarbon measurement. If this is the case, the potential artifacts of this approach need to be explicitly discussed. For example, if a soil is respiring at a lower rate, it might imply that the “labile” pool is much smaller. Continuing to incubate this soil for a longer time period would potentially allow for the decomposition of less readily available C, causing the radiocarbon value to be skewed to a more depleted value.
The proposed source and interpretation of the laboratory-respired C is inconsistent in the manuscript. In the methods and results section, the pool of respired C is described as being transient and representative of “…the majority of C leaving the soil via heterotrophic respiration (Lines77-79).” The text in this section of the paper strongly implies that the radiocarbon value of the CO2 captured during incubation reflects the radiocarbon value of C being respired in the field, specifically that which is cycling on a decadal scale. However, in the discussion, when explaining depleted radiocarbon values of CO2 from subsurface granite soils, the laboratory-respired CO2 is described as being sourced from persistent stabilized pools (lines 434-438). Here, the incubation conditions are described as alleviating constraints on decomposition, allowing for decomposition of C which is stabilized under field conditions. Similarly, in the results section, some respiration data was excluded from analysis due to unexpectedly depleted values resulting from “… disturbance of the soil during sample extraction and preparation (lines 363-365).” No doubt these statements imply a high degree of complexity associated with the interpretation of incubation radiocarbon data. It may be necessary to delve into more of this complexity and the philosophical justifications behind one interpretation or another.
Short-range-order mineral abundance as an explanatory variable:
One of the main findings of this work is the dependence of soil C concentration/stock and radiocarbon abundance on the abundance of oxalate-extractable metals. The manuscript also features the lack of correlation between soil C variables and crystalline oxyhydroxide phases.
The acronyms PCM (poorly crystalline metal oxides) and CRM (crystalline metal oxides) are introduced. Because the manuscript relies so heavily on the relative abundance of these two pools of extractable metals, it is important to be specific and accurate about what they represent. Almost all of the minerals that are selectively dissolved are more accurately described as hydroxides and oxyhydroxides rather than oxides. The term PCM includes pyrophosphate-extractable pools of Fe and Al, which are amorphous (not poorly crystalline). Truly using the short-range-order pool of Fe and Al rather than the oxalate-extractable pool would necessitate the subtraction of the pyrophosphate numbers from the oxalate numbers. Similarly, the CRM value excludes Al from crystalline hydroxide phases since Al is not redox soluble. Including Al phases in one pool (PCM) while excluding them from the other (CRM) negates the true comparison of the explanatory power of these mineral pools.
In a more general sense, testing the explanatory power of extractable metals for C characteristics in these soils does not seem to be novel. See results from previous work:
From the abstract of Rasmussen et al. (2005): “We found highly significant, positive correlations between Al-humus complexes, SRO Al minerals, and total C content.”
From the abstract of Rasmussen et al. (2008): “This study corroborates the varied response in soil C mineralization by parent material and highlights how the soil mineral assemblage and litter type may interact to control conifer forest soil C response to climate change.”
From the abstract of Rasmussen et al. (2018): “Results indicated that short-range order mineral phases were the dominant factors accounting for the variation in soil carbon content and residence time.”
Climate as a categorical variable:
Using climate as a categorical variable does not appear to be appropriate. Vegetation is covarying with climate, and this strongly influences C stocks and turnover. This fact is acknowledged in Rasmussen et al. (2018) where the categorical variables are parent material and ecosystem instead of parent material and climate. This fact is further driven home by the Rasmussen et al. (2008) paper where differences in litter chemistry between vegetation types were so large that they had significant explanatory power for variance in soil respiration. It is also possible that NPP and/or above- vs. below-ground C allocation might vary between “warm”, “cool”, and “cold” sites. There’s also the issue of differences in fire regime across the different ecosystems (Rasmussen et al., 2018). Perhaps the fact that temperature had no explanatory power as a continuous variable is actually a more interesting result? What if litter chemistry and mineralogy were important and climate was not? Wouldn’t this be an interesting finding to explore?
Minor line-by-line comments:
Line 10: The results associated with the repeated sampling do not seem to be highlighted anywhere in the abstract.
Line 16: Could it be that climate was highly significant as a categorical variable because as a categorical variable it encompasses all the other variation across sites? How is using climate as a categorical variable different than using site as a categorical variable? Statistically, the tests would be identical, but the interpretation of the result would be very different.
Line 21: The acronym PCM? Has the convention changed from using "short range order" to "poorly crystalline metal" (SRO has been swapped for PCM)?
Line 38-39: “… insight into the relevant time scales associated with these key soil forming factors.” Something doesn't quite work with the meaning of this sentence. Is "time scales" referring to the fact that temperature changes much faster than the SRO mineral content of the soil?
Line 53: The sorptive capacity of minerals is dependent on many things in addition to the density of hydroxyl groups. Perhaps modify the first part of this sentence to be more general and then highlight the especially strong bonding associated with ligand exchange?
Line 65: You might specify that the bomb-derived radiocarbon is the tracer.
Line 72: Would it be appropriate to cite Graven's or Sierra's paper on the decline of the bomb curve here? I think it might be important to point your readers to additional information on this specific point, as many soil scientists are unaware of this fact.
Lines 77-79: The phrasing here might need a little clarification. Some might argue that the majority of the C leaving the soil via heterotrophic respiration under field conditions would likely have a similar radiocarbon value as that year's atmosphere (See figure 9 of Phillips et al., 2013; full reference given below). Some might additionally argue that the C respired in a laboratory incubation is not the same pool of C respired under field conditions, and that the radiocarbon value of lab-respired C is some metric of the mean age of the C that may be more available for decomposition under conditions of disturbance.
Lines 141-142: “Samples were collected from the A horizon only in 2001, and from both the A and B horizons in 2009 and 2019.” This does not seem correct according to the associated publications, or the data presented in this paper.
Lines 150-151: “We omitted the 2009 samples from the incubation experiment to save on time and analysis costs, and because sample material was only available from a single profile at each site.” Doesn't this strongly effect your ability to examine the dynamics of this "decadally cycling pool"? Also, it looks like there were three pits sampled in 2009 according to Rasmussen et al., 2018?
Line 176: Here you switch to SRO, then back to PCM below.
Line 184: “… the difference of dithionite-citrate extractable Fe and ammonium-oxalate extractable Fe as a proxy for CRM abundance.” This obviously is excluding gibbsite, which cannot be quantified with selective dissolution... wouldn't the andesite and/or granite soils contain gibbsite in some quantity? Is that important in the context of the influence of crystalline oxides on SOM?
Lines 196-197: “…values of Δ14C < 0 indicate radioactive decay of 14C, which has a half-life of 5730 years.” Samples with values <0 per mil can still have bomb C in them. Also, all radiocarbon values reflect some level of radioactive decay. Maybe just emphasize that higher values indicate faster cycling, with values above zero confirming the incorporation of significant amounts of bomb C, implying decadal cycling?
Equations 1-4: How are these random error terms constrained and/or defined?
Line 257: “… substantial variation in SOC concentration between 2001, 2009, and 2019 in the surface mineral layers.” Were the samples taken during the same month/season across the years? Were there significantly different drought conditions, el nino, atmospheric river events, fire, etc. across the years?
Lines 362-364: This seems a bit of a "throw away" explanation for these outlier values. These samples wouldn't have been any more or less disturbed during sampling than the others, would they? I think it's fine to exclude strong outliers (how else can you possible do statistics on systems with such high variance?), but it seems like a more process-based or mechanistic explanation could be found given all the ancillary data and the high level of expertise of the author team.
Lines 394-395: What is the hypothesis associated with this testing the explanatory power of crystalline Fe oxyhydroxides? Given the information in the literature, why would you expect a relationship between crystalline Fe/Al oxyhydroxides and 14C? Wouldn't you expect a relationship between pyrophosphate-extractable Fe/Al and %C and/or 14C given the pH of the soils?
Lines 470-474: This is not the only conclusion that can be reached by the data presented. Indeed, the carbon could be cycling more quickly in the p. pine ecosystems regardless of what mineral surfaces are available to provide sorptive protection, especially since only the suface soils were incubated… also these conclusions seem highly similar to the conclusions of the Rasmussen et al. (2018) paper: “These results suggest a link between the degree of soil weathering and C storage capacity, with a greater divergence in storage capacity and residence time in the Inceptisols, Entisols, and Andisols of the white fir and red fir ecosystems relative to minimal variation in the highly weathered Ultisols and Alfisols of the ponderosa pine ecosystem.”
Baisden, W.T., Parfitt, R.L., Ross, C., Schipper, L.A. and Canessa, S., 2013. Evaluating 50 years of time-series soil radiocarbon data: towards routine calculation of robust C residence times. Biogeochemistry, 112(1), pp.129-137.
Phillips, C.L., McFarlane, K.J., Risk, D. and Desai, A.R., 2013. Biological and physical influences on soil 14 CO 2 seasonal dynamics in a temperate hardwood forest. Biogeosciences, 10(12), pp.7999-8012.
Rasmussen, C., Throckmorton, H., Liles, G., Heckman, K., Meding, S. and Horwath, W.R., 2018. Controls on soil organic carbon partitioning and stabilization in the California Sierra Nevada. Soil Systems, 2(3), p.41.
Rasmussen, C., Southard, R.J. and Horwath, W.R., 2008. Litter type and soil minerals control temperate forest soil carbon response to climate change. Global Change Biology, 14(9), pp.2064-2080.
Rasmussen, C., Torn, M.S. and Southard, R.J., 2005. Mineral assemblage and aggregates control carbon dynamics in a California conifer forest. Soil Science Society of America Journal, 69(6), pp.1711-1721.