Work

Data-driven reconstruction of soil neutralization curves for lime requirement prediction in Sub-Saharan African soils

Year: 2026

Type: article

Abstract: Reliable lime requirement (LR) estimation is essential for managing soil acidity in Sub-Saharan Africa, where highly weathered soils exhibit strong variability in buffering capacity and neutralization behavior. This study compared complementary approaches for predicting soil neutralization curves and LR from a multi-country dataset of acidic soils. A parametric titration–incubation conversion (TIC) framework was developed to reconstruct long-term incubation-equivalent curves from short-term titration responses. In parallel, machine-learning (ML) models were built from a minimal chemical signature (CSM; soil pH and soil organic matter), visible–near-infrared spectral signatures (SSM-VisNIR), and mid-infrared spectral signatures (SSM-MIR). Their performance was benchmarked against conventional single-acidity-indicator equations from the literature (SAI-LIT) and against regionally recalibrated equations (SAI-CAL). The TIC framework successfully reconstructed incubation-equivalent responses from titration data, including under an independent country hold-out application. Among predictive approaches, SSM-MIR combined with Cubist achieved the highest accuracy, with R 2 = 0.94 and root mean square errors (RMSEs) of 0.18 pH units for titration and 0.28 pH units for incubation. The CSM also showed strong predictive ability despite relying only on routine soil analyses. In contrast, SAI approaches were generally weak and highly context dependent, although pH SMP retained partial predictive value. Overall, methods that reconstruct the full soil neutralization trajectory clearly outperformed static single-indicator equations. These findings support a shift from static, single-indicator recommendations to dynamic, curve-based decision tools tailored to different analytical capacities in Sub-Saharan Africa.

Source: Geoderma

DOI: 10.1016/j.geoderma.2026.117901

Authors: Ismail Kouera, Lotfi Khiari

Institutions: Université Mohammed VI Polytechnique, Université Laval

Language: English

Topic: Soil Management and Crop Yield

Subfield: Soil Science

Field: Agricultural and Biological Sciences

Domain: Life Sciences

SDGs: Zero hunger, Life on land

Open Access status: gold

Estimated APC paid: $3,320