ter Laak Thomas L, Gebbink Wouter A, Tolls Johannes
RAS-Institute for Risk Assessment Sciences, Utrecht University, Yalelaan 2, PO Box 80176, 3508 TD Utrecht, The Netherlands.
Environ Toxicol Chem. 2006 Apr;25(4):933-41. doi: 10.1897/05-229r.1.
Environmental exposure assessment of veterinary pharmaceuticals requires estimating the sorption to soil. Soil sorption coefficients of three common, ionizable, antimicrobial agents (oxytetracycline [OTC], tylosin [TYL], and sulfachloropyridazine [SCP]) were studied in relation to the soil properties of 11 different soils. The soil sorption coefficient at natural pH varied from 950 to 7,200, 10 to 370, and 0.4 to 35 L/kg for OTC, TYL, and SCP, respectively. The variation increased by almost two orders of magnitude for OTC and TYL when pH was artificially adjusted. Separate soil properties (pH, organic carbon content, clay content, cation-exchange capacity, aluminum oxyhydroxide content, and iron oxyhydroxide content) were not able to explain more than half the variation observed in soil sorption coefficients. This reflects the complexity of the sorbent-sorbate interactions. Partial-least-squares (PLS) models, integrating all the soil properties listed above, were able to explain as much as 78% of the variation in sorption coefficients. The PLS model was able to predict the sorption coefficient with an accuracy of a factor of six. Considering the pH-dependent speciation, species-specific PLS models were developed. These models were able to predict species-specific sorption coefficients with an accuracy of a factor of three to four. However, the species-specific sorption models did not improve the estimation of sorption coefficients of species mixtures, because these models were developed with a reduced data set at standardized aqueous concentrations. In conclusion, pragmatic approaches like PLS modeling might be suitable to estimate soil sorption for risk assessment purposes.
兽药的环境暴露评估需要估算其在土壤中的吸附情况。研究了三种常见的可离子化抗菌剂（土霉素[OTC]、泰乐菌素[TYL]和磺胺氯哒嗪[SCP]）的土壤吸附系数与11种不同土壤的土壤性质之间的关系。在自然pH条件下，OTC、TYL和SCP的土壤吸附系数分别为950至7200、10至370和0.4至35 L/kg。当人工调节pH时，OTC和TYL的变化增加了近两个数量级。单独的土壤性质（pH、有机碳含量、粘土含量、阳离子交换容量、羟基氧化铝含量和羟基氧化铁含量）无法解释土壤吸附系数中超过一半的变化。这反映了吸附剂与吸附质相互作用的复杂性。整合上述所有土壤性质的偏最小二乘（PLS）模型能够解释高达78%的吸附系数变化。PLS模型能够以六倍的精度预测吸附系数。考虑到pH依赖的物种形成，开发了物种特异性的PLS模型。这些模型能够以三到四倍的精度预测物种特异性的吸附系数。然而，物种特异性吸附模型并没有改善对物种混合物吸附系数的估算，因为这些模型是在标准化水浓度下使用减少的数据集开发的。总之，像PLS建模这样的实用方法可能适用于为风险评估目的估算土壤吸附。
