Monitoring PCBs in polar bears: lessons learned from Svalbard Espen O. Henriksen, a Øystein Wiig, a,b Janneche Utne Skaare, c,d Geir Wing Gabrielsen a and Andrew E. Derocher* a a Norwegian Polar Institute, N-9296 Tromsø, Norway b Zoological Museum, University of Oslo, PO Box 1172 Blindern, N-0318 Oslo, Norway c Norwegian School of Veterinary Science, PO Box 8146 Dep., N-0033 Oslo, Norway d National Veterinary Institute, PO Box 8156, N-0033 Oslo, Norway Received 22nd March 2001, Accepted 6th July 2001 First published as an Advance Article on the web 13th September 2001 Monitoring pollutants in the biota of the Arctic is a high priority activity of the circumpolar Arctic nations. Polar bears (Ursus maritimus) are one species that have been selected for monitoring, owing to their high trophic position in the Arctic marine ecosystem and high contaminant levels. Considerable research has been directed at understanding the effects of pollutants, and ultimately these effects are tied to temporal trends in pollutant levels. This paper reports on the state of contaminant monitoring of polar bears in the Norwegian Arctic and provides recommendations for future monitoring programmes. PCB-153 decreased significantly in plasma collected from polar bears sampled at Svalbard during the 1990s. Future monitoring efforts should sample annually at the same location, at the same time of year and analyse 10–25 samples per year. Introduction The establishment of the Arctic Monitoring and Assessment Programme (AMAP) reflected the agreement of the circum- polar Arctic countries to monitor pollutants and their possible effects on humans and wildlife. 1 Although governments have agreed to monitor pollutants, few such data series exist for wildlife of the European Arctic. Most published data must be categorised as surveys, which are investigations that include measurements from many different locations and/or species over a limited period of time. 2 In contrast, monitoring is defined as ‘the repeated measurement in the same place and on the same substrate at fixed intervals of time’. 2 Typically, most variation in animal contaminant levels is associated with factors that are not expected to change systematically over time in the population. These factors could be the fat content of the analysed matrix, the sex, age and physical condition of the animal or seasonal changes in contaminant exposure. 3,4 Furthermore, trends in toxicology data are often difficult to discern owing to short time series, natural background variation and changes in analytical sensitivity. 5,6 The reduction of variation associated with such factors will increase a monitoring programme’s ability to detect change over time and thus increase statistical power. To optimise monitoring programme design, the objectives and performance requirements must be predefined. The performance criteria, statistical power and significance level a, correspond to the quality of the two possible conclusions of a trend study: no significant trend or significant trend. Thus, the interpretation of an apparent trend must be judged in the light of the significance level a, which quantifies the risk that the apparent trend is no more than a result of random variation. Conversely, a conclusion of no significant trend must be judged in the light of the programme’s probability to actually detect a trend of a given size. This probability is defined as the programme’s statistical power. 7 . In the Norwegian Arctic, the highest concentrations of persistent organic pollutants have been found in top predators, such as glaucous gulls (Larus hyperboreus), Arctic fox (Alopex lagopus) and polar bears (Ursus maritimus). 8–10 Contaminants reach particularly high levels in polar bears because they are the apex predators in Arctic marine ecosystems and rely largely on the blubber of seals for food. 11 Studies have examined the spatial patterns of organochlorine contaminants in polar bears, 10,12 but temporal trend data are lacking. Recent studies have indicated that hormone regulation and immune function of polar bears in the Norwegian Arctic are negatively affected by polychlorinated biphenyls (PCBs). 13,14 It is likely that contaminant effects will change with contaminant levels, so that monitoring temporal trends is critical to understanding the threats to wildlife populations. During the 1990s, more than 500 polar bear samples were analysed for contaminants in conjunction with various research projects in which monitoring for temporal trends was only one of several aims. 13–15 In the present study, we analyse and evaluate the applicability of the data obtained for determining temporal trends. The primary objective was to develop insights to improve monitoring of pollution in polar bears at Svalbard. In particular, we considered three aspects: (i) the levels of variance in different sampling matrices; (ii) adjustment for confounding factors to reduce irrelevant variance; and (iii) statistical power of different sampling strategies based on the variance in existing data. Methods Contaminant data Plasma, blood cells, milk and subcutaneous fat samples were collected from live-captured polar bears in the Svalbard area from 1990 to 1998 using methods approved by The Norwegian Animal Research Authority. Analyses of chlorinated hydro- carbons were performed at The Environmental Toxicology Laboratory at The Norwegian School of Veterinary Science/ The Norwegian Veterinary Institute. Details on capture, sampling and pollutant analyses have been described pre- viously. 12,15 The laboratory was accredited by Norwegian Accreditation in 1996 as a testing laboratory for organochlor- ines (OCs) according to the requirements of NS-EN45001 (Norwegian Standard–European Standard 45001) and ISO/ IEC (International Standardisation Organization/International Electrotechnical Commission) Guide 25. The laboratory’s analytical quality was approved in inter-calibration tests containing components and matrices relevant to our work, DOI: 10.1039/b102683f J. Environ. Monit., 2001, 3, 493–498 493 This journal is # The Royal Society of Chemistry 2001