Insights Into the Declined Efficacy of In Situ Deep Soil Benzene Biostimulation: An Investigation Across Four Sites Over Three Years

Biostimulation is a widely used approach for remediating deep‐layer oil‐contaminated soils. However, at four petroleum hydrocarbon‐contaminated sites in Saskatchewan, Canada, we observed a marked decline in the effectiveness of benzene biostimulation over a three‐year period. The underlying causes of performance decline are poorly understood. To investigate the factors contributing to the reduced efficacy of biostimulation, this study hypothesizes that either the delivery of amendments was affected by soil matrix or the prevalence of benzene‐degrading microbes declined in areas requiring remediation. Deep soil samples were collected annually and analyzed for benzene concentration, water‐soluble ions, and the abundance of functional microbial genes associated with benzene degradation. A generalized linear mixed model (GLMM) was used to examine the relationship between the binary remediation outcome (success or failure) at the sample scale and water‐soluble ion concentrations, with site treated as a random effect. A linear model was applied to investigate the relationship between failure rate of remediation and soil properties at the site scale. The GLMM identified soil pH, along with soluble PO43−, Ca2+, SO42−, NO3− and NO2−as significant contributors to the effectiveness of biostimulation at sample scale. Notably, Ca2+ and PO43− exhibited comparable importance but opposite effects, with Ca2+ negatively and PO43− positively associated with successful remediation. The linear model found that soil water‐soluble Ca2+ and SO42− were positively correlated with the rate of declined remediation outcomes at site scale (p < 0.05). We inferred that at sites with moderately high background SO42−, decade‐long natural attenuation rendered the benzene more recalcitrant. High soil‐soluble Ca2+ could sequester the phosphate introduced by amendments, forming precipitates that reduced phosphorus availability. Given that the amendments contained nitric acid, an observed increase in pH or a decrease in electrical conductivity in the samples after biostimulation relative to pre‐biostimulation conditions, suggests that fewer amendments reached the polluted plume. This may indicate that the initial infiltration pathways became clogged, potentially due to Ca–P precipitation. Moreover, the decline in functional genes linked to anaerobic benzene degradation suggests insufficient microbial capacity to utilize the amendments. This emphasizes the need to tailor biostimulation strategies for successful in situ biostimulation, ensuring effective delivery of amendments, particularly over long‐term practices, and to sustain microbial activity under field conditions.