the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Jul 2026
Decoupling phosphorus availability and fixation in tropical soils: roles of iron and aluminum oxides across agroecological gradients in Nigeria
Abstract. Phosphorus (P) deficiency and fixation are major constraints to crop productivity in tropical soils, yet their large-scale interactions remain poorly understood. This study provides a national-scale assessment of how short-range-ordered iron (Feₒₓ) and aluminum (Alₒₓ) oxides, soil properties, and climate jointly regulate P availability across Nigeria. A total of 1,382 topsoil samples (0–20 cm), collected using a hierarchical Soils4Africa framework, were analyzed for extractable P, Feₒₓ, Alₒₓ, soil organic carbon (SOC), texture, and pH, and integrated with precipitation data. We developed a phosphorus sorption index (PSI) to quantify fixation potential as the molar ratio of oxide abundance to available P. Soil properties varied strongly along agroecological gradients, with Feₒₓ, Alₒₓ, SOC, precipitation, and PSI increasing from semi-arid savannas to humid forest systems, while pH declined. Contrary to the expected inverse relationship, extractable P was positively associated with Feₒₓ, Alₒₓ, SOC, and precipitation, but negatively related to PSI. Multivariate analyses clarified this apparent paradox: SOC exerted the strongest positive effect on P availability, whereas Alₒₓ had a strong negative effect consistent with sorption, and precipitation influenced P both directly and indirectly through oxide accumulation. These results demonstrate that P availability and fixation are not simply inversely coupled but are co-regulated by climatic and biological processes. The PSI effectively captured spatial variation in fixation potential, with the highest values in humid regions characterized by intensive weathering. These findings highlight the dominant role of agroecological gradients in structuring P dynamics and reveal contrasting nutrient constraints across regions: oxide-rich humid soils are limited by fixation, whereas northern savannas are constrained by low inherent P. This study provides a mechanistic basis for site-specific phosphorus management and supports the development of sustainable nutrient strategies to improve P-use efficiency in tropical agroecosystems.
- Preprint
(4866 KB) - Metadata XML
-
Supplement
(128 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2815', GEORGES MARTIAL NDZANA, 30 Sep 2026
-
RC2: 'Comment on egusphere-2026-2815', V. Logah, 01 Nov 2026
The manuscript explores an important issue of phosphorus fixation in acid soils of the tropics. Drawing on a national-scale dataset from the Soils4Africa project, the study evaluates how soil properties and climatic factor (precipitation) modulate soil P availability across Nigeria. Although the rationale of the manuscript is good, it requires substantial improvements before it can be considered for publication in SOIL. These revisions relate primarily to the Methodology and Results and discussion sections, as indicated below:
Major Comments
1. Inadequate information on agro-ecological zones
Although annual rainfall amounts were provided for the agro-ecological zones, important information on rainfall distribution patterns, the duration of the rainy season(s), relative humidity, and other climatic factors is lacking. Furthermore, no description of the vegetation characteristics of the agro-ecological zones was provided.
Recommendation: Authors to provide adequate descriptions of the climatic conditions and vegetation characteristics of the agro-ecological zones.
2. Conflicting description of calibration and validation sets
In lines 175-176, authors state that 20% of the soil samples were used as a hold-out validation set. In line 179, they indicate that 20% of the total soil samples were selected using the Kennard-Stone algorithm for calibration and subjected to wet chemistry analytical procedures. Consequently, this raises a number of questions detailed below.
- Was only 20% of the samples subjected to wet chemistry?
- Was this 20% used for calibration and the remaining 80% predicted spectroscopically?
- If only 20% had reference laboratory measurements, how was the validation set constructed?
- What proportion was used for calibration, validation, and independent testing?
Recommendation: Authors to clarify the calibration/validation design.
3. Inadequate information on spectral pretreatment
The manuscript appears to excessively summarize the Methodology section which weakens the reproducibility of the study. For example, key processing methods in R were mentioned including rubber-band baseline correction, Savitzky-Golay, mean centering, scattering correction, unit variance scaling, derivatives. However, there were no details on which scatter correction method was used, whether first or second derivative was applied, neither was there any information on derivative window size and derivative polynomial order, etc. It is important to note that different pre-processing combinations can produce significantly different prediction accuracies.
Recommendation: I invite authors to clearly provide these details to allow for workflow reproducibility.
4. No description of model optimization
There is insufficient information on model optimization. For the Partial Least Squares (PLS) model, the number of latent variables used was not reported, nor was the procedure for determining the optimal number of components described. For the Random Forest approach, number of trees have not been reported. It is important to set it to a large value to ensure that it does not decrease the predictive performance. More so, authors did not report mtry parameter and neither described the tuning procedure. The absence of these methodological details limits the reproducibility of the modelling approach.
Recommendation: Authors to provide description of model optimization.
5. Weak validation strategy
Authors report a 20% hold-out validation. Generally, a single hold-out split is considered weak for spectroscopic validation.
Recommendation: Authors to consider including more robust approaches. For example, repeated cross validation, k-fold cross validation, nested cross validation.
6. No report on moisture and equilibration conditions
Authors reported that soils were ground to <0.5 mm but did not indicate whether samples were subjected to the traditional air-drying or was oven-dried at low temperature. In addition, no information was provided on sample storage conditions or equilibration procedures before spectral scanning. Residual moisture strongly affects MIR spectra.
Recommendation: Authors to provide complete sample preparation information.
7. Lack of soil distribution information in each zone
While the manuscript identified the major soil groups (viz., Acrisols, Ferralsols, Lixisols, and Luvisols) across the agroecological zones, it is unclear how soil sampling was distributed among the applicable soil groups in the zones.
Recommendation: It would be valuable to provide additional information or data showing the distribution of sampled soils among sampled soil groups in each agroecological zone. Although Figure 1 presents geological information, the relationship between the geological formations and the major soils is not clearly articulated. Strengthening this linkage would enhance the understanding and interpretation of the findings for the wider scientific community.
8. Fragmentation of Results and Discussion
The Results and Discussion section appears overly fragmented, with the discussion lacking sufficient depth and integration. Furthermore, Section 3.3, which addresses the core theme of the manuscript, is neither comprehensively reported nor adequately discussed.
Recommendation: Authors to merge closely related sub-sections to improve the flow and coherence of the manuscript and discuss trends or observations with scientific rigor.
Minor comments
- Climatic variable(s) not specified in Methods section.
- Tables 1 and 3 do not have units for any of the parameters measured.
- To give a clear overview of the textural classes, it is important to provide data also on sand and silt distributions at least in supplementary submission.
Citation: https://doi.org/10.5194/egusphere-2026-2815-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 251 | 107 | 36 | 394 | 61 | 34 | 34 |
- HTML: 251
- PDF: 107
- XML: 36
- Total: 394
- Supplement: 61
- BibTeX: 34
- EndNote: 34
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript addresses an important topic in tropical soil fertility by examining the relationship between phosphorus (P) availability, phosphorus fixation, and oxalate-extractable Fe and Al oxides across Nigeria using a large national-scale dataset (n = 1,382). The study is novel in its geographical scope and potentially valuable for nutrient management in tropical agroecosystems. The manuscript is generally well structured and presents interesting findings, especially the concept that P availability and fixation can be partially decoupled along agroecological gradients. However, several methodological, analytical, and interpretational issues should be addressed before publication.
Recommendation: Major Revision
Major Comments
The manuscript indicates that only 20% of samples underwent wet chemistry analysis and that MIR spectroscopy models were used to predict soil properties for the remaining samples. However, model validation statistics are not reported.
Critical metrics are missing:
Without these metrics, it is impossible to assess the reliability of the predicted soil properties that support all subsequent analyses.
Recommendation: Include a comprehensive calibration and validation table and provide prediction performance figures in supplementary materials.
The manuscript introduces:
as a phosphorus sorption index.
Several concerns arise:
The negative correlation between PSI and available P may be partly inherent to index construction rather than an independent ecological phenomenon.
Recommendation: Provide theoretical justification and sensitivity analysis. Discuss limitations of using PSI as a proxy rather than a direct measurement of sorption capacity.
Because available phosphorus is included directly in the PSI equation, conclusions such as:
"PSI was negatively correlated with P"
are mathematically expected to some extent.
The manuscript should clarify:
The kriging approach is inadequately described. Important information is missing:
Although a PSI map is presented, no geostatistical diagnostics are shown.
Recommendation: Report variogram parameters and kriging validation results.
The SEM is one of the manuscript's key analytical tools, yet important information is absent:
Without goodness-of-fit measures, it is difficult to evaluate the robustness of the SEM conclusions.
The manuscript argues that positive correlations between Feox and available P reflect broader pedogenic conditions and biological cycling.
While plausible, this explanation is not directly tested.
Alternative explanations include:
The discussion should acknowledge these alternatives and avoid overinterpreting observational correlations as causal mechanisms.
The study spans a large geographic area, yet key drivers of phosphorus availability are absent:
These variables can strongly affect available P and may confound observed relationships.
The limitations should be explicitly discussed.
Minor Comments
Section title:
"3.1 Geological Influence and interaction with agroecology"
appears after another section also numbered "3.1".
Renumber appropriately (likely 3.2).
Several equations are poorly rendered:
These should be reformatted using proper mathematical notation.
Examples include:
Professional copyediting is recommended.
Several duplicated citations exist:
Tables 1–3 provide means but not pairwise significance groupings (letters or post-hoc results).
The regression table is useful but should additionally report: