Evaluation of agriculture-based phytoremediation in Pacific island ecosystems using trisector planters.

It is difficult to directly evaluate the efficacy of phytoremediation of petroleum hydrocarbon contaminants embedded in deep soil layers, especially if the contaminants are of relatively low concentration and are unevenly distributed. This report describes the greenhouse and laboratory experiments carried out to evaluate a field demonstration project. A trisector planter was designed to simulate field conditions, including soil profiles and field management of the trees selected. The third or bottom section of the planter was spiked with known quantities of 6 diesel-fuel components and the reduction of their concentrations was monitored after 200 days under the influence of the plant root systems. Results are statistically compared; among the three tree species used, milo (Thespesia populnea) and kou (Cordia subcordata) are more effective than false sandalwood (Myoporum sandwicense) in reducing the concentration of the spiked contaminant. Enumerations of populations of hydrocarbon-degrading microoorganisms in the bottom section suggest that biodegradation may be affected by the response of microorganisms to both the "close rhizosphere" (soil within 1 mm of the root) and the "expanded rhizosphere" (soil in the bottom section after root removal). Root exudates leached from the upper sections could be responsible for the expanded rhizosphere effect in the bottom section.