A multivariate statistical method and a heuristic method were employed to examine the structural features associated with the persistence or degradation of 287 chemicals tested with the standard biochemical oxygen demand (BOD) procedure. The data base consisted of 196 “degradable” chemicals with a half‐life of less than 15 d (e.g., Theoretical BODs (ThOD) > 16%) and 91 “persistent” chemicals with a half‐life of more than 15 d. The multivariate statistical analysis consisted of (1) calculating 54 molecular connectivity indices, five physicochemical properties and eight principal components from the molecular connectivity indices, (2) clustering the chemicals on the basis of the principal components and one of the physicochemical properties, Kow and (3) discriminating between persistent and degradable chemicals using the molecular connectivity indices and the physicochemical properties as discriminating variables within each cluster. The heuristic approach used the results of the multivariate analyses and the literature on biodegradation to identify a series of structural features associated with degradable and persistent chemicals. The best iteration of the multivariate technique correctly predicted 85% of the degradable chemicals and 94% of the persistent chemicals. In contrast, the heuristic approach correctly predicted 91% of the degradable chemicals and 96% of the persistent chemicals. Twelve structural features or chemical classes were associated with degradable chemicals and 16 structural features were associated with persistent chemicals. The analysis is presented as a potential technique for rapidly assessing the relative degradability of discrete organic chemicals. The structural features identified are offered as tentative hypotheses to be examined with a larger and more diverse data set.
- Multivariate analysis