<![CDATA[Variations in land-use types and management intensity significantly alter soil physical-chemical properties and dictate carbon dioxide (CO2) emissions fluxes. This study aimed to model the control of soil physical-chemical characteristics on soil CO2 emission rates across various tropical mineral land-use types in Seginim District, South Bengkulu Regency. Utilizing an observational edaphic dataset across six dominant land-use types (corn, oil palm, natural forest, vegetables, coffee, and rubber) with three replications (N=18), advanced statistical modeling via Pearson correlation and multiple linear regression was performed. Soil CO2 emissions were measured in-situ using the modified closed-chamber method followed by alkali absorption titration, while edaphic variables (soil moisture, temperature, bulk density, and pH) were evaluated using standard laboratory procedures. Data were analyzed using Pearson correlation and multiple linear regression. The results revealed that intensive land-use types, namely corn (51.65 ± 2.31 kg/ha/day) and vegetables (43.80 ± 2.75 kg/ha/day), doubled CO2 emissions compared to natural forest (21.87 ± 1.36 kg/ha/day). Soil moisture and bulk density acted as the master variables controlling the emissions. Soil moisture alone showed a strong negative correlation, explaining 83.72% of the emission variance (r = -0.91; p < 0.01). Integrating all edaphic components into a multivariate regression model significantly increased the predictive power to 97.22% (R2 = 0.9722; p < 0.001), driven by the severe physical inhibition effect of bulk density (β = -30.83; p < 0.001). Conversely, soil temperature and pH exerted no significant partial effects due to microclimatic buffering. This study concludes that soil physical properties rigidly govern CO2 emissions across different tropical mineral land uses. Consequently, climate change mitigation strategies must prioritize protecting soil physical structures through water management and compaction prevention rather than chemical manipulation.]]>
Copyrights © 2026
