DDT and its metabolites were analysed from 2002 to 2011 using the same method as

described for PCB6. From 2006 several other organochlorine pesticides were also analysed using GC/MS as described by Berntssen et al. (2011b). For quality control, an in-house control sample was analysed with each run, and the CRM SRM-1946 from the National Institute of Standards and Technology (Gaithersburg, USA) was analysed at least once a year. The pesticides selleck chemical analysed were: aldrin, dieldrin, alpha-endosulfan, beta-endosulfan, endosulfan-sulphate, alpha-hexachlorocyclohexane, beta-hexachlorocyclohexane, gamma- hexachlorocyclohexane, cis-chlordane, trans-chlordane, oxy-chlordane, cis-nonachlor, trans-nonachlor heptachlor A, hexachlorocyclobenzene, isodrin, mirex and toxaphene (40 + 41, 26, 32, 42a, 50, 62). Not all pesticides were measured in each sample; the number of replicates for each pesticide is given in Appendix 1. This method was accredited in 2005, and remained accredited to and including 2011. The median was chosen as the descriptor of the population mean due to the lack of normality of the data, and a large number of measurements below the LOQ. Median is presented with interquartile ranges to illustrate variability. When more than 50% of the results were below the LOQ the median was not calculated. Since the uncertainty

increases when one approaches the LOQ, regression analyses were excluded when more than 50% of the analyses were below 1.5× LOQ. Regression was used for evaluating time trends for the different contaminants. Differences between years were also determined using the non-parametric Kruskal–Wallis with post hoc paired comparisons. Differences were regarded mTOR inhibitor as significant when p < 0.05. Statistical Urocanase analyses were performed using

Statistica 9 (StatSoft Inc., Tulsa, USA), and Graphpad Prism 5.04 (Graphpad software Inc., San Diego, CA, USA). A total of 1025 samples from 2005 to 2011 were analysed for elemental mass fraction of Hg, As, Pb and Cd. For Cd, the measured levels in 933 of the total 1025 samples were < LOQ (0.01 mg kg− 1 w.w.), whilst 994 measurements of Pb were < LOQ (0.04 mg kg− 1 w.w.). In contrast, the measured levels of Hg were < LOQ (0.03 mg kg− 1) in only seven samples, and As was detectable > LOQ in all samples. Since most of Cd and Pb data were < LOQ, statistical analyses of time trends were not feasible. A linear regression showed a decline during the 6 years of sampling for As and Hg mass fraction in fillet (Fig. 1). This time trend was verified using the non-parametric Kruskal–Wallis test. The median elemental mass fractions of As and Hg in fillets for the time period were 1.07 and 0.029 mg kg− 1 w.w. respectively. A total of 432 samples were analysed for dioxins and dl-PCBs from 1999 to 2011. For data from 2005 and earlier, only 1998 WHO TEQ was available. Thus a conversion regression described by Bhavsar et al. (2008) was used to convert data to WHO-TEQ 2005. The regression analyses performed by Bhavsar et al.