The role of geochemistry in organic carbon stabilization in tropical rainforest soils

The following text represents the supplementary results and short discussion on geochemical soil parameters, SOC 10 stocks and stabilization assessed at valley positions of contrasting parent material geochemistry investigated in tropical rain mountain forest in the border region of the Congo and Nile basement. We show that, in valley positions, significant variation in the size of SOC stocks are related to changes in hydrological conditions and alluvial processes compared to non-valley positions. Geochemical differences in SOC stabilization mechanisms between regions remain in place, but are less prominent in valleys than at non-valley positions. 15


Supplemented information on valley positions
The following text represents the supplementary results and short discussion on geochemical soil parameters, SOC For valley positions, pyrophosphate extractable oxide mass (0.02 to 0.66 mass%) and oxalate extractable oxide mass (0.03 to 4.29 mass%) were low compared to DCB extractable oxide mass (0.11 to 10.79 mass%) (Fig. S1).
In general, valley positions showed a comparable picture compared to non-valley positions (Fig. 3, S1) with the exception that valley soils in the mixed sedimentary region had considerably lower amounts of pedogenic oxides at all soil depths (< 0.27 mass%) compared to valley soils in the mafic and felsic region.

Abundance of C fractions
In the felsic magmatic region, the abundance of microaggregate associated C and the ratio of m / s+c was more than twice as high in valley topsoils (m / s+c: 2.5 ± 0.4) compared to non-valley topsoils (m / s+c: 1.2 ± 0.5 to 1.4 ± 0.5) (Figs. 3a, S1a). Additionally, SOC>250 µm in topsoil of valley positions of the mixed sedimentary region was about threefold higher in valleys (22.2 ± 12.3 %) compared to the corresponding non-valley positions (6.5 ± 7.1 45 to 6.7 ± 4.0 %) (Figs. 4b, S2b), and the ratio of m / s+c exhibited a much higher variance in topsoils (Figs. 3a, S1ab).

Figure S2: (a) Δ 14 C across geochemical regions in valley positions (n = 1 per bar). (b) SOCbulk and fractions across geochemical regions in valley positions (n = 3 per bar). Letters indicate significant differences between geochemical
regions per soil layer. Asterisks indicate no significant differences in means (p > 0.05). In case of Δ 14 C values, error bars are smaller than symbols. cPOM -coarse particulate organic matter, m -waterstable microaggregates, s+c -free silt and clay fraction.  (Table S1). 65

Rotated principal component explained variance and loadings
For valley soils, five rPCs were determined (rPCv) explaining 92.7 % of the variance in the valley position subset 80 (Table A4, Fig. S3). rPC1v (Eigenvalue 10.7, explaining 38.2 % of the variance) represents solid phase mineralogy, in which total metal oxide concentration, total P and oxalate extractable oxides concentration had strong positive loadings ( ≥ 0.98), while total Si, pH / clay ratio and sand content had strong negative loadings ( ≤ -0.69). rPC2v, (Eigenvalue 9.65, explaining 34.5 % of the variance) represents the chemistry of the soil solution and silt content where CEC base saturation, pH and CEC base saturation / clay ratio showed strong positive 85 loadings ( ≥ 0.92) while CIA, SOCorganic and exchangeable acidity showed negative loadings (≤ -0.78). rPC3v (Eigenvalue 2.5, explaining 9 % of the variance) represents bioavailable P and Al weathering. Bioavailable P showed positive loadings (0.69), while Aldcb / Alt ratio showed negative loadings (-0.91). rPC4v (Eigenvalue 1.65, explaining 6 % of the variance) represents root input and exchangeable acidity, both having strong positive loadings (≥ 0.55). rPC5v (Eigenvalue 1.46, explaining 5 % of the variance) represents Fe weathering and litter C 90 stock in which both the Fedcb / Fet ratio and Litter C stock (SOClitter) had strong positive loadings ( ≥ 0.66).  angles between vectors display the degree of auto-correlation between variables. Small angles represent positive correlations and high degree of autocorrelation, diverging angles represent negative correlations and high degree of autocorrelation, high angles indicate no correlations between variables and/or rPCs.

Soil C stabilization driven by soil chemistry and parent material rather than topography
In contrast to our initial hypothesis that topography affects C stabilization in tropical forest soils through lateral material movements, we found no indication of this in our analysis (Table A1). This is supported by Drake et al.
(2019) who showed that C fluxes from pristine forest catchments within the same study area are characterized by 115 young C from organic layers, with no indication of erosion of older, mineral stabilized C. Similarly, Wilken et al. (2021), using plutonium tracers to analyze soil redistribution along forest hillslopes did not detect any signs of decadal erosional processes of soils for the same sites as used in this study. Even though material fluxes due to lateral water fluxes along slopes cannot be excluded over longer time scales (e.g. 100 to 1,000 years), our results pointed to no recent natural or anthropogenic induced erosional processes in the investigated sites. However, there 120 was a detectable effect on SOCbulk between positions that are affected by riverine and alluvial dynamics with regularly high water saturation (valleys) vs. well drained positions (non-valleys), with valleys generally having higher SOCbulk than their non-valley counterparts. Gleyic properties and fluvial sedimentation processes recorded in valley soils during profile description (Doetterl and Fiener, 2021) supported this interpretation. Overall, we found that in valley positions differences in SOC variables are rather related to changes in hydrological conditions 125 and alluvial processes than by chemical soil properties.