A multivariate analysis technique, polytopic vector analysis (PVA) (Ehrlich and Crabtree, 2000, Johnston et al., 2002 and Ramsey et al., 2005), was applied selleck chemical to extract additional information from the 15 diagnostic ratios used to identify sediment samples containing MC-252 oil. After excluding six of the 29 samples with missing ratios (noted in Table 3), the remaining 23 samples containing all
15 diagnostic ratios were input into PVA to determine the least number of indicator diagnostic sample-sets that captured the variance of these 23 samples plus the MC-252 source oil (a total of 24 sample-sets of diagnostic ratios). The indicator sample-sets were identified by deriving a simplex or encapsulating surface defined by vertices lying dominantly in the positive orthant (physically realistic solutions) that contained www.selleckchem.com/products/nu7441.html all input diagnostic ratios (represented as vectors) within the simplex. Next, the similarity of each sample-set to each indicator sample-set was calculated based on distances between the coordinates defining each sample-set and simplex vertices (Ehrlich and Crabtree, 2000 and Ramsey et al.,
2005). In the final PVA processing, the diagnostic ratio set defining the MC-252 sample was set as one of the simplex vertices in order to directly assess the likelihood of each sediment sample containing MC-252 oil. The quality of the similarity analyses performed by PVA was evaluated initially based on two criteria. First, the similarity measures associated with the sediment samples should align with the designations, match (included the two probable match samples), inconclusive, and non-match determined in the oil source-fingerprinting and 17-DMAG (Alvespimycin) HCl diagnostic ratio analysis. Once the
first criterion was met, sediment samples comprising the inconclusive category were evaluated based on their similarity to MC-252 and on their physical proximity to locations of sediment samples designated as match or non-match. If the similarity measure and spatial proximity (<100 m) both indicated high alignment with samples comprising the match category, those inconclusive sediment samples were considered to contain MC-252 oil and assigned to the PVA-match category. Inconclusive sediment samples failing one or both criteria remained in the inconclusive category. Diagnostic ratio analysis separated the 29 sediment samples into match, probable match, inconclusive, and non-match categories (Table 3). The use of the supplemental alkyl DBTs/Phens ratios moved samples 33 Shore and 34 Interior from the probable match to match category, resulting in 9 match, 8 inconclusive, and 12 non-match sediment samples prior to PVA.