Skip to main content

A global, cross-system meta-analysis of polychlorinated biphenyl biomagnification

Cite this dataset

Prince, Kimberly (2020). A global, cross-system meta-analysis of polychlorinated biphenyl biomagnification [Dataset]. Dryad.


Studies evaluating the mechanisms underpinning the biomagnification of polychlorinated biphenyls (PCBs), a globally prevalent group of regulated persistent organic pollutants, commonly couple chemical and stable isotope analyses to identify bioaccumulation pathways. Due to analytical costs constraining the taxonomic and geographic scope, sample size, and the range of compounds analyzed for most studies, and study-to-study variation in methodologies and analytical resolution, how PCBs biomagnify at food web, regional, and global scales remains uncertain. To overcome these constraints, we compiled diet (stable isotopes) and lipid-normalized PCB data from peer-reviewed studies reporting both values and used complementary analyses to evaluate the relative importance of global key PCB drivers and assess ecosystem- and ocean-wide biomagnification trends of sum total PCB concentrations (PCBST), and the concentrations of seven individual PCB congeners, and their sum (PCBå7). We discovered that the number of congeners analyzed, region, and class were the most important factors predicting PCBST, while, similarly, region, class and feeding location were the best predictors of PCBå7 and all seven congeners. In addition, biomagnification analyses revealed that PCBST, PCBΣ7 and the seven individual PCBs all demonstrate a higher propensity for biomagnification in marine relative to freshwater food webs and within the Atlantic Ocean relative to the Pacific. We further found that some congeners exhibiting relatively high trophic magnification factors (TMFs) in the Atlantic exhibited low TMFs in the Pacific (such as PCB 118), while the order of individual congener TMFs relative to one another remained consistent across marine and freshwater ecosystems. Our analyses demonstrate that novel insights regarding PCB concentrations across taxonomic, food webs, regional and global scales can be gleaned by leveraging existing data to overcome analytical constraints. 


To synthesize stable isotope and PCB data, we conducted a search in April 2019 on Web of Science ( using topic words “polychlorinated biphenyl* or PCB*” AND “stable isotope*” with no time restriction. To expand the coverage of this search, we applied the same search criteria and incorporated any additional studies from the following journals that commonly publish on this topic: Environmental Research, Chemosphere, PLOS one, Environmental Toxicology and Chemistry, Environmental Science and Technology, Science of the Total Environment, Environmental Pollution, Ecotoxicology and Environmental Safety, Marine Pollution Bulletin, Journal of Wildlife Diseases, Marine Ecology Progress Series, Environmental Science Pollution and Restoration,Chemistry and Ecology, Environment International, and Frontiers in Ecology and the Environment. Additional studies referenced within studies identified in our initial search were also incorporated in our database. Across all studies, we only included those that published original data and reported both PCB concentration and stable isotope data, resulting in a total of 231 studies published between 1995 and 2019.

Since studies varied in reported units of measurement, we further reduced this database to the subset of 73 studies that reported PCB concentrations in ng/g lipid-normalized weight (lw). In these studies, we extracted 815 unique sum total PCB concentration (PCBST) values, i.e. the summed concentration of all congeners measured for a given sample. PCBST was overwhelmingly the most commonly reported response variable, and only 31 of the 73 studies provided concentrations for individual congeners in either the main text or supplement files in addition to PCBST values. In the few studies that only reported individual PCB congener concentrations, we calculated PCBST ourselves. We herein present analyses focused on evaluating the relative importance of different drivers of PCBST, as a measure of aggregate PCB exposure in line with prior studies34–36, at global, regional, and food web scales given the large and geographically well-distributed characteristics of this dataset (Fig. 1). In our PCBST analyses we include the number of congeners analyzed an explanatory variable to help control for between-study differences in this important dimension of analytical effort (see Data Analysis below). Though details on which congeners analyzed were available, without concentrations of each congener we were unable to include them in the analysis. Thus, the total number of congeners analyzed in each study was used. 

We complement the PCBST dataset with one in which the number and composition of congeners analyzed are standardized. In this second dataset, we focus only on the seven most commonly analyzed and reported PCB congeners (PCBs 28, 52, 101, 118, 138, 180, 153) as well as their sum (PCBΣ7), which we derived from 13 of the 73 studies. This particular set of seven congeners happens to be the full suite of indicator PCBs recommended for monitoring by the International Council for the Exploration of the Sea (ICES) given their high concentrations in commercial mixtures and wide chlorination range . 

For each dataset, we then categorized samples by geographic region (i.e. country and, if relevant, continent or ocean including associated seas and bays) and ecosystem type: marine, freshwater, brackish (mix of freshwater and saltwater), terrestrial and multiple systems (if the sampled organism utilizes multiple ecosystems). We also cataloged geographical coordinates when available and, if not provided, estimated them based on the study site description using Google Maps. In addition, we recorded the taxonomic family of each sample (class), the type of tissue sampled (i.e. fat, blubber, blood, red blood cells, plasma, egg, embryo, heart, kidney, liver, muscle, milk, spleen, pylori, whole body or multiple tissues), the composition and total number of congeners analyzed, and relevant characteristics of the sampled organism including its biomass, age, and trophic level. If trophic level was not provided, we assigned the sampled species a trophic level using or used the organism’s consumer status, which we based on published information about its diet. Primary producer, primary consumer, secondary consumer, tertiary consumer, or apex predator organisms corresponded to trophic levels 1-5. To assess the validity of the assigned trophic levels, we first fit a linear model between isotopic nitrogen and trophic level using 252 samples for which both values were reported in the study. We then found that 92% of the assigned trophic level values for the remaining 526 samples in our data set were within the 95% prediction interval of this linear model, indicating that they correspond well with study-derived trophic levels. 

We further classified organisms by feeding location (organisms occurring exclusively within aquatic systems were categorized as benthic, demersal, benthopelagic, pelagic, bathypelagic, or bathy-demersal and primarily land-based species were categorized as: exclusively-terrestrial (e.g. sparrows, spiders, owls, moths, pigeon), terrestrial/marine (e.g. Artic foxes, Polar bears, herring gull), or terrestrial/freshwater (dragonflies, amphibians, ducks), and by feeding behavior (filter feeder, deposit feeder, autotroph, herbivore, carnivore, or omnivore). When necessary, we used Web Plot Digitizer ( to extract isotope data, trophic level, and PCB concentration data. 

Usage notes

To push PCB biomagnification analyses toward the forefront of the ecotoxicology field, future studies can continue to add to this database through the reporting of PCB and stable isotope data using standardized methods and units of measurement (e.g. lipid-normalized concentrations), thereby collectively advancing knowledge on the global transport and fate of PCB concentrations. Further, we encourage others to leverage the data compiled in our existing database to contextualize their findings relative to other organisms sampled in a similar region or from a taxonomic class. Such communal datasets provide one of the few comprehensive tools to understand and manage how PCBs move into and across ecosystems at the global scale.