Determination of trace elements in red snapper, Lutjanus campechanus, otoliths using ICPMS to evaluate their affinity for oil and gas platforms

Abstract

From Results: "Discriminant function analysis was successful in differentiating between our two sites using all 17 elements that were analyzed with otolith microchemistry. Two approaches were used in the discriminant function analysis to estimate the misclassification rate of the 61 observations into
site 1 (oil and gas platforms) and site 2 (artificial reefs). The first was the resubstitution approach. It correctly classified 29 of the 30 observations from site 1 but misclassified one observation into site 2. This approach also correctly classified all 31 observations from site 2. Therefore, this approach only had a misclassification rate of 1.67%. However, the resubstitution approach is known in general to be overly optimistic and the second approach used, the crossvalidation approach, gives a more realistic misclassification rate. The cross-validation approach correctly classified 25 of the 30 observations from site 1 and misclassified five of them into site 2. It correctly classified 27 of the 31 observations form site 2 and misclassified 4 observations
into site1. This approach has a more realistic misclassification rate of 14.78%. A stepwise discriminant function was then performed on the 61 observations, and it identified the most influential elements used to discriminate between oil and gas platforms and artificial reefs in Louisiana. The elements identified as the most important in the analysis were Silver 109, Zinc 66, Vanadium 51, Uranium 238, Lead 206 and Nickel 58. Using only these six elements to rerun
the original discriminant analysis yielded a cross-validation approach misclassification rate of only 5%, reduced from the original estimate of 14.78%."