the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Tropical Andosol organic carbon quality and degradability in relation to soil geochemistry as affected by land use
Sastrika Anindita
Peter Finke
Steven Sleutel
Abstract. Land use is recognized to impact soil geochemistry on the centennial to millennial time scale, with implications for the distribution and stability of soil organic carbon (SOC). Juvenile volcanic soils in tropical areas are subject to much faster pedogenesis, with then also possibly a significant mediation by land use on much shorter centennial or even decadal scale. Very scarce observational evidence exists and so such indirect implications of land use on SOC cycling are largely unknown. We here investigated SOC fractions, substrate specific mineralization (SOC or added plant residue), and net priming of SOC in function of forest or agricultural land use on Indonesian volcanic soils. The content of oxalate-extracted Al and exchangeable Ca correlated well with OC associated with sand–sized aggregates. The concomitant near doubling of the proportion of SOC in sand-sized aggregates compared to forest and likewise contrasts in Al and Ca suggest an enhanced formation of Al- (hydr)oxides and liming promoted aggregation and physical occlusion of OC. This was importantly also consistent with a relatively lesser degradability of SOC in the agricultural sites, though the net priming of SOC and degradability of added 13C-labelled ryegrass was not found to depend on land use. We expected that the formation of amorphous Al after conversion of native forest to agriculture would mainly have promoted mineral-association of SOC compared to under pine forest but found no indications for this. Enhanced weathering but improved small scale aggregation of tropical Andosols caused by conversion to agriculture may thus partially counter the otherwise expectable decline of SOC stocks following cultivation. Such indirect land use effects on the SOC balance appeared relevant for correct interpretation and prediction of the long-term C-balance of (agro)ecosystems with soil subject to intense development, like the here studied tropical Andosols.
Sastrika Anindita et al.
Status: final response (author comments only)
-
RC1: 'Comment on soil-2022-13', Anonymous Referee #1, 25 Feb 2022
This is a highly competent paper with very interesting content. The impact of landuse on OC stabilisation and storage is a major issue and volcanic soils have not been investigated intensively in this respect. The paper describes the comparison of forested and agricultural areas using quite sophisticated methods including 13C labelling, soil fractionation and incubation studies. I find the manuscript highly interesting and professional.
Some issues should maybe addressed befor acceptance:
Line 51: Reading the cited reference it turned out that the comparison of agriculture and native forested land was performed after 15 and 5 years after conversion, not after centuries or decades. So you might revisit the cited paper.
2.1. can you exclude small scale climatic differences between the sites? Is the orientation of sites the same and the precipitation?
The yearly organic matter inputs into the agricultural soils by manures is massive. How does this influence the overall results as compared to sites receiving e.g. only mineral fertilizers?
Results section: I do wonder, why you did not inlcude OC stocks of the profiles, as you estimated bulk densities of the soil horizons, this is not a bigt deal, but would add to overall conclusions? So absolute OC amounts per horizon could also be an interesting value, although, relating SOC to equivalent soil masses would be even better (Haden von, A.C., Yang, W.H., DeLucia E.H., 2020. Soils' dirty little secret: Depth-based comparisons can be inadequate for quantifying changes in soil organic carbon and other mineral soil properties. Glob Change Biol 26, 3759–3770.)
Ad all figures. Letters are too small.
Line 253: Comma after "Relatively"
Ad 4.2.: Of course, the discussion of the impact of land-use on SOC ihn texture fractions is not easy. The measurement of 13C in texture frations after the incubation would be highly interesting to show, where young SOC fractions end up. Sometimes the youngest material ends up in the smallest fraction (Gerzabek et al. 2001. Soil organic matter pools and 13C natural abundances in particle size fractions of a long-term agricultural field experiment receiving organic amendments. Soil Science Society of America Journal 65, 352-358). That means that in forest soils, where the mineralization often is hampered by less favorable situation for microbes, a intermediates of the mineralization process could show up in the smallest fraction. Does this influence your interpretation?
Conclusions: Before drawing final conclusions that weathing is the major factor for differences between forest and agricultural sites, it should be clear what is the approximate net input of OC into the soils of the different sites. What is the impact of the massive organic fertilizer input on agricultural plots and how does the overall SOC stocks develop under forest and agricultural land, taking into account BD and, even better, the equivalent mass method?
Â
Â
Citation: https://doi.org/10.5194/soil-2022-13-RC1 -
AC1: 'Reply on RC1', Sastrika Anindita, 10 Jul 2022
Thank you for the comments and suggestions. Our responses are listed below in the same order as the referee's comments.
- Line 51: We have checked the article by Gerzabek et al. (2019) and indeed a comparison of SOC stock, pedogenic Al and Fe was made between native forest and agriculture after 5 and 15 years of conversion. We propose to revise the sentence accordingly to: "A recent study on the Galapagos island demonstrated that even 15 years conversion of native forest to cultivated land strikingly accelerated soil weathering (Gerzabek et al., 2019)"
- Regarding the small climatic differences between sites: all the sites have a strong leaching regime (between 1850 – 2600 mm year-1) and a similar slope and exposition. For this reason, we do not expect differences in weathering and leaching between the sites, even though there are some differences in climate.
- Regarding the OC stock: we agree to include OC stocks of each soil carbon pool per sampled layer as per the referee’s advice. The outcome of a statistical comparison of OC stocks between the land-use groups yielded the same result as that obtained for SOC contents (as bulk densities were in fact highly comparable between the soils). Accordingly, reporting SOC stocks would not bring about major changes to the discussion section or to the conclusions.
- Figures: We will increase the font size in all figures
- Line 253: Will be revised accordingly
- Ad 4.2, Regarding our interpretation about the mineralisation process that could show up in the smallest fraction when the situation is less favorable for microbes: we agree that for the soil incubation experiment, a comparison of the entrance of the exogenous OM-C (untransformed or microbially processed) into the various separated soil fractions (MAOM, POM, occluded POM)) could provide extra insight into whether or not potentially contrasting governing SOM-protection mechanisms predominate between the studied soils. However, the contrast in 13C-content of the used plant material (δ13C +50‰) and native SOC (δ13C  -13 to -27‰) is probably too small for proper tracking of ryegrass-C entrance into soil fractions, particularly given the large SOC background concentration of most of the included soils. This would most likely obstruct robust statistical comparison of ryegrass C preservation in function of land use (forest and agricultural land use), particularly as the soil respiration assays already revealed wide variation in % ryegrass mineralized between the forest soils. Moreover, even if we were to dispose of a substrate with more strongly contrasting δ13C vs. SOC than the one used here, we still think that the primary conclusion of this study - long-term formation of pedogenic Al and amorphous materials and accumulation of Ca by liming stimulated occlusion of POC into microaggregates – would not have been well testable by a simple lab incubation experiment such as the one performed here. The formation of soil aggregates depends on plant, soil microbial and soil faunal activity as well as on fluctuations in environmental conditions (like moisture content). In addition, the diminution of the added OM into smaller-sized particles that may more readily end up inside microaggregates could be of importance. The simplistic approach used here: short-term incubation of small soil cores at constant moisture level without plants and no inoculation is likely unfit to realistically reproducible actual aggregate occlusion as it were to occur on the longer time scale in the field. We are therefore reluctant to also fractionate soil cores at the end of the performed soil respiration assays and measure δ13C levels of soil fractions – as this would involve a substantial extra amount of work with probably limited added value to support or disprove the current interpretations.
- Conclusions, net-input OC, the overall SOC stock under forest and agricultural land, and comparison with sites receiving e.g. only mineral fertilizers: We acknowledge that data on OC inputs would have been very useful to more strongly ascertain that differential pedogenesis indeed caused observed differences in aggregate occluded SOC and other fractions between the two investigated land-uses. At all three agricultural sites farmers indeed apply substantial quantities of manure. This is the default practice in the Mt. Tangkuban Perahu and Mt. Burangrang areas. We in fact did not find any field where only mineral fertilizer is used, and hence comparison with sites that do receive manure would not be possible. Obviously, one can expect such large manure amendments (please see the below table) to non-negligible contribute to the SOC balance. We propose to clearly list organic fertilizer C inputs and balance these vs. plant-C inputs in the revised version. Unfortunately, measuring plant-C inputs would require a huge amount of extra work and is also methodologically extremely challenging, especially measurement of below-ground C inputs from the considered tropical native forest and secondary pine forest seems out of reach for this study.
As an alternative, we propose to introduce estimates of vegetation C inputs for the included land uses into our discussion. When estimating C-inputs for the arable sites, we could base these on known residue C-inputs for the crop rotations and the known use of exogenous OM. For the forested sites, ranges of vegetation C-input estimates could be taken over from the literature. For this, we only used literature references on comparable native forest and secondary pine forests in Indonesia. The below table presents estimated plant and manure C inputs to the soil for each site (next to SOC stocks). We propose to add these C-input estimates to Table 1 and three new literature references.
Site
land use
Plant C -input
(Mg ha-1 yr-1)
Organic fertilizer C – input
(Mg ha-1 yr-1)
Total SOC stock (0-20 cm)
(Mg ha-1 yr-1)
Total SOC stock (0-80cm) (Mg ha-1yr-1)
NF
natural forest
6.7 – 12.2(1)
-
56.9
134.8
PF1
pine forest
3.6 – 9.1(2)
-
30.4
54.4
PF2
pine forest
3.6 – 9.1(2)
-
33.0
72.1
AG1
agriculture
1.8
11.50
40.3
72.0
AG2
agriculture
1.6
6.4
71.4
139.1
AG3
agriculture
1.0
6.8
61.1
105.8
(1)Guillame et al. (2018), https://doi.org/10.1038/s41467-018-04755-y and Hertel et al. (2009), https://doi.org/10.1016/j.foreco.2009.07.019
(2)Bruijnzeel (1985), http://www.jstor.org/stable/2559453
Although the estimated ranges in C-inputs are forcefully very wide they do readily reveal that C inputs to the agricultural plots are probably similar to or even above those in the forest soils. Thus, we recognize that it was not possible to conclude that the higher OC stock in agricultural compared to pine forest soils was solely due to the observed enhanced soil weathering as a result of conversion from forest to agricultural land use. We propose to change L405-409 of the conclusion into, "Based on the present study, we postulate that the enhanced formation of amorphous minerals and Alo under agriculture with high OM inputs promoted the development of stable soil aggregates and OC occlusion therein and this would in part counter otherwise expected losses of SOC compared to secondary forest. However, the contribution of large OM input vs. land-use conversion per sé could not be elucidated here and this will require study of other tropical Andosol forest-agricultural land use pairs with detailed inventory of OC inputs".
Citation: https://doi.org/10.5194/soil-2022-13-AC1
-
AC1: 'Reply on RC1', Sastrika Anindita, 10 Jul 2022
-
RC2: 'Comment on soil-2022-13', Anonymous Referee #2, 13 Jun 2022
The current study attempted to examine land-use effect on SOC quality in a volcanic region of Indonesia by selecting 6 sites (3 forest and 3 agricultural), focusing 3 depth layers, and using two approaches – physical fractionation to distinguish different types of SOC and laboratory incubation to assess C degradability including possible priming effect. Tropical Andisols are clearly less studied than temperate ones and the land-use effect on SOC and other soil fertility related parameters is clearly an important topic in soil and agricultural science. The authors clearly put lots of effort to conduct both physical fractionation study and lab incubation study. While the results are potentially valuable, I found several serious limitations and strongly question some of the result interpretations.
First, I think the authors need to carefully consider if the observed soil chemical variations found here really resulted from the land-use difference. There is no information on how the authors judged the six sites were similar in terms of pedogenic factors except for the land use. Quite a variation in Alo and Feo contents is present within forest sites and within agricultural sites. I assume the sites are on a volcanic mountain slope. Rather small differences in topography and aspect can create significant differences in soil properties in the tropics due to much longer land history and more intensive weathering and erosion as well as (re)depositions of tephra. Local people usually select specific land forms for specific management (e.g., ag vs. forest). The high variability in Alo and Feo even at 60+ cm depth among the sites suggest that the six sites don’t have the same parent material. Volcanic landscape is complex. These areas might have received various types of tephra from nearby volcanoes and also from at distance time to time. Tephra (esp. ash) is often redistributed by wind and water. Thus, more detailed information on the six sites is clearly needed to judge the validity of testing “land-use effect” from these sites. Also I cannot find any information on the soil sample replication per site.Â
Second, I wonder the effectiveness of Zimmermann fractionation which was developed for Switzerland agricultural soils that are wildly different from tropical Andisols. While the isolation of POM is reasonable, Zimmermann method gives ambiguous separation of the SOC pools that are associated with soil mineral matrix in various ways. Particularly the distinction between S+A (mostly sonication-resistant aggregates of >63 um) and s+c (silt & clay size particles after the sonication) is rather problematic for following reasons. First, aggregate can be <63 um and the sonication energy used is too weak to break most aggregates in Andisols that are known to have high physical stability. Second, the aggregation strength and aggregate size are strongly affected by the amounts of short-range-order minerals, other metals, and OM. The six soils studied here showed a wide variation in mineralogical properties such as Alo and Feo (Table 1). Thus, what the authors recognized as S+A and s+c likely change among the six soils. This probably contributed to some of the unexpected results. I recommend the authors to characterize S+A and s+c fractions further (e.g., for oxalate-extractable Al and Fe). This would strongly help to interpret what these fractions represent. Otherwise, it seems impossible to deduce any C protection mechanisms responsible in these fractions.
I also found puzzling statements and interpretations on C storage and stabilization mechanisms. For instance, the authors cannot assume that C present in “S+A” fraction is occluded OM in aggregate. Clays and esp. SRO minerals present in S+A fraction (probably as microaggregates) can directly sorb relatively large amounts of OM. The authors should report C:N of each fraction in Table 2, which help to interpret what types of OM in S+A fraction and in other fractions. In general, the authors should to be more critical when interpretting their results.Â
For these reasons above, the conclusion drawn by the authors have very weak basis. While the data is potentially valuable, I think the manuscript at its current state should be rejected.
Citation: https://doi.org/10.5194/soil-2022-13-RC2 -
AC2: 'Reply on RC2', Sastrika Anindita, 11 Jul 2022
Thank you for the comments and suggestions. Our responses are listed below in the same order as the referee's comments.
- Regarding the parent materials between sites: Testing the similarity of parent materials is an interesting point but was problematic to evaluate because not all the samples have C horizons, so the signal of the parent material is obscured. Before our study, previous studies reported the age in our study area for 1 m depth approximately between 8000 – 10000 years (Utami et al., 2019, https://doi.org/10.1016/j.catena.2018.09.024; Chartres and Van Reuler, https://doi.org/10.1111/j.1365-2389.1985.tb00322.x). Thus, we did not expect inhomogeneous parent materials in our study area. Detailed information and soil properties on the six sites can be found in our recent publication (Anindita et al., 2022, https://doi.org/10.1016/j.geoderma.2022.115963). We have taken and analysed replicate (duplo) samples for each layer. Based on that study we found sufficient grounds to indeed assume that the parent materials in our study are similar. The area might receive various types of tephra from nearby or at distance and redistribute by wind, but the estimated mineral content (by quantitative X-ray diffraction) was found to be comparable between all sites, except NF. Also, a comparison of weathering degree and total oxide composition of the site revealed a close likeliness of all soils, except again the NF site. As explained in the current manuscript, the NF site is considered an exception because it is located within the 1.5 km distance from the crater, with the presence of new ejecta on the top of soil. Accordingly, for most observations (SOC fractions, C-mineralization data), the effect of land-use conversion in the current manuscript was based on a comparison between the pine forest sites and the agricultural sites only, with comparable parent materials. The history and management of land were therefore concluded to have effects on the differences of soil properties between sites. We understand that the inclusion of the NF forest site data could have been misleading and we will carefully rescan the entire manuscript for unambiguous phrasing on this matter. If this is found necessary by the editor, we will also add data from our previous research, such as (i) weathering index: SiO2/(Al2O3 + Fe2O3 + TiO2) and (K+Ca)/Ti, and (ii) estimated mineral amount, (iii) texture, (iv) %volcanic glass, (v) code for soil classification. However, such would imply a direct repetition of previously published information, for which we would need to ask reproduction permission from the published of Geoderma, Elsevier.
- Regarding the Zimmermann-fractionation method and weak sonication energy level: We agree with the referee that the “Zimmermann-fractionation” most likely yields composite soil fractions, as would in fact most soil fractionation methods. However, aggregate stability is indeed particularly strong in Andosols and we share the referee’s scepticism on the ability of the applied 22 J ml-1 ultrasonication step to sufficiently disrupt sand-sized soil aggregates. As a consequence, the currently isolated S+A fraction (sand + aggregates > 63µm; after ultrasonication) would possibly also contain a substantial part of the silt+clay associated OC (s+c). After rechecking we can indeed acknowledge the presence of undisrupted microaggregates in our S+A fractions and we would follow the referee’s advice to further subdivide it into >63µm sized occluded-POM and silt+clay associated OM. For this, a stronger mode of soil dispersion followed by wet sieving at 63µm and analysis of the resulting two-size fractions would be needed. We also now tested the successfulness of this procedure by subjecting the S+A fractions of three of the soil samples to 400 J ml-1 ultrasonication and this led to full dispersion of >63µm aggregates (other tests at 200 J ml-1 proved insufficient). In these three test samples, the %of SOC of the S+A OC lowered by 8-31% vs. the original S+A OC estimate. If the editor agrees, we propose to thus further subdivide the S+A fraction into a sand + occluded-POM fraction and into s+c OC for all soils (3 replicates x 6 sites x 3 depths). The amount of C in the latter fraction could be added to the already separated s+c. Relevant discussion parts will be revised accordingly with these updated S+A 400 J ml-1 and s+c OC data.
Citation: https://doi.org/10.5194/soil-2022-13-AC2
-
AC2: 'Reply on RC2', Sastrika Anindita, 11 Jul 2022
-
RC3: 'Comment on soil-2022-13', Anonymous Referee #3, 14 Jun 2022
The submitted study examines the relationship between organic carbon storage/stability and land-use change in soils of the tropics. The chosen approach pursues the idea that the soil geochemistry changes due to the history of land use and that this causes a change in the stabilization of the soil organic carbon. The authors base this assumption on the fact that changes in soil geochemistry have an effect on the soil mineral phase, which in turn indirectly affects carbon stabilization. The connection of the mineral phases with their characteristic properties and their changes by exogenous factors is a key point for the prediction of soil carbon stabilization under a changing environment. Six sites under three different land uses were selected for the study. andic cambisols (primary forest and pine forest land use) were described at two sites and aluandic andosols (pine forest and horticulture land use) at the other four sites. Despite the global distribution and the high agricultural relevance of Andosols (land-use change), there are many gaps in knowledge, especially with regard to the behavior of the mineral phases in tropical regions. Although the approach of the study is very intresting, there are some ambiguities in the methodological approach and evaluation of the results.
The novelty of the study is justified by the fact that an effect on the soil mineral phase is assumed to be triggered by land use change. This assumption is based on a single case study that, however, covers fundamentally different soils. Furthermore, an own study is given, which has not yet been published. This makes it very difficult for the reader to follow the chosen approach of the submitted study. Perhaps the integration of data from the other study would be useful for a better understanding, since otherwise the present study cannot stand alone.
This would possibly also invalidate another point of criticism, which relates to the locations used. Their description shows that not all land uses have received a comparable parent material for soil formation through the prevailing volcanism (e.g. "The proportion of primary minerals is higher in the younger NF soil as it is closer (within 1.5 km) to the crater of Mt Tangkuban Perahu and received ash more recently."). It can be assumed that possible changes in the properties of the soil mineral phases due to changes in land use are already superimposed by differences in the parent material. This makes the provision of detailed mineralogical data all the more important for understanding the sites. Furthermore, this should also better allow for a clearer link between existing and suspected historical geochemistry of the soil and its impact on the properties of present-day mineral phases.
Due to the objective of the study, the mineral phases should be characterized as precisely as possible and what changes occur with longer-lasting agricultural use. The description of the fractionation scheme used gives me the impression that there can be overlaps at least between S+A and s+c fractions. It is conceivable that mineral-associated OC is contained in aggregates of the S+A fraction, which could not get into the s+c fraction due to the relatively low application of 22 J/mL. Which does not allow an evaluation of the mineral phase characteristics and their effect on the OC stabilization. Provideing the database for deriving the selected energy level would support the results shown in the study and the conclusions derived from them.
For the reasons listed, I recommend that the submitted work be rejected. However, due to the high relevance, I recommend that the authors resubmit the study after it has been thoroughly revised.
Citation: https://doi.org/10.5194/soil-2022-13-RC3 -
AC3: 'Reply on RC3', Sastrika Anindita, 11 Jul 2022
Thank you for the comments and suggestions. Our responses are listed below in the same order as the referee's comments.
- Regarding different parent materials between sites: Our previous study found that the selected soils were comparable based on the weathering degree and the estimated mineral content, except for the NF. For a detailed description of the minerals and soil properties, we referred to our recent publication in Geoderma (Anindita et al., 2022, https://doi.org/10.1016/j.geoderma.2022.115963). If the editor finds it necessary, we can ask permission from Elsevier to reproduce part of these data, such as (i) weathering index: SiO2/(Al2O3+Fe2O3+TiO2) and (K+Ca)/Ti, (ii) estimated amount of mineral, (iii) texture, (iv) volcanic glass%, (v) code for soil classification from our previous research. According to literature and historical map, the areas near Tangkuban Perahu and Burangrang mountains were forests and most of the areas were converted into pine forest and agricultural sites. Only the NF site represents an original forest. However, this site is located within the 1.5 km distance from the crater, so there was a possibility presence of new ejecta in the topsoils and we confirmed this difference from the total oxide content (Anindita et al., 2022, on request we can ask for reproduction in the revised version). On the other hand, the weathering index and the estimated amount of minerals of pine forest and agricultural soils were comparable, thus we did a comparison between these two land-use types (please see also our response to an alike comment by referee 2). We apologize for apparently not having made this sufficiently clear in our original manuscript.Â
In Anindita et al., 2022 we concluded that only NF has a deviating geochemistry of the top one meter of soil, the other soils have similar parent materials and similar age as well. Other studies reported oxides content below one-meter depth which possibly has different parent materials.
- Regarding the overlaps between S+A and s+c fractions: We agree that the 22 J mL-1 >63µm S+A fraction contains mineral-associated OC as well and propose to remediate this by adding an extra fractionation step. This would then allow specific quantification of POM occluded in the S+A fraction. After rechecking, we can indeed acknowledge the presence of undisrupted microaggregates in our S+A fractions. We also now tested the successfulness of the dispersion procedure by subjecting the S+A fractions of three of the soil samples to 400 J ml-1 ultrasonication and this led to full dispersion of >63µm aggregates. In these three test samples, the %of SOC of the S+A OC lowered by 8-31% vs. the original S+A OC. If the editor agrees, we propose to further subdivide the S+A fraction into a sand + occluded-POM fraction and into s+c OC for all soils (3 replicates x 6 sites x 3 depths). The amount of C in the latter fraction could be added to the already separated s+c. Relevant discussion parts will be revised accordingly with these updated S+A 400 J ml-1 and s+c OC data.
As explained in our responses above, a detailed comparison of mineralogy and weathering stage presented previously (Anindita et al, 2022) does demonstrate the alikeness of the included pine forest and agricultural sites in terms of parent material, validating to use these sites to investigate the impact of land-use on soil mineralogy and soil carbon quality. With some modification to the text and tables, this first major concern could be accommodated. With respect to referee 3’s second main comment, we are convinced that ambiguity in the current interpretation can be resolved by introducing the new data on OC in sand + particulate OC in the S+A fraction and s+c therein a moderately revised version of the current manuscript.
Citation: https://doi.org/10.5194/soil-2022-13-AC3
-
AC3: 'Reply on RC3', Sastrika Anindita, 11 Jul 2022
Sastrika Anindita et al.
Sastrika Anindita et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
410 | 170 | 21 | 601 | 8 | 10 |
- HTML: 410
- PDF: 170
- XML: 21
- Total: 601
- BibTeX: 8
- EndNote: 10
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1