The parameters requiring the fewest fish (4–16 fish per site) were EROD and ECOD activity, serum SDH, and biliary PAH metabolites. Analysis of HSP70, LSI, GSI and CF required considerably more fish per site (13–106). These numbers HDAC inhibitor mechanism generally increased in direct proportion to requirements for decreasing amplitudes of the difference from reference values. For EROD and ECOD activity, only 4–12 fish/site were needed to detect a 3-fold induction. Previous studies with other fish species gave similar results. Flammarion and Garric (1999) estimated that 13 fish/sex/season/site were required to detect a 2-fold induction of EROD activity at α = 0.05 in chub (Leuciscus cephalus). Similarly,
Beliaeff and Burgeot (1997) calculated for a variety of fish species that 10 fish were required to detect a 3-fold EROD activity induction at α = 0.10. The required number of fish computed in the present investigation was comparable to numbers reported in the published literature for field studies, where EROD activity is, on average, investigated using n = 7 fish per site (and laboratory studies use on average five fish per treatment, Oris and GSI-IX in vivo Roberts, 2007).
Some acute field exposures may cause large and significant difference with very few fish. For example, following an oil spill, a significant EROD induction in rockfish (Sebastes schlegeli) and in marbled flounder (Pseudopleuronectes yokohamae) was detected using only n ⩾ 3 fish per site ( Jung et al., 2011). The field sampling from which the black bream data set was extracted was conducted this website outside of the reproductive season for this species to avoid a gender bias in EROD activity. While EROD activity is unbiased by gender in this case, other parameters such as GSI and reproductive parameters in general could not be investigated properly using this data set because the fish were not sexually mature. While a 10% change in these parameters required that 43–106 fish be sampled, the field data suggest that only 13–36 fish per site would be sufficient, as inter-site
differences in LSI and GSI often varied by more than 10%. Four factors will influence the required number of samples (n) to collect. The first, the significance level α, is almost uniformly accepted at α = 0.05, meaning that for 1 in 20 comparisons, there may be a false positive and incorrect conclusions about effects. Lowering α causes n to increase dramatically but it may be practical to collect a larger number of samples if the biomarker analyses are inexpensive, or if more fish are needed for other responses. The second factor is the desired minimum detectable difference amongst sites, which will be specific to each location and to each biomarker. No obvious rulings exist for the magnitude of change that can be appropriate to specific situations (Hanson et al., 2010). For each biomarker, we estimated a biologically or environmentally relevant degree of change between reference and impacted fish (Table 1).