| 08 Oct 2026
Direct P K-edge X-ray absorption spectroscopy evidence of phosphorus immobilization in anoxic peat soils
Abstract. Rewetting of drained agricultural peatlands is widely promoted to mitigate greenhouse gas emissions, yet concerns remain that reductive dissolution of Fe(III) (hydr)oxides may mobilize bound phosphorus (P) and increase eutrophication risk. However, P concentrations following rewetting can vary by orders of magnitude among peatlands, and the processes controlling this variation remain unclear. In this study, we investigated the mechanisms controlling P retention in a rewetted fen (Løvenborg, Denmark) using synchrotron-based P K-edge X-ray absorption spectroscopy (XAS) and µ-X-ray fluorescence (µ-XRF), bulk soil chemistry, selective extractions, and porewater analyses. The study site is characterized by near-neutral pH (6.8–7.2), elevated calcium (Ca) concentrations (~ 0.5 mM), and substantial pools of ~200 mmol oxalate extractable Fe and Al. Fe(III) reduction reached up to 79% of the oxalate-extractable Fe pool, yet dissolved P remained below detection limits (< 0.1 µM) in all samples, demonstrating a clear decoupling between Fe(III) reduction and P release. Calculated mobilized P greatly exceeded measured porewater P, indicating rapid re-immobilization within the soil matrix. Bulk P K-edge X-ray absorption near-edge structure (XANES) spectra were similar under oxic and anoxic conditions and were dominated by organic and Fe/Al-associated P, with no clear bulk evidence of vivianite or hydroxyapatite. In contrast, µ-XANES data collected from hotspots of P identified using µ-XRF revealed localized domains of Fe- and Al-associated P together with Ca phosphates and vivianite-like Fe(II) phosphate phases that are not observed in bulk spectra. Thermodynamic modeling shows strong supersaturation with respect to vivianite and weaker supersaturation for Ca-P. The results of this study imply that P dynamics in neutral, Ca-rich peat soils are governed by microscale mineralogical heterogeneity that is not captured by bulk analytical techniques, suggesting that conventional Fe-based or bulk redox indicators may overestimate P release following peatland rewetting.
Comments to the manuscript egusphere-2026-5564 submitted to Biogeosciences by Adrian F. Florea, Dominik H. Zak, Case van Genuchten, Camille Rivard, and Hans Christian B. Hansen
General comment:
The paper of Florea et al. addresses a timely issue with broad environmental relevance, namely the risk of P mobilization into the hydrosphere after rewetting of agriculturally used drained peatlands, which may result in eutrophication of downstream freshwater and ecosystems. The paper addresses relevant scientific questions within the scope of Biogeosciences and presents novel concepts, ideas and data. Combining traditional wet-chemical soil and soil solution analyses with synchrotron-based XANES, µ-XANES, and µ-XRF techniques, the authors assess major P binding forms in a rewetted Danish peatland and discuss key biogeochemical processes derived from the data. The paper is clearly written and concise, the tables and figures are well-designed and easy to understand. New substantial conclusions have been reached; namely the co-occurrence of different P retention processes after peat rewetting at the micro-scale, including the formation of vivianite and Ca-P forms, likely hydroxy apatite in micro-sized hotspots. The scientific methods and assumptions are valid and clearly outlined. The results are largely (but in my opinion not always) sufficient to support the interpretations and conclusions. The description of experiments and calculations are mostly complete and precise to allow their reproduction by fellow scientists – minor exceptions are mentioned below. The authors give proper credit to related work and clearly indicate their own contribution. The title clearly reflects the contents of the paper. The abstract provides a concise and complete summary. The overall presentation of the manuscript is well structured and clear. The language is fluent and precise. Mathematical formulae, symbols, abbreviations, and units are mostly correctly defined and used – minor issues are mentioned below. The number and quality of the references as well as the amount and quality of supplementary material are appropriate. Based on the good scientific significance of the study, the good scientific quality, and the good presentation quality of the manuscript, I recommend acceptance of the manuscript after moderate revision. In the revised version of the manuscript, my following concerns and suggestions should be addressed:
Regarding the µ-XRF plots, I suggest the inclusion of additional RGB panels with the end members P-Ca-Fe to assess the co-localization of P with Ca vs. Fe as well as P-Al-Fe to assess the co-localization of P with Fe vs. Al. The already presented RGB panels with the endmembers P- Al- Si show that a large part of the P in the investigated soils is free organic P as well as Al+Si-bound P, probably in clay minerals and/or Al-Si-hydroxy polymers. The suggested addition of the two additional panels for each sample would additionally allow assessing the co-localization of the P amount bound to Fe and Ca vs. Al. This information then can be compared with the results of the LCF conducted in the selected ROIs as suggested earlier. When doing so, the limitations of both methods should be discussed.
Finally, I have some problems with the repeated statement in the Discussion that in contrast to a readily formation of vivianite, the formation of Ca phosphates is kinetically restricted in the Loevenborg peat. In my opinion, this reasoning is not supported by the reported marked oversaturation of the soil solution with respect to vivianite, whereas the SI of hydroxy apatite is close to zero.
Specific comments:
Line 65
The terms “Al oxides” and “residual Fe(III) oxides” are very general. Please add examples to specify these ”oxides” in more detail. What about Al-hydroxy-(poly)cations sorbed on clay minerals? According to Gerard (Gérard F 2016: Clay minerals, iron/aluminum oxides, and their contribution to phosphate sorption in soils—A myth revisited. Geoderma, 262, 213-226) the latter can be important P sorbents in circum-neutral soils – but probably not so much in sandy environments like Loevenborg.
Line 76
I wonder if the peatland rewetting also resulted in soil pH changes (increase), because Fe reduction processes are associated with proton consumption.
Line 78
I wonder whether potential pH increases may affect the equilibrium HS-/S2- and thus the formation of Fe(II) sulfides in addition to/rather than vivianite?
Line 96
Do these methods also extract Fe(II) complex-bound to SOM? The latter Fe species may probably be relevant in anoxic peatlands.
Line 103
Maybe also the paper of Werner et al. (2017) should be cited here.
Line 123
When (at which time in the year) did you sample the soils? Have the soils been rather anoxic or oxic in the most recent time before sampling? This should be stated in more detail here, because this information is relevant in the Discussion (lines 501-520).
Line 125
When had the site been rewetted? Give exact date and indicate the time between rewetting and soil sampling.
Line 127
I would expect that most pores of the peat (at least the fine and medium pores) are filled with water up to the surface due to capillary rise from the groundwater table which is at most 30 cm below the surface. However, the capillary water may contain oxygen to produce the partially oxic features in the uppermost peat as reported in Figure 1 of your study. Please clarify in more detail.
Line 150
It would have been interesting to additionally see data on dissolved organic P (DOP) in the soil solution here. Maybe there is some or even considerable P export as DOP in addition to the negligible export of inorganic P.
Line 156
Insert abbreviation for residual sorption capacity (RSC) here, because this term is used in equation 1 without prior introduction.
Line 160
Is the sum (Alox + Feox) calculated on a mass or mole basis? The common way to sum up the two elements for retention studies is either the mole basis, or calculating the RSC on a mass basis as (Al + 0,5 Fe), because the atomic mass of Fe is twice that of Al.
Line 163
What is PSC? The term has not been introduced so far. Is it equal to RSC?
Please clarify in the revised version of the manuscript.
Line 164
“Porg” as calculated by subtracting molybdate-reactive (orthophosphate) P from total P as determined by ICP-OES may also contain molybdate-non-reactive inorganic P forms like polyphosphates. Please clarify.
Line 169
Please add a rationale for investigating oxic and anoxic soil samples and shortly specify sampling and sample processing details for both variants in the manuscript in addition to the description in the Supplementary Information. It is important for the reader to know what both variants are intended for and what the different results of both variants may tell us.
Line 175
Give more experimental details for the XANES studies, e.g. on beam intensity, monochromator, detector, chamber type (vacuum, He filled, ….), beam energy calibration...
Line 195
Please be more specific here and mention in detail all P reference spectra used in your study as well as their source in a table in the Supplementary Information. Please clarify also in the spectra of “P sorbed to….” is organic (e.g. IHP, others) or inorganic (probably orthophosphate). K-edge XANES spectra of organic and inorganic P sorbed to the same Fe mineral sometimes have different shapes (particularly pre-edge signal intensities) – see Prietzel & Klysubun 2018: Phosphorus K-edge XANES spectroscopy has probably often underestimated iron oxyhydroxide-bound P in soils. Synchrotron Radiation 25, 1736-1744.
Line 198
Why did you only make “fingerprint” assessments? Additional spectra deconvolution (e.g. by LCF) would substantially strengthen your paper.
Line 207
A total C content of about 28% means that the SOM content is about 50-55%, implying that about half of the material in the peat samples is inorganic. Is this true? Which minerals may contribute to the inorganic share? If there is information available on this issue it should be mentioned and used for P retention argumentation in the Discussion.
Line 220
(Table 1): Do the samples contain carbonate? The high soil Ca contents (about 2%) together with the high pH suggest a CaCO3 content of 5%. The presence or absence of carbonate should be stated in the text, because carbonate may act as important P retention agent and also explain the Ca-P forms observed in the micro-XANES measurements. If there is CaCO3 in the samples, retention of inorganic and organic P by CaCO3 is important (see Celi et al. 2000; Interaction of inositol phosphate with calcite. Nutrient Cycling in Agroecosystems 57, 271-277; Prietzel et al. 2022; Soil phosphorus status and P nutrition strategies of European beech forests on carbonate compared to silicate parent material. Biogeochemistry 158, 39-72) and should be discussed in the paper.
Line 254
This applies only for molybdate-reactive orthophosphate-P, but not total P, am I correct? If so, please clarify.
Line 298+
It would be nice to have LCF results here. Then the relative shares of free organic P, Al-bound P, Fe(III)-bound-P and Ca-bound P could be estimated at least semi-quantitatively here and presented explicitly.
Line 299
(Figure 2): I would put the anoxic sample spectra above, because all oxic samples were first anoxic, and oxidized after sampling, am I correct? I also would change the spectra sequences within the anoxic and oxic spectra sets, starting with most reduced L1. It should be explained if the “P sorbed to Al oxide” in the reference spectrum is organic or inorganic. Moreover, Al oxides do not exist in soils, so why has this spectrum been used as reference? The authors should comment on this in the revised version of the manuscript or replace the P-Al reference.
Line 310+
Again, I would start with the anoxic samples, because all oxic samples were first anoxic, and oxidized after sampling, am I correct?
Line 314
What about spatial associations of P with Ca (and Fe)? Please add additional RGB panels in Figures 3 and 4 (see also general comment and comments to lines 333, 402).
Line 322
Why not also organic P? LCF may clarify.
Line 333
My suggestion for the authors is to add RGB panels with the end members P-Fe-Ca (and P-Fe-Al) to address P co-localization with Fe, Ca, Fe-Al mixed minerals).
Line 369
Should read: “and/or”
Line 402
My suggestion for the authors is to add RGB panels with the end members P-Fe-Ca (and P-Fe-Al) to address P co-localization with Fe, Ca, Fe/Al-mixed minerals)
Line 476
Add: “µ-XRF,”?
Line 484
The considerable oversaturation of the soil solution with respect to vivianite suggests that vivianite formation is hampered by kinetic constraints and/or that the Fe and P species included in the analysis do not properly represent the actual species present in the soil solution.
Line 488+
Line 496+
In my opinion, the high oversaturation of the soil solution with respect to vivianite in the study suggests the opposite, namely a kinetically inhibited vivianite formation – see last comment.
Line 509
This statement can probably be tested using visual MINTEQ.
Line 511
Here the sampling date becomes important… (see comment to line 123).
Line 519
This is probably also true for vivianite formation and may be a reason for the observed high oversaturation of the soil solution with respect to vivianite…