Multivariate Modelling of Total Petroleum Hydrocarbon Reduction in Crude Oil-Contaminated Soil Using Costus afer Plant

Abstract

Phytoremediation is a cost-effective and environmentally friendly approach for the treatment of crude oil-contaminated soil. Costus afer is widely available in the Niger Delta region and has shown strong potential for the remediation of polluted sandy loam soils. This study developed a predictive model describing the phytoremediation performance of Costus afer in reducing total petroleum hydrocarbon (TPH) concentrations over a 90-day treatment period. A multivariate regression model was applied to quantify the relationship between TPH removal and two independent variables: crude oil volume (0.5, 1.0 and 1.5 litres) and remediation time (30, 60, and 90 days). Model evaluation showed close agreement between measured and predicted TPH removal, with correlation coefficients ranging from 0.808 to 0.828. Prediction errors were low, as indicated by root mean square error values between 3.06 and 5.60, while residual prediction deviation values ranged from 2.05 to 8.21. Over 80% of the variance in TPH removal was explained by the model (R2 > 0.80), confirming strong reliability. The model provides a practical tool for estimating crude oil volume and remediation time required for effective cleanup of contaminated sandy loam soils using Costus afer plants of known age. This study supports the use of predictive modelling for optimizing phytoremediation system design and management in crude oil-contaminated environments.