Without research on those factors, the source of variation cannot be controlled, and the inherent variability might be so high that the biomarker is invalidated as part of a field monitoring program. Minimizing the effects of confounding factors can reduce systematic sampling error. For example the data set used in the present exercise included only non reproductively-active Ku-0059436 cost adult fish to reduce the high variability of EROD activity

among female fish at the onset of spawning. Estrogen is known to down-regulate the cyp1a gene, so that assays of EROD activity in sexually maturing female fish approaching spawning will inflate the variance of EROD activities of a mixed sample of male and female fish ( Forlin and Haux, 1990). If the biomarker selected is influenced by the gender of the fish, the data provided in Table 3 represents the number of fish per sex to selleck chemical be collected at each site, assuming that the variance is equal between sexes. It is worthwhile to note that in field studies, seasonality in biomarkers of fish health often introduces variability that is higher than inter-site variability ( Hanson et al., 2010), making it increasingly difficult to relate cause and effects. A rigorous

sampling program with an adequate number of fish sampled will offer a reasonable potential to offset high seasonal variability. While the influence of confounding factors might be minimized, the analytical variability can still be surprisingly high. In an inter-laboratory round-robin, Munkittrick et al. (1993) found that EROD activities measured in sub-samples of fish livers varied considerably. For seven laboratories reporting EROD activities measured with 9000g supernatants (S-9 fractions), the coefficients of variation of arithmetic mean EROD activities of six fish per site sampled from reference and pulp

mill sites ranged from 46–80% (calculated from Table 2, Munkittrick et al., 1993). However, the variation in induction (i.e. the proportional increase in activity between reference and exposed sites) was much less, with a cv of only 30% among the seven independent labs. This indicates that the variance among labs was likely related Dynein to differences in methods that affected induced and uninduced fish equally. Standardization and improvement of analytical protocols can reduce analytical variability (van den Heuvel et al., 1995), thereby increasing the probability of detecting an inter-site difference. Because this variability is entirely within the control of the monitoring agency, it can be beneficial to develop Quality Assurance/Quality Control (QA/QC) protocols for each biomarker. For example, in addition to variations among fish of EROD activity, variation in EROD assays can be generated from each step of the assay, including preparation of S-9 fractions, the biochemical assay, and the analysis of data.