Depth-stratified soil organic carbon dynamics across forest–cultivation gradients in Nkanu East Watershed, Southeast Nigeria: implications for agroecological management

1 Introduction

Soil organic carbon (SOC) dynamics in tropical watersheds are fundamentally shaped by land-use conversion, yet conventional sampling approaches systematically underrepresent subsoil contributions that constitute the majority of profile carbon stocks. In sub-Saharan Africa, where approximately 65% of the population depends on smallholder agriculture, the conversion of forest to continuously cultivated land triggers cascading changes in soil physicochemical properties, hydrological functioning, and carbon storage capacity (Elwell, 1986; Hanson, 1976; Lal, 2004). Nigeria exemplifies these pressures: with a population exceeding 200 million and an agricultural sector dominated by subsistence and semi-commercial smallholders, forest-to-cultivation conversion has proceeded with limited soil conservation measures (Ezemedo and Igbokwe, 2013). The Nkanu East Watershed of Enugu State, located within the derived savanna agroecological zone of southeastern Nigeria, typifies this pattern: extensive areas formerly under tropical rainforest and secondary regrowth have been converted to continuous cultivation of cassava, yam, maize, cowpea, rice, and sweet potato over 40–50 years (Anikwe, 2010; Anikwe et al., 2026).

Soil organic matter (SOM) occupies a central position in tropical agroecosystem functioning, influencing nutrient availability, water retention, aggregate stability, cation exchange capacity (CEC), and pH buffering (Brady and Weil, 2017; Nair et al., 2021; Nwite, 2017). In the highly weathered Ultisols that dominate southeastern Nigeria, where low-activity clay minerals (primarily kaolinite) and high rainfall impose severe constraints on inherent fertility, SOM assumes particular importance as the principal source of negative charge and nutrient retention (Brady and Weil, 2017; Six et al., 2002). Soil organic carbon—the quantifiable carbon fraction of SOM, typically estimated as SOM/1.724—serves as both a diagnostic indicator of soil health and a manageable pool in the global carbon cycle (Anikwe, 2010; Lal, 2004).

The vertical distribution of SOC follows a predictable surface-concentrated pattern, yet subsoils (20–100 cm) contain approximately 50–67% of total profile SOC stocks in tropical forest ecosystems (Jobbágy and Jackson, 2000; Lorenz and Lai, 2005). This challenges the surface-centric perspective embedded in many soil carbon inventories. Recent research has reinforced the significance of deep-profile carbon: subsoils (>30 cm) contribute 25–55% of total profile SOC stocks in tropical ecosystems, with the proportional contribution increasing under certain vegetation types and geological settings (Bhattacharyya et al., 2007; Wright, 2016). Rumpel and Kögel-Knabner (2011) identified root-derived carbon, dissolved organic matter transport, and mineral-organic associations as the primary processes sustaining deep-profile carbon pools.

Despite the established importance of depth-resolved assessment, the majority of soil carbon studies in tropical agroecosystems continue to rely on fixed-depth sampling protocols that aggregate measurements across 0–15 cm, 15–30 cm, and 30–60 cm increments (Celik, 2005; Dungait et al., 2012; Poeplau and Don, 2013). This pooling approach obscures critical information about the vertical distribution of carbon and the depth-specificity of land-use effects. Guo and Gifford (2002) conducted a landmark meta-analysis of soil carbon stocks and land-use change, but their analysis could not evaluate whether carbon losses were concentrated in surface horizons or distributed throughout the profile because primary studies reported only pooled totals.

In the Nkanu East Watershed, preliminary evidence indicates that two locations—Nkerefi (Ihukulu) and Ugbawka (Amauzam)—differ substantially in texture, acidity, and organic carbon levels despite similar climate and land-use histories. The Typic Paleustult soils at Nkerefi are more strongly acidic (pH ~4.5–5.0) and clay-rich (sandy clay loam), while the Ugbawka soils are moderately acidic (pH ~5.0–5.7) and sandier (sandy loam) (Anikwe, 2010; Oguike and Mbagwu, 2009). This study addresses the following research gap: there is insufficient horizon-level data on organic matter stratification in Nigerian agroecosystems, constraining both the accuracy of national carbon inventories and the development of depth-sensitive management recommendations.

This study makes four concrete contributions: (i) first horizon-level (A/B/C) assessment of SOC stratification and stocks across forest-cultivation gradients in Nkanu East Watershed; (ii) demonstration that subsoil (B and C horizon) carbon contributes 51–73% of total profile stocks, challenging surface-centric management paradigms; (iii) quantification of location-specific land-use effects mediated by three-way interactions between location, land use, and horizon; and (iv) agroecological management recommendations grounded in depth-resolved soil health indicators.

2 Materials and methods

2.1 Study area and site selection

2.1.1 Geographic, climatic, and geologic setting

The study was conducted in Nkanu East Local Government Area (LGA) of Enugu State, southeastern Nigeria, within the Nkanu East Watershed (approximately 6°30′N, 7°30′E; elevation ~233 m above sea level). The area falls within the tropical humid climatic zone, characterized by a bimodal rainfall distribution spanning April–July and September–November, with total annual precipitation ranging from 1,700 to 2,060 mm (Lal, 2004). Mean air temperature varies seasonally from 27 °C during the harmattan period (December–February) to 31 °C in the dry season, while relative humidity fluctuates between 60% in the dry season and 80% during the rainy periods (Lal, 2008). The LGA covers approximately 804.6 km² with a population density that places considerable pressure on agricultural land (Brady and Weil, 2017).

Geologically, the soils are derived from sedimentary shale parent material of the Nanka Sands formation, weathered under high-temperature, high-rainfall conditions typical of the region (Brady and Weil, 2017). Soils in the watershed are classified as Typic Paleustult (USDA Soil Taxonomy) (Jobbágy and Jackson, 2000), representative of highly weathered, acidic Ultisols common to the southeastern Nigerian lowlands.

2.1.1.1 Precipitation patterns and anthropogenic amendments

The bimodal rainfall distribution (1,700–2,060 mm yr^-1) concentrates erosive energy during April–July and September–November, with a pronounced dry season (December–March) that limits biological activity and organic matter decomposition. No mineral fertilizers, organic amendments, or liming materials were applied to the cultivated plots during the study period; management consisted of manual tillage, crop residue removal for livestock fodder and fuel, and continuous monoculture or simple rotation of cassava, yam, maize, cowpea, rice, and sweet potato without fallow periods.

2.1.2 Site selection: contrasting pedogenic contexts

Two towns within the watershed were purposively selected to capture contrasting pedogenic environments: Nkerefi (Ihukulu area) and Ugbawka (Amauzam area). Site selection followed a stratified purposive approach based on reconnaissance soil surveys, land use history documentation, and accessibility. Nkerefi (Ihukulu) represents a lower-elevation site (~87 m a.s.l.) with finer-textured soils (sandy clay loam) and stronger pedogenic development, whereas Ugbawka (Amauzam) occupies a higher topographic position (~117–165 m a.s.l.) with sandier soils (sandy loam) and more rapid internal drainage (Batjes, 1996).

2.1.3 Land use definitions and criteria

At each location, two land use types were identified: Continuously Cultivated Land (CCL) and Forested Land (FL). CCL plots were defined as areas under annual or near-annual cultivation for 40–50 years (minimum four to five decades), involving continuous cropping of cassava (Manihot esculenta), yam (Dioscorea spp.), maize (Zea mays), cowpea (Vigna unguiculata), rice (Oryza sativa), and sweet potato (Ipomoea batatas) without fallow periods, organic amendment, or conservation tillage. FL plots were defined as areas supporting natural secondary forest or long-term fallow vegetation (>50 years), characterized by closed canopy cover with species including Elaeis guineensis (oil palm), mixed shrubs, indigenous trees, herbs, and grasses. The FL plots served as reference baselines representing the pre-cultivation soil condition, acknowledging that these are secondary forests rather than primary old-growth vegetation (Six et al., 2002).

2.2 Soil sampling and horizon identification

2.2.1 Plot establishment and profile pit excavation

Soil sampling was conducted in April 2023 during the early rainy season to ensure consistent moisture conditions across sites. At each of the four location–land use combinations, a 20 m × 20 m (0.04 ha) representative plot was demarcated. Geographical coordinates, elevation, and slope were recorded using a handheld GPS device. Within each plot, two replicate soil profile pits measuring 1.5 m × 1.5 m in horizontal dimension and extending to a maximum depth of 1.5 m or to the point of impenetrable parent material were excavated manually. The replicate pits were spaced a minimum of 10 m apart within each plot to capture within-site spatial variability. A total of eight profile pits were excavated across the entire study (2 locations × 2 land uses × 2 replicate profiles).

Sample size limitation: The study design yielded n = 2 profiles per location–land use combination (n = 8 profiles total). This limited replication constrains statistical power and increases uncertainty regarding reported significant differences. All statistical results should be interpreted as indicative patterns requiring confirmation with expanded replication (n ≥ 4).

2.2.2 Horizon-based sampling protocol

This study employed a pedogenetic horizon-based sampling approach rather than the conventional fixed-depth increment method. Horizon-based sampling was chosen because diagnostic soil horizons (A, B, C) represent natural boundaries of carbon distribution, nutrient transformation, and structural organization that fixed-depth increments may arbitrarily bisect (Schmidt et al., 2011).

At each profile pit, master soil horizons were identified and delineated in situ based on the following diagnostic criteria (Nair et al., 2021):

2.2.2.1 A horizon (topsoil)

The uppermost mineral horizon, characterized by darker coloration (higher organic matter content), granular or sub-angular blocky structure, and the highest biological activity. Depth ranged from 0 to 20–52 cm across profiles.

2.2.2.2 B horizon (subsoil)

The zone of maximum illuviation, distinguished by brighter color (higher chroma), stronger structure development (sub-angular to angular blocky), and clay accumulation. Upper boundaries ranged from 20 to 52 cm; lower boundaries extended to 74–145 cm.

2.2.2.3 C horizon (parent material transition)

The weathered parent material horizon showing little pedogenic alteration, with rock structure partially preserved and minimal organic matter. Upper boundaries ranged from 74 to 145 cm depth.

Horizon boundaries were marked in the field, and each horizon was described for color (Munsell notation), texture class, structure, consistence, and root density.

2.2.3 Sampling depths, thickness measurements, and imputation protocol

Within each identified horizon, disturbed composite samples were collected for chemical analysis by combining multiple sub-samples along the exposed face. Undisturbed core samples were taken using stainless steel cylinders (100 cm³ volume, 5.1 cm diameter, 5.0 cm height) for bulk density determination.

The actual thickness of each horizon was measured in situ to the nearest centimeter. Horizon thickness varied substantially among profiles (A: 20–52 cm; B: 42–94 cm; C: 15–28 cm), reflecting genuine pedogenic heterogeneity. For six cultivated profile samples (25% of the dataset) where explicit depth documentation was compromised during fieldwork, horizon midpoints and thickness values were imputed using horizon-specific defaults derived from the observed distribution across all profiles: A horizon midpoint = 15 cm, B = 50 cm, C = 120 cm; default thicknesses of A = 25 cm, B = 60 cm, C = 25 cm were applied for carbon stock calculations only (Oguike and Mbagwu, 2009).

2.2.3.1 Imputation sensitivity

A sensitivity analysis was conducted to evaluate the influence of imputed values on SOC stock calculations. Varying imputed horizon thicknesses by ±10% changed total profile stocks by approximately 2–4%—modest relative to the 44% between-location difference. This suggests that imputation uncertainty does not substantially alter the principal conclusions.

A total of 48 horizon-level samples were collected (2 locations × 2 land uses × 2 replicate profiles × 3 horizons × 2 sampling types [disturbed + undisturbed]), with four C-horizon physical samples excluded from bulk density analysis due to core damage during extraction from the compacted subsoil.

2.3 Laboratory analysis

All soil analyses were conducted at the Research Laboratory of the Department of Soil Science, Federal University of Agriculture, Umudike, Nigeria. Analytical procedures followed established protocols as summarized in Table 1.

Table 1. Summary of laboratory analytical methods.

Walkley–Black correction factor clarification: The 1.30 correction factor was applied to all Walkley–Black organic carbon determinations to account for incomplete oxidation (approximately 77% recovery). This replaces the 1.33 factor indicated in earlier versions of Table 1. The correction factor was applied consistently across all samples to ensure relative comparability; absolute values may still underestimate true SOC by 14–23%. Soil organic matter (SOM) content was derived from organic carbon using the conventional van Bemmelen factor: SOM (%) = OC (%) × 1.724. This factor assumes that soil organic matter contains approximately 58% organic carbon by mass (Guo and Gifford, 2002).

2.4 Carbon stock calculation by horizon

2.4.1 Horizon-specific SOC stock formula

Soil organic carbon stock was computed at the horizon level rather than by fixed-depth increments. The horizon-specific carbon stock (SOCstock,i; kg C ha^-1) for the i-th horizon was calculated as:

SOCstock,i=10,000×hi×ρb,i×OCi

where hi is the horizon thickness (m), ρb,i is the bulk density (Mg m^-3, equivalent to g cm^-3), and OCi is the organic carbon concentration (kg kg^-1, equivalent to % × 0.01). The factor 10,000 converts the product to kg C ha^-1 (1 ha = 10,000 m²). For computational convenience, the equivalent form used was:

SOCstock,i(kg C ha−1)=1,000×hi(cm)×ρb,i(g cm−3)×OCi(%)

2.4.2 Profile summation to 0–130 cm depth

Total profile SOC stock to 130 cm depth was obtained by summation across the three diagnostic horizons:

SOCprofile=∑SOCstock,i (i=A,B,C)

The 130 cm depth was selected as the maximum reliable sampling depth across all profiles; where the C horizon extended beyond 130 cm, the stock was calculated for the actual sampled thickness and then proportionally adjusted to the 130 cm reference depth. Profile-level stocks were computed separately for each of the two replicate pits, and the mean of the two replicates was used as the experimental unit for statistical comparison.

2.4.3 Equivalent soil mass considerations and imputation for missing data

For four C-horizon samples where bulk density could not be measured due to core damage during extraction from compacted subsoil, ρb was imputed as the mean of the A and B horizon bulk densities for that same profile. This linear interpolation assumes a gradual increase in bulk density with depth, consistent with the compaction gradients observed in the measured profiles (Dexter, 1988).

The equivalent soil mass (ESM) approach was considered but not applied as the primary metric. The ESM approach corrects for bulk density changes that accompany land use conversion (Rumpel et al., 2020), but in this study, the horizon-based framework inherently accounts for depth–density interactions by using measured horizon thicknesses.

2.5 Statistical analysis

All statistical analyses were performed using SAS software (version 9.4, SAS Institute, Cary, NC). A three-way analysis of variance (ANOVA) was conducted using the General Linear Model procedure (PROC GLM) with the following factorial structure:

Yijkl=μ+Li+Uj+Hk+(LU)ij+(LH)ik+(UH)jk+(LUH)ijk+ϵijkl

where Yijkl is the response variable, μ is the overall mean, Li is the location effect (i = Nkerefi, Ugbawka), Uj is the land use effect (j = CCL, FL), Hk is the horizon effect (k = A, B, C), (LU)ij, (LH)ik, (UH)jk are two-way interaction terms, (LUH)ijk is the three-way interaction, and ϵijkl is the residual error. The mean of the two replicate profiles served as the observation unit.

2.5.1 ANOVA assumptions

Normality of residuals was assessed using the Shapiro-Wilk test, and homogeneity of variances was evaluated using Levene’s test. Where assumptions were violated, data were log-transformed prior to analysis. Significance was assessed at α = 0.05.

Where ANOVA indicated significant main effects or interactions, Tukey’s Honestly Significant Difference (HSD) test was applied for pairwise mean comparison. Tukey’s HSD controls the family-wise error rate across all pairwise comparisons.

Pearson product-moment correlation coefficients (r) were computed among all pairs of soil physicochemical variables using PROC CORR. Where linear relationships were indicated, simple linear regression equations were fitted using PROC REG.

A horizon-level nutrient depletion ratio (NDR) was computed for each element as the ratio of the A-horizon concentration to the C-horizon concentration: NDR = CA-horizon/CC-horizon. An NDR < 1.0 indicates subsoil enrichment; NDR > 1.0 indicates surface enrichment.

3 Results

3.1 Soil physical properties by horizon

3.1.1 Bulk density stratification

Bulk density (BD) varied with land use, horizon, and their interaction. At Nkerefi, cultivation increased A-horizon BD from 1.33 ± 0.01 ag cm^-3 under forest to 1.75 ± 0.03 ag cm^-3 under cultivation, a 31.6% increase. The B and C horizons showed attenuated but comparable patterns: cultivated BD of 1.69 ± 0.00 and 1.66 ± 0.03 ag cm^-3 exceeded forest values (1.36 ± 0.01 and 1.38 ± 0.00 ag cm^-3) by 24.3% and 20.3%, respectively. Three-way ANOVA confirmed a highly significant land use effect on BD (F = 16.99, df = 1, p < 0.001), strongly moderated by location (FLoc×LU = 17.62, df = 1, p < 0.001) and the land use × horizon interaction (FLU×Hor = 4.16, df = 2, p < 0.05). The three-way interaction was also highly significant (F = 156.75, df = 1, p < 0.001).

At Ugbawka, the A horizon showed the opposite trend: cultivated BD (1.36 ± 0.00 ag cm^-3) was marginally lower than forest BD (1.38 ± 0.01 ag cm^-3). This reversal likely reflects textural differences—Nkerefi cultivated profiles contained 34.8% clay in the C horizon versus 7.5–9.1% at Ugbawka—predisposing Nkerefi soils to structural degradation under tillage.

3.1.2 Porosity, moisture content, and hydraulic conductivity

Total porosity remained stable across all treatments (47.0–50.3%; Table 2), with no significant main or interaction effects (all F < 0.05, p > 0.89). This apparent paradox—higher BD without lower porosity—arises because particle density also increased under cultivation, and because cultivation likely shifted pore size distribution (reducing macroporosity, increasing microporosity) rather than eliminating total pore space. Moisture content ranged from 8.9% (Ugbawka Forest, C horizon) to 16.1% (Ugbawka Forest, A horizon) but showed no significant treatment effects. Saturated hydraulic conductivity (Ksat) was uniformly low (0.015–0.049 cm h^-1), typical of fine-textured tropical soils. Nkerefi Forest exhibited the highest Ksat (0.044–0.049 cm h^-1), while Ugbawka showed the lowest (0.015–0.026 cm h^-1; FLoc = 45.2, df = 1, p < 0.001).

Table 2. Soil physical properties by location, land use, and diagnostic horizon.

Values are mean ± SD (n = 2 profiles). BD, bulk density; MC, moisture content; Ksat, saturated hydraulic conductivity. Large porosity SDs reflect high between-profile variability at n = 2.

3.1.2.1 Particle density considerations

The increase in bulk density under cultivation at Nkerefi reflects compaction of the solid soil matrix and reduction in total pore space, not changes in particle density. Particle density (the mass of soil solids per unit volume of solids) is an intrinsic property of soil minerals and organic matter and does not change with cultivation. The apparent stability of total porosity despite increased BD arises from concurrent changes in soil structure and pore size distribution, not from altered particle density.

3.2 Organic carbon and organic matter stratification

3.2.1 Horizon-level concentration patterns

Organic carbon (OC) and organic matter (OM) exhibited consistent surface enrichment across all four location–land use combinations, with highest concentrations in the A horizon declining through B and C horizons (Figure 1). Ugbawka Cultivated displayed the highest surface enrichment: A-horizon OC reached 2.26 ± 0.90% and OM reached 4.12 ± 1.64%, substantially exceeding all other treatments. Ugbawka Forest also showed elevated A-horizon OC (1.74 ± 0.36%) and OM (3.04 ± 0.87%). At Nkerefi, both forest and cultivated A horizons showed lower OC (1.25 ± 0.04% and 1.22 ± 0.10%, respectively), with minimal land use difference at the surface.

3.2.2 OC decline patterns: 37.7–52.8% A→C decline

The magnitude of A→C OC decline varied substantially between locations and land uses (Table 3). Overall declines ranged from 37.7% (Nkerefi Cultivated) to 52.8% (Ugbawka Cultivated). At Nkerefi, the decline trajectories were similar between forest and cultivated plots (45.4% versus 37.7% A→C), suggesting cultivation has not fundamentally altered the vertical carbon distribution. At Ugbawka, the larger decline under cultivation (52.8%) reflects the exceptionally high A-horizon OC (2.26%) combined with B and C-horizon values comparable to or lower than forest equivalents.

Table 3. Organic carbon (OC) and organic matter (OM) stratification by location, land use, and horizon.

Values are mean ± SD (n = 2 profiles). Decline A→C = [(OC-A − OC-C)/OC-A] × 100.

3.2.3 Decline rates and anomalous B→C patterns at Ugbawka

A→B decline rates ranged from 0.572% OC/cm (Nkerefi Cultivated) to 1.044% OC/cm (Ugbawka Cultivated). At Nkerefi, OC continued declining from B to C at 0.290–0.339% OC/cm. At Ugbawka, however, B→C decline rates were negative (−0.131% OC/cm Forest; −0.091% OC/cm Cultivated), indicating C-horizon OC marginally exceeded B-horizon values.

This B→C reversal at Ugbawka departs from the monotonic decline model. Three mechanisms may explain this anomaly: (i) organic illuviation—downward transport of dissolved or colloidal organic compounds may enrich deeper horizons; Ugbawka’s higher clay content in B and C horizons (up to 33.8%) provides greater adsorption surface for leached carbon; (ii) parent material effects may contribute if the C horizon contains material with inherently higher organic carbon than the overlying B horizon; (iii) pedogenic boundary effects: the B horizon at Ugbawka corresponds to a zone of intense clay accumulation that may dilute carbon, while the C horizon retains more original carbon stock.

3.3 Carbon stock distribution by horizon

3.3.1 Horizon-specific SOC stocks

Horizon-specific SOC stocks were calculated as SOCstock = 1000 × thickness (cm) × BD (g cm^-3) × OC (%). Stocks varied substantially across horizons, locations, and land uses (Table 4; Figure 2).

Table 4. Soil organic carbon stocks by horizon and total profile stocks.

SOC stock = 1000 × thickness (cm) × BD (g cm^-3) × OC (%). Mean ± SD (n = 2 profiles).

At Nkerefi Forest, the A-horizon stock (86,862 ± 1,148 kg C ha^-1) was the highest single-horizon contribution, representing 49.4% of the total 0–130 cm profile. At Nkerefi Cultivated, A-horizon stock fell to 50,885 ± 3 kg C ha^-1 (27.1% of total), reflecting a thinner A horizon (23.9 cm versus 52.2 cm) and higher BD. The B horizon dominated the Nkerefi Cultivated profile at 108,292 ± 21,219 kg C ha^-1 (57.6% of total), owing to substantial thickness (69.6 cm).

At Ugbawka, B-horizon stocks were the largest absolute pools observed: 121,359 ± 7,378 kg C ha^-1 under forest and 136,905 ± 68,446 kg C ha^-1 under cultivation, reflecting greater horizon thickness (83.3 cm and 94.2 cm) and sustained moderate OC concentrations.

3.3.2 Total profile stocks and location vs. land use effects

Total profile SOC stocks ranged from 175,796 kg C ha^-1 (Nkerefi Forest) to 253,524 kg C ha^-1 (Ugbawka Forest). The location effect was pronounced: Ugbawka stored 44% more carbon than Nkerefi on average. The land use effect was inconsistent: at Nkerefi, cultivated stocks (187,871 kg C ha^-1) marginally exceeded forest stocks (175,796 kg C ha^-1; +6.9%), while at Ugbawka, cultivated stocks (240,447 kg C ha^-1) were 5.2% lower than forest (253,524 kg C ha^-1). At Nkerefi, the compensatory effect of a thicker, denser B horizon under cultivation offset lower OC concentrations.

The B horizon was the dominant carbon reservoir in cultivated profiles (57.6% at Nkerefi, 56.9% at Ugbawka). Combined B + C horizon stocks accounted for 72.9%, 62.2%, 72.7%, and 62.3% of total profile stocks at Nkerefi Cultivated, Nkerefi Forest, Ugbawka Cultivated, and Ugbawka Forest, respectively. The 0–60 cm depth interval captured 74–88% of total 0–130 cm stocks, leaving 12–26% below 60 cm.

3.4 Chemical properties and nutrient dynamics by depth

3.4.1 pH stratification

Soil pH showed strong location-dependent stratification (Figure 3). At Nkerefi, pH was moderately acidic (4.50–5.03), with the cultivated C horizon recording the lowest value (4.50 ± 0.09). At Ugbawka, pH was systematically higher (FLoc = 16.25, df = 1, p < 0.001), ranging from 4.76 to 5.74 ± 0.01. The location × horizon interaction was highly significant (F = 11.38, df = 2, p < 0.001), and the three-way interaction was also significant (F = 91.77, df = 1, p < 0.001).

3.4.2 Nitrogen and phosphorus distribution

Total nitrogen (N) followed OC’s depth pattern, with A-horizon N ranging from 0.11 ± 0.00% (Nkerefi) to 0.14 ± 0.01% (Ugbawka). Three-way ANOVA confirmed highly significant horizon (F = 16.40, df = 2, p < 0.001) and location (F = 7.85, df = 1, p < 0.05) effects, with a highly significant three-way interaction (F = 66.82, df = 1, p < 0.001). The strong OC–N correlation (r = 0.778, p < 0.001) confirms expected C:N coupling.

Available phosphorus (P) displayed an inverse depth pattern, increasing from A to C in forest soils. At Nkerefi Forest, P rose from 14.25 ± 0.71 mg kg^-1 (A) to 31.30 ± 0.57 mg kg^-1 (C); at Ugbawka Forest, from 17.66 ± 0.23 to 36.22 ± 3 mg kg^-1. Under cultivation, P peaked in the B horizon at Nkerefi (39.35 ± 0.71 mg kg^-1) rather than the C horizon.

3.4.3 Exchangeable bases, CEC, and base saturation

Exchangeable calcium (Ca) and magnesium (Mg) showed strong depth stratification (Table 5). At Nkerefi Forest, both increased markedly with depth: Ca rose from 3.04 ± 0.06 to 6.50 ± 0.10 cmol kg^-1; Mg from 1.40 ± 0.00 to 3.67 ± 0.04 cmol kg^-1. At Ugbawka Cultivated, the A horizon was strongly enriched in both Ca (4.76 ± 0.10 cmol kg^-1) and particularly Mg (4.07 ± 0.31 cmol kg^-1)—the highest Mg observed—suggesting substantial amendment inputs.

Table 5. Chemical properties by location, land use, and diagnostic horizon.

Values are mean ± SD (n = 2 profiles). N, total nitrogen; P, available phosphorus (Bray-1); Ca, exchangeable calcium; Mg, exchangeable magnesium; ECEC, effective cation exchange capacity; BS, base saturation.

3.4.4 Nutrient depletion ratios

The nutrient depletion ratio (A-horizon value/C-horizon value) quantifies the direction and magnitude of depth redistribution (Table 6). At Nkerefi Forest, Ca (0.47), Mg (0.38), P (0.46), and ECEC (0.50) all showed ratios well below 1.0, confirming strong subsoil enrichment. Cultivation at Nkerefi partially reversed this pattern: Ca increased from 0.47 to 0.56, Mg from 0.38 to 0.78, and P from 0.46 to 0.87.

Table 6. Nutrient depletion ratios (A-horizon/C-horizon) by location and land use.

Ratios = A-horizon mean/C-horizon mean. Values < 1.0 indicate subsoil enrichment; values > 1.0 indicate surface enrichment.

At Ugbawka Forest, Ca (1.23), Mg (1.73), and BS (1.74) already exceeded 1.0, indicating surface enrichment under native vegetation. Cultivation dramatically amplified this: Ca rose to 2.09, Mg to 4.93, and ECEC to 2.81—the highest ratios observed.

3.5 ANOVA results and correlations

Three-way ANOVA (Location × Land Use × Horizon) partitioned variance across all factors and interactions (Table 7). Organic carbon showed significant location (F = 5.33, df = 1, p < 0.05) and horizon (F = 9.84, df = 2, p < 0.001) effects, but no land use main effect (F = 0.28, df = 1, p = 0.61). The location × horizon interaction was highly significant (F = 10.85, df = 2, p < 0.001), reflecting different depth-distribution patterns between Nkerefi (monotonic decline) and Ugbawka (non-monotonic B→C reversal).

Table 7. Three-way ANOVA summary for selected soil properties.

F-values with degrees of freedom (1, 2) for main effects and interactions. Significance: *p < 0.05, p < 0.01, *p < 0.001, ns, not significant.

Pearson correlation analysis across all 24 horizon samples revealed: OC–OM (r = 0.998, p < 0.001), OC–N (r = 0.778, p < 0.001), OC–pH (r = 0.688, p < 0.001), and BS–OC (r = 0.433, p < 0.05). Bulk density showed a weak negative correlation with OC (r = −0.274, p = 0.21).

4 Discussion

4.1 Organic matter stratification in tropical soils

The consistent A-horizon enrichment of organic carbon across all treatments confirms that tropical soil carbon is surface-concentrated, driven by litterfall, root exudation, and biological turnover (Brady and Weil, 2017; Jobbágy and Jackson, 2000). Nkerefi A-horizon OC (1.22–1.25%) falls within the range reported for degraded tropical Ultisols under low-input management, whereas Ugbawka Cultivated reached 2.26%—likely reflecting sustained organic inputs from crop residue retention or manure application (Nair et al., 2021; Schmidt et al., 2011).

The depth-decline rates (0.572–1.044% OC per cm, A→B) exceed the global tropical average of approximately 0.3–0.6% OC per cm (Jobbágy and Jackson, 2000), suggesting rapid carbon attenuation in these southeastern Nigerian soils. The steepest decline at Ugbawka Cultivated (1.044% OC/cm) creates a sharply truncated profile where carbon is concentrated in the layer most vulnerable to erosion and oxidation (Blanco-Canqui and Lal, 2008; Dungait et al., 2012).

The total A→C decline magnitude (37.7–52.8%) and A-horizon contribution to total stocks (27.1–49.4%) are broadly consistent with Batjes (2016), who estimated that the upper 30 cm of tropical soils contain 50–60% of profile carbon. The cultivated soils’ lower A-horizon percentages (27.1–27.3%) reflect A-horizon thinning under tillage rather than subsoil enrichment.

The B→C reversal at Ugbawka challenges the monotonic decline model. Organic illuviation provides the most plausible explanation: downward transport of dissolved organic carbon may enrich the C horizon, particularly where Ugbawka’s clay accumulation zone (up to 33.8% clay) creates a textural discontinuity that promotes organo-mineral complexation (Brady and Weil, 2017; Kögel-Knabner et al., 2008). This interpretation aligns with Dungait et al. (2012), who argued that SOM persistence is governed by spatial accessibility to decomposers rather than intrinsic recalcitrance.

4.2 Land use effects on depth-distributed soil properties

The 31.6% increase in A-horizon bulk density at Nkerefi approaches the 1.8 ag cm^-3 threshold above which root penetration is critically restricted in clay-rich tropical soils (Blanco-Canqui and Lal, 2008; Dexter, 1988). Critically, compaction extended through the B and C horizons (24.3% and 20.3% increases), indicating that decades of continuous tillage have degraded the entire profile at this site.

At Ugbawka, cultivated A-horizon BD (1.36 ag cm^-3) was marginally lower than forest (1.38 ag cm^-3), illustrating texture’s mediating role. Nkerefi cultivated profiles contained up to 34.8% clay versus 7.5–9.1% at Ugbawka; sandy loam textures resist compaction because sand particles provide structural scaffolding (Brady and Weil, 2017). The absence of significant porosity differences (~48–50% across all treatments) indicates that cultivation shifted pore architecture—reducing macropore continuity—rather than eliminating total pore volume (Dexter, 1988).

4.2.1 Why total SOC stocks did not consistently decline under cultivation

Two compensatory mechanisms explain this. First, B-horizon thickening under cultivation increased carbon-bearing soil mass: at Nkerefi, the cultivated B horizon was 69.6 cm thick versus 63.2 cm under forest; at Ugbawka, 94.2 cm versus 83.3 cm. Combined with higher BD, this thicker B horizon compensated for lower OC concentrations. Second, the cultivated sites likely received historical organic inputs that maintained surface carbon. The Ugbawka Cultivated A-horizon OC of 2.26% strongly suggests residue retention or manure application (Altieri, 2018; Wezel et al., 2009).

4.2.2 However, this conclusion should be interpreted with caution

The observed SOC stocks may partly reflect differences in bulk density and horizon thickness rather than true carbon accumulation. The forested plots, being secondary regrowth, may never have accumulated the large carbon pools expected of mature tropical forest (Guo and Gifford, 2002). Concentration-based comparisons would conclude that cultivation reduces carbon; stock-based comparisons incorporating horizon architecture reveal a more nuanced picture—but one that requires temporal monitoring to distinguish steady state from transient dynamics.

4.3 Agroecological implications

The B horizon was the largest carbon reservoir in cultivated soils (108,292–136,905 kg C ha^-1; 56–58% of total profile), challenging the topsoil-only paradigm dominant in sub-Saharan African soil health assessment (Altieri, 2018; Minasny et al., 2017). The strong OC–N correlation (r = 0.778) indicates that carbon management is simultaneously nitrogen management throughout the profile, and the OC–pH correlation (r = 0.688) suggests organic matter mediates acidity buffering at depth (Schmidt et al., 2011).

4.3.1 Location-specific management recommendations

4.3.1.1 Nkerefi (Ihukulu)

The strongly acidic pH (4.50–5.03), severe compaction (BD 1.75 ag cm^-3), and low base saturation (55.8–75.4%) define this as a degradation hotspot. Recommended interventions include: (i) agricultural lime at 2–4 t ha^-1 yr^-1 to raise pH above 5.5; (ii) compost or manure incorporation at 5–10 t ha^-1 yr^-1 to reduce BD and increase surface carbon; (iii) deep-rooted cover crops (Mucuna pruriens, Crotalaria ochroleuca) during fallow to penetrate compacted B horizons; (iv) cassava–maize rotation with leguminous cover crops in a 2:1 ratio; and (v) contour bunding to protect the thin A horizon (23.9 cm under cultivation).

4.3.1.2 Ugbawka (Amauzam)

The higher surface carbon (2.26% OC), lower BD (1.36 ag cm^-3), and less acidic pH indicate partially effective current management. Recommended interventions include: (i) continued crop residue retention and compost use; (ii) agroforestry trees (Gliricidia sepium, Leucaena leucocephala) to channel carbon to the B horizon; (iii) deep-placement Ca, Mg, and P fertilization at 30–50 cm to address the steep A-to-B nutrient gradient; and (iv) avoidance of deep ploughing that would disturb the B horizon and oxidize its substantial carbon stock (136,905 kg C ha^-1).

These recommendations are proposed as future management perspectives rather than direct conclusions from this study, as the practices were not experimentally evaluated herein.

4.4 Methodological considerations and limitations

4.4.1 Sample size constraints (n = 2 profiles per treatment)

The small sample size limits statistical power. Large standard deviations (e.g., Ugbawka Cultivated B-horizon stock: 136,905 ± 68,446 kg C ha^-1) reflect high between-profile variability. The consistent three-way interaction structures (F = 4.44–156.75, p < 0.01) across multiple variables provide some reassurance that observed patterns are robust, but the absence of significant land-use main effects for OC and pH may partly reflect insufficient power.

4.4.2 Walkley–black method considerations

The 1.30 correction factor was applied consistently to account for approximately 77% oxidation recovery. Absolute values may underestimate true SOC by 14–23%, but relative comparisons remain valid. The near-perfect OC–OM correlation (r = 0.998) supports internal consistency. If cultivation altered SOM chemical composition, oxidation efficiency may differ between land uses (Six et al., 2002); dry combustion or POM/MAOM fractionation would provide more mechanistic insight.

4.4.3 Imputation sensitivity

C-horizon bulk density was imputed for four profiles as the mean of A and B horizon values. Sensitivity analysis indicates a ±10% BD error would change total stocks by approximately 2–4%—modest relative to the 44% between-location difference. Horizon boundary subjectivity adds further uncertainty, though consistent identification by a single experienced pedologist minimized observer bias.

4.4.4 Representativeness

The investigation was conducted at only two locations within a single watershed. Conclusions should be restricted to the investigated environments and should not be generalized to all tropical soils or southeastern Nigeria.

4.4.5 Reference dated 2026

Anikwe et al. (2026) is a companion manuscript currently under review at Frontiers in Soil Science that presents complementary ecosystem services assessment from the same watershed. The manuscripts share authorship and study sites but report distinct datasets and research questions; there is no overlap in primary data or conclusions.

5 Conclusion

This study demonstrates that horizon-level assessment reveals depth-specific land use effects that pooled-depth approaches systematically obscure in tropical Ultisols. Subsoil horizons (B/C) contain 51–73% of total profile SOC stocks, with the B horizon alone storing 56–58% of carbon in cultivated soils—exceeding the A-horizon contribution by a factor of two at Ugbawka. Location effects dominated over land use effects: Ugbawka stored 44% more carbon than Nkerefi, and cultivation increased bulk density by 31.6% at Nkerefi but had no comparable effect at Ugbawka, where sandy loam texture resisted compaction. These findings indicate that surface-only sampling protocols miss the majority of profile carbon stocks and that agroecological management in tropical watersheds must be calibrated to site-specific soil conditions through whole-profile assessment.

6 Future research directions

6.1 Temporal monitoring

Establish a replicated temporal monitoring network (n ≥ 4 profiles per treatment) tracking SOC stocks at 3- to 5-year intervals using equivalent soil mass correction to distinguish concentration changes from bulk-density artefacts.

6.2 Root carbon contributions

Quantify root biomass distribution and rhizodeposition rates by horizon using isotope-based partitioning (δ¹³C or ¹⁴C labelling) to resolve whether subsoil carbon is sustained by downward transport or in situ root inputs.

6.3 Intervention trials

Evaluate the efficacy of deep-rooting cover crops (Mucuna pruriens, Crotalaria spp.), biochar incorporation to 40–60 cm, and reduced tillage depth in protecting B-horizon carbon at both Nkerefi and Ugbawka to test whether the same intervention produces different outcomes across location-specific edaphic templates.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Stéphane Cordeau, UMR Agroécologie, France

Reviewed by: Amit Anil Shahane, Central Agricultural University, India

Atika Mouaddine, Teacher éducation nationale, Morocco