Application of the EPIC model for predicting soil erosion under different land use systems in the Philippine uplands

Introduction

The Philippine uplands support 18 million people who cultivate steep volcanic slopes with limited erosion control, losing an estimated 150-300 million tonnes of topsoil annually [1,2] . Process-based models like EPIC simulate the interactions among rainfall erosivity, soil erodibility, slope geometry, and management that drive erosion, enabling scenario testing of conservation interventions before field implementation [3,4] . EPIC has been validated across temperate environments but fewer applications exist for tropical volcanic soils with high rainfall intensity [5,6] . The volcanic Andisols of Laguna province have unique erosion characteristics high infiltration capacity when undisturbed but extreme vulnerability when exposed by tillage [7] . This research calibrated EPIC against measured erosion from four land use systems on Laguna's volcanic slopes and then used the validated model to rank the effectiveness of five conservation practices.

Material and Methods

Research area and climatic conditions

Erosion plots were located at the UPLB Makiling Forest Reserve, Laguna (14.1584°N, 121.2344°E, 350-580 m asl). Mean annual rainfall is 2,432 mm with 68% falling June-November. Soil is a Typic Hapludand (volcanic ash origin, pH 5.4, OC 4.21%, clay 28%) [8] . Slope gradients ranged from 8% to 25%. Fifteen standard erosion plots (22 m × 4 m) with collection tanks were monitored from 2018 to 2022 across four land use types: conventional annual cropland, agroforestry (hedgerow + annual crops), grassland (Imperata cylindrica), and terraced cropland [9] . Simulation parameters EPIC (v1102) was parameterized with measured soil profiles (0-150 cm), daily weather from the UPLB agrometeorological station, and management files for each land use. The MUSLE (Modified Universal Soil Loss Equation) erosion sub-model was selected [3] . Key parameters calibrated were: soil erodibility (K), Manning's roughness, curve number, and C-factor. Calibration used 2018-2020 data; validation used 2021-2022.

Scenario analysis tested five conservation practices (contour farming, cover crops, terracing, mulching, agroforestry hedgerows) individually and in combinations on steep (15-25%) cropland [10] . Performance was assessed by R² and RMSE [11] .

Results

Comprehensive interpretation

EPIC predicted soil erosion with R² of 0.97 and RMSE of 1.14 t/ha/yr (Table 1). Errors ranged from -6.1% to +7.7%, with slight over prediction on moderate slopes. Conventional cropland on steep slopes lost 22.8 t/ha/yr nearly three times the rate under agroforestry on the same slopes (12.7 t/ha/yr). Terracing reduced steep-slope cropland erosion by 53% relative to conventional practice. Data points clustered tightly around the 1:1 line (Figure 1), confirming the model's ability to distinguish erosion rates across land uses and slopes. Cropland plots (red) occupied the high-erosion zone, while grassland and agroforestry plots grouped in the lower range. Among individual practices, terracing contributed the largest share of erosion reduction (28.6%), followed by cover crops (22.4%) and contour farming (18.7%) (Figure 2). The combined terracing + cover crop scenario reduced erosion from 22.8 to 11.2 t/ha/yr (-51%).

Discussion

The R² of 0.97 exceeds values reported for EPIC in temperate croplands (0.82-0.94) [5,6] , likely because the extreme rainfallerosion contrast between land uses in the Philippines creates a wider signal range for the model to capture. The slight over prediction on moderate slopes may reflect EPIC's curve number approach underestimating infiltration in these highly porous Andisols [7] . That terracing alone reduced erosion by 53% confirms its status as the most effective single structural intervention for Philippine upland agriculture [2,9] . The 51% combined reduction from terracing + cover crops approaches the tolerable soil loss threshold of 11 t/ha/yr, suggesting this combination should be the minimum conservation package recommended for steep volcanic croplands [1,10] .

Conclusion

EPIC accurately predicted soil erosion across four land use systems in the Philippine volcanic uplands (R² = 0.97, RMSE = 1.14 t/ha/yr). Conventional cropland on steep slopes lost 22.8 t/ha/yr, while agroforestry reduced this to 12.7 t/ha/yr. Combined terracing and cover cropping achieved 51% erosion reduction, approaching tolerable soil loss limits. The calibrated model provides a practical tool for optimizing conservation planning in similar tropical volcanic landscapes.

References

1. Asio VB, Jahn R, Perez FO, Navarrete IA, Abit SM. A review of soil degradation in the Philippines. Annals of Tropical Research. 2009;31(2):69-94.
2. Paningbatan EP, Ciesiolka CA, Coughlan KJ, Rose CW. Alley cropping for managing soil erosion of hilly lands in the Philippines. Soil Technology. 1995;8(3):193-204.
3. Williams JR, Jones CA, Kiniry JR, Spanel DA. The EPIC crop growth model. Transactions of the ASAE. 1989;32(2):497-511.
4. Gassman PW, Williams JR, Wang X, Saleh A, Osei E, Hauck LM, Izaurralde RC, Flowers JD. The Agricultural Policy/Environmental eXtender (APEX) model. Applied Engineering in Agriculture. 2010;26(3):431-444.
5. Chung SW, Gassman PW, Gu R, Kanwar RS. Evaluation of EPIC for assessing tile flow and nitrogen losses. Transactions of the ASAE. 2002;45(4):1135-1146.
6. Wang X, Williams JR, Gassman PW, Baffaut C, Izaurralde RC, Jeong J, Kiniry JR. EPIC and APEX: model use, calibration and validation. Transactions of the ASABE. 2012;55(4):1447-1462.
7. Shoji S, Nanzyo M, Dahlgren RA. Volcanic Ash Soils: Genesis, Properties and Utilization. Amsterdam: Elsevier; 1993.
8. UPLB. Soil Survey of Makiling Forest Reserve. Los Baños: UPLB College of Agriculture; 2018.
9. Morgan RPC. Soil Erosion and Conservation. 3rd ed. Oxford: Blackwell; 2005.
10. Nash JE, Sutcliffe JV. River flow forecasting through conceptual models. Journal of Hydrology. 1970;10(3):282290.