EMERGING PERSISTENT ORGANIC POLLUTANTS (POPS) IN THE WESTERN SOUTH ATLANTIC AND ANTARCTIC BIOTA

Persistent organic pollutants (POPs) are largely synthetically produced chemicals that are known to persist in the environment, bioaccumulate, have the potential to be transported long distances, and cause adverse effects. There are legacy POPs that have been around for decades and have either been banned or strictly regulated, but are still found in the environment; and there are emerging POPs that are either not yet or are very newly regulated. This research focuses on contributions to the global dataset of emerging POPs by investigating hydrophilic perfluoroalkyl substances (PFAS) in surface waters and at depth of the Western South Atlantic; as well as hydrophobic polybrominated diphenyl ethers (PBDEs) in Antarctic biota (plankton, krill, fish, fur seal milk). PFAS were found in all surface waters (ΣPFAS 20.3 – 525.8 pg/L) and at depths of up to 5526 m. This confirms the infiltration of these compounds into our global oceans. PBDEs were detected at the highest concentrations in Antarctic plankton (plankton > krill > fur seal milk > fish). This is contrary to the biomagnification seen in many legacy compounds and indicates the potential for biodilution and species-specific metabolic processes occurring. These data contribute to the growing knowledge of emerging pollutants in the southern hemisphere, which is generally less prominently covered in terms of pollution studies.

are categorized largely as either acids or sulfonates, differentiated by the functional group at the end of the carbon chain. The carbon-fluorine bond that is imparted during production is extremely strong, which makes these chemicals extremely useful in a wide range of consumer and industrial applications (e.g. non-stick cookware, stain-repellent fabrics, food packaging and water-resistant apparel). At the same time, it also makes them extremely stable and resistant to environmental degradation. Historical production focused largely on the eight-carbon (C8) chemistries with perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) being both the most heavily produced and utilized compounds. PFOA and PFOS are also the degradation end products of numerous other fluorinated (precursor) compounds, thus making their environmental concentrations the greatest. PFAS are now found ubiquitously in surface waters, wildlife, and humans (Giesy & Kannan 2001, Yamashita et al. 2005, Houde et al. 2006, 2011a, Lindstrom et al. 2011, Benskin et al. 2012 are allowable exemptions to Annex B and while it may be easier to regulate production say, within the United States, there has been evidence that there is and may continue to be "knock-on" production of C8 and other PFAS in some countries within Asia. In addition there is increasing production of shorter-chained (e.g. C4) PFAS as replacements of the eight-carbon compounds, along with the increased use of more volatile precursor compounds that have the capability of undergoing long-range transport through the atmosphere and ultimately degrading into compounds such as PFOA and PFOS (Paul et al. 2009, Lindstrom et al. 2011).
One of the unique physiochemical properties of PFAS is that unlike many of their legacy POP counterparts, they are largely hydrophilic, making their ultimate environmental fate our world's oceans (Yamashita et al. 2005). Atmospheric transport and deposition has been described for the neutral and volatile precursor compounds such as fluorotelomer alcohols (FTOHs) and perfluorinated sulfanomido alcohols (FOSEs).
These can undergo oxidation to ionic PFAS, and quickly deposit thereafter. The other main transport mechanism that has been proposed is via the hydrodynamics of ocean circulation transporting these compounds away from source regions (Ahrens et al. 2010, Butt et al. 2010). Sources to riverine systems include nonpoint ones such as rainfall, runoff, and snowmelt, but wastewater treatment plants are also known point-source contributors of PFC loads to river systems (Furl et al. 2011).
While much of the world's surface waters have been surveyed for PFAS, little is known about PFAS at depth. If the oceans are indeed the final sink for these compounds, we need a more thorough understanding of how PFAS are moved to depth and affected by ocean circulation. Yamashita et al. (2008) proposed that perfluorinated acids may be used as novel tracers for global ocean circulation due to their high water solubility, persistence, measurability, and the fact that they (at least the acids) are less bioavailable than many other POPs. Vertical profiles obtained from the Labrador Sea, the mid Atlantic Ocean, the South Pacific Ocean, and the Japan Sea provided some of the first available data to look at PFAS beneath the surface waters and it was estimated that approximately 1% of PFOA emissions since production began (around 60 years ago) has been transported to deep ocean waters. (Yamashita et al. 2008a) Later, Lohmann et al. (2013) proposed vertical eddy diffusion as a primary mechanism of PFOA removal from the surface, estimating that it accounts for 13% of the removal of PFOA from surface waters compared to 4% via deep water formation, with the surface waters (i.e. top 100m) storing at least 21% (Lohmann et al. 2013).
In this study, vertical seawater profiles were collected at 12 stations throughout the amended with 4mL Ammonia Solution), and 5mL water (RECETOX Milli-Q water), followed by the passing of 1 L sample water through the cartridge at approximately 1-2 drips per second. Upon near completion of the sample passing through the cartridge, 4mL of a wash buffer solution (water, acetic acid, and ammonium acetate) was used to rinse the sides of reservoirs and cartridges used, then allowed to pass through the cartridge. Cartridges were centrifuged at 4000rpm for 2 minutes, followed by elution into two fractions; the first fraction being eluted with 6mL methanol and the second fraction with 8mL basic methanol. All eluent was captured in pre-cleaned (basic methanol and methanol-rinsed) polypropylene falcon tubes and blown down with N 2 gas on a heated evaporation plate. Samples were reconstituted to 1mL volume with 0.5mL MeOH and 0.5mL water. Prior to analysis, samples were vortexed and 100µL was transferred to an autosampler vial.
Separation and detection was achieved using HPLC/MS/MS (Agilent 1100 liquid chromatograph, Applied Biosystem QTRAP 5500 mass spectrometer). A Phenomenex column (50 mm x 2.1 mm) with 3 μm particles was used combined with a mobile phase (methanol + 5 mM ammonium acetate) gradient elution. 10µl of sample was injected onto the column, which had a temperature of 25°C; flow of the mobile phase was set at 200µl/min. Capillary voltage in the MS ion source was set at 4500V, temperature at 450°C, and the electrical voltage multiplier at 2000V. Evaluation and quantification of data was based on a standard calibration set (0.04 to 40 ng/ml). Two MRM transitions for each compound were used to identify analytes.

Quality Assurance and Control
Fraction 1 recoveries of mass-labeled compounds averaged <1.2% for all carboxylates and sulfonates, thus concentrations for PFAS presented are all from fraction 2. FOSA, which is targeted to elute in fraction 1, recovery only averaged 21.5 ± 2.21% (standard error) and is not discussed further. Other compounds with poor recoveries of mass-labeled compounds (mean ± standard error) are perfluorobutanoic acid (MPFBA,7.38 ± 0.29%) and perfluorododecanoic acid (MPFDoDA, 29.03 ± 2.78%). MPFDoDA serves as the mass-labeled counterpart for perfluorotridecanoic acid (PFTrDA) and perfluorotetradecanoic acid (PFTeDA) and as such, none of these compounds will be discussed further. were determined individually for each sample as 10 times the signal to noise ratio. In many cases the MQL had quite a large range and was within or occasionally, above, the normal range of what is expected to be found in ocean waters (i.e. 10s to 100s of pg/L).
We recommend that in future work with saltwater, a freshwater flush of ultra-pure cartridge-cleaned water equivalent to 2% of the total sample volume (i.e. 20mL in the case of a 1L sample) be run through the cartridge prior to elution. All samples that were classified as < MQL were substituted for "0" in terms of means and sums. This is likely under representing PFC concentrations, but due to the variability in MQLs among samples, any other means of substitution would likely lead to an unacceptable overestimation of concentrations.
The eight-carbon PFOA and PFOS have been the most widely produced and utilized PFAS, and as such, they are typically the dominant compounds detected in the surface oceans (Yamashita et al. 2008b, Benskin et al. 2012 PFNA, and PFDS from the surface to depth, they appear to be nearly uniform at all depths, which is cause for concern of contamination and/or problems with these compounds, and so, for now, we will focus on the remaining compounds (SI Figure 1. in 42% of our surface water samples, our range of <MQL -32.5 pg/L also falls within that of prior surface water PFHxA concentrations (< 5 -75 pg/L) (Benskin et al., 2012, SI). We thus conclude that for the most common PFAS analyzed in our current work and previous studies, the range of concentrations is comparable, implying we achieved robust analytical results and contamination-free sampling.

Patterns of PFAS within the Atlantic Ocean
The samples collected in this study can be categorized into approximately 3 different groups in terms of regions sampled. The initial stations sampled closet to Uruguay were thought to possibly contain some influence of the Rio de la Plata, followed  (Dreyer et al. 2009).

Detection of PFAS at depth
Unexpectedly, PFAS were detected at most depths at the majority of stations, with the deepest detection at 5526 m. Initially, we expected PFAS to spike off the coast of Uruguay, a reflection of the Rio de la Plata influence, as seen in Benskin et al (2012).
However, the initial stations sampled closest to the river mouth reflected some of the lowest concentrations detected, and we believe that our sampling stations may have been too far offshore to in fact capture any riverine influence. Although, salinity at the time of sampling Station 1 was 34.9 PSU at the surface (5m), which was slightly less than that of Station 2 (salinity of 36.01 PSU at the surface). Guerrero et al. (1997) completed an assessment of the physical oceanography of the Rio de la Plata Estuary and found that in the months surrounding our sampling scheme (i.e. April -August), the flow out of the estuary drifts along the Uruguayan coast to the NNE direction (Guerrero et al. 1997).
Additionally, the stations sampled by Benskin et al. (2012) were much closer to the river mouth than our further offshore stations (Benskin et al. 2012 were the only two stations sampled along the same latitude and there appears to be a transport of pollutants carried west in this region, which was also seen in surface waters. The efficient vertical mixing of PFAS seen at these stations could be due, in part, to the complex dynamics of currents that occurs near the edge of ocean basins (Knauss 1997 (Speich et al. 2007, Bengtson Nash et al. 2010. While this range is fairly large, the lower end of the estimate is well within the time period during which PFAS have been widely produced (ca. 60 years).
World Ocean Circulation Experiment (WOCE) data from line A17, which has a very similar cruise track to ours, running along the South American coast (10°N to 51°S; 60°W to 30°W), show the infiltration of chlorofluorocarbon-11 (CFC-11) in AABW as far north as the equator and in NADW at 24°S, 20°S, and 10°S in samples taken from 1994 (Schlosser et al. 2001). Given that our samples were taken nearly 20 years later, it is not unreasonable to see PFAS present in waters today that contained CFCs then. Schlosser et al. (2001) present compiled CFC data reported by Warner and Weiss (1992) focused on AAIW, and note its renewal is on a decadal timescale. In 1992, CFC-11 was found in the majority of AAIW in the South Atlantic, with the water mass age averaging around 35 -37 years when located off the easternmost tip of Brazil at approximately 10°S, and 1 -3 years old when located south of 50°S and closest to its formation source region (Schlosser et al. 2001).
It is important to note, however, that in order for these deep and intermediate PFDoDA occasionally at very low levels (Wei et al. 2007, Ahrens et al. 2010).
Additionally, volatile precursor compounds are known to undergo long range atmospheric transport and have been detected in the Antarctic atmosphere (Dreyer et al. 2009, Del Vento et al. 2012). However, the concentrations of PFAS reported so far are not enough to account for the concentrations we detect in our subsurface and deep water masses, so there must be some other mechanism contributing to the transport of these compounds to depth.
The most notable trait of our profiles is the exceptionally high PFC content found along latitude 5.7°S. Perhaps, in a region where there is divergence of the South Equatorial Current to the North Brazil Current along with the presence of Equatorial countercurrents, there is turbulence and mixing of subsurface waters, causing efficient mixing of hydrophilic substances, such as PFAS, to depth. Alternatively, the biological pump may be an important transport mechanism that has not yet been investigated (Gonzalez-Gaya, presentation 2014). As our water samples were raw unfiltered seawater, it is plausible that PFAS are being transported to depth sorbed to particles and colloids.
Regardless these data confirm the detection of PFAS at depths as great as 5526m and at concentrations greater than those detected in the early 2000s ( We expect over time that the concentrations of shorter chain and precursors compounds may increase with the reduced usage of eight-carbon chemistries. Continued monitoring of PFAS at depth and in target water masses, along with further comparisons to existing ocean tracers, such as CFCs and Tritium, will further confirm the utility of PFAS as water mass tracers. Additional field data could potentially pair agreeably with modeling efforts to predict ocean mixing and circulation below the surface.      throughout the EU and US (two of the largest production areas), there is still production of others occurring (i.e. decaBDE) and a massive reserve of products exists around the globe that will have these chemicals leaching out of them for years to come (Hites 2004, Chiuchiolo et al. 2004, Sacks & Lohmann 2012.

MANUSCRIPT IN PREPARATION FOR SUBMISSION TO ENVIRONMENTAL SCIENCE
The continent of Antarctica is arguably home to some of the most untouched land on the planet. However, even in this remote region, the effects of humans are not unseen.
Scientific and military exploration has now been going on for decades and the summer season can witness over 100 active facilities operated by 30 different nations (COMNAP 2014). Research that takes place in Antarctica spans an enormously broad range of fields, from astrophysics and deep-sea oceanography to the more controversial-as-of-late topic of climate change. While pollution in Antarctica is typically orders of magnitude lower than concentrations found elsewhere around the globe, the fact remains that organic contaminants, particularly volatile ones, do reach the region via long range environmental transport through processes such as global fractionation and cold condensation (Wania & Mackay 1996). Legacy contaminants such as PCBs and organochlorine pesticides (OCPs) have been reported along with emerging contaminants such as PBDEs and perfluoroalkyl substances (PFAS) in numerous environmental matrices from the region (Chiuchiolo et al. 2004, Corsolini et al. 2006, Borghesi et al. 2008, Brault et al. 2013. in 2004 (Chiuchiolo et al. 2004, Corsolini et al. 2006. The distribution of PBDEs in the base of the Antarctic food web was complex, but biomagnification of PBDEs between phytoplankton and adult krill was not observed (Chiuchiolo et al. 2004). While PBDE congeners have been shown to bioaccumulate, it is apparent that PBDE metabolism may be species specific and variations in arctic food chains have been observed (Wolkers et al. 2004, Kelly et al. 2008. Kelly et al. (2008) presented evidence from a Canadian Arctic marine food web in which many PBDEs appeared to exhibit negligible biomagnification, with the exception of BDE-47, which did demonstrate food web biomagnification, albeit at a much lower level than PCBs (Kelly et al. 2008).
Trophic magnification factors (TMFs) represent an average biomagnification factor within food webs and are becoming increasingly useful tools in food web studies by allowing inter-ecosystem comparison. The application of TMFs has previously been used in the Arctic region and elsewhere globally, but to our knowledge, this will be their first use in the Antarctic ecosystem (Hobson & Welch 1992, Hobson & Ambrose 1995, Fisk et al. 2001, Hop et al. 2002, Kelly et al. 2008, Houde et al. 2011b, Borgå et al. 2012).
The PBDE data presented in this study are from a sample set of plankton, krill, fish, and fur seal milk post-2000 and will contribute to the unique dataset that comes from the remote Antarctic. Specific goals in this research were to (i) determine which PBDEs are being detected in Antarctic biota; (ii) establish trophic magnification factors (TMFs) to see if any observable biomagnification or biodilution is occurring; and (iii) to potentially establish temporal trends of congeners over a time period where global regulations and restrictions on production were being implemented (i.e. 2000s).

Sample Collection
Breast milk samples were collected from Antarctic fur seals (Arctocephalus (including mature females), or gravid females. At some sites, numerous krill were collected and provided enough biomass for "replicates" to be performed. In this case the replicates consisted of different individual krill from the same sample collection and were averaged and presented as one sample (SI Table 2.2 and 2.12 where * indicates multiple samples). Fish samples consisted of silverfish and myctophids collected in the same manner as krill.

Sample extraction
Seal milk extraction was conducted in two batches. The first batch, which consists of samples from the 5 austral summers spanning from 2000/01 -2010/11, was extracted at the Virginia Institute of Marine Science (VIMS) following previously established POP procedures as reported in Brault et al. (2013) (Risebrough et al. 1976, Chiuchiolo et al. 2004, Geisz et al. 2008, Brault et al. 2013. In short, seal milk was freeze-dried, homogenized, sub-sampled (1 g dry-weight), solvent extracted (65:35 DCM: Acetone), and analyzed for several POPs including DDT, PCBs, and chlordane.
Bulk lipid analysis was also performed as reported elsewhere (Brault et al. 2013) in order for sample concentrations to be lipid-normalized. Following analysis at VIMS, sample extracts were shipped to the University of Rhode Island's Graduate School of Oceanography (URI-GSO) to be analyzed for polybrominated diphenyl ethers (PBDEs) (Brault et al. 2013).
The second batch of seal milk samples (samples from 2011/12 and 2012/13) was extracted at URI-GSO as follows: a 2mL aliquot of each seal milk sample was transferred to 50mL centrifuge tubes where samples were spiked with a surrogate mixture, vortexed for 1 minute, and left overnight in the fridge. The following day, each sample was vortexed for 1 minute, followed by extraction three times in an ultrasonic bath with 20mL each of n-hexane/acetone (2:1) for 5 minutes. After each extraction, the organic layer was separated by centrifugation at 4000 rpm for 5 minutes. Combined extracts were evaporated, solvent exchanged to n-hexane, and brought to a final volume of 5mL.
200µL (from the 5mL) was taken for determination of percent lipid. Extracts were treated with sulfuric acid (concentrated) in an ice bath to remove lipids. Treated extracts were then partitioned on water (to remove excess acid), evaporated to 1mL, and cleaned on SPE cartridges (6 cc) filled with 2 g silica and topped with 1 g acidic silica (40%).
PBDEs were eluted with 50mL n-hexane/DCM (60:40). To determine percent lipid, the 200 µL aliquot to pre-weighed aluminum boats and left to dry overnight. The boats were re-weighed, with the difference in weights representing % lipid in the 200 µL, which was then extrapolated to % lipid in the 5 mL sample.
Plankton samples were investigated under a microscope in order to attempt basic species identification (i.e. diatoms or Phaeocystis sp.) and remove any visible zooplankton. Samples were then manually homogenized, freeze-dried at -80°C for approximately 72 hours, and solvent-extracted. Following analysis at VIMS, sample extracts were shipped to URI-GSO to be analyzed for polybrominated diphenyl ethers (PBDEs).
Each krill sample consisted of multiple individual krill that were homogenized prior to freeze-drying with a Virtis "45" tissue homogenizer (Virtis Co. Inc.), freezedried, and solvent extracted; if there were enough individuals, location replicates were measured (i.e. a different batch of homogenized krill from the same station collection).
Following analysis at VIMS, sample extracts were shipped to URI-GSO to be analyzed for polybrominated diphenyl ethers (PBDEs). Fish samples consisted of whole fish and followed the same extraction procedure as the krill samples. Further details on sample preparation can be gathered from Brault et al. (Brault et al.).

PBDE Analysis
All samples were analyzed for mono-through hepta-brominated congeners (BDE-  Results presented below are only for compounds that were detected > 30% of the time. Only peak areas with a signal to noise (S/N) ratio > 3 were considered quantifiable and for all samples with S/N < 3 or negatively corrected blank values, a value of "0" was used, which is likely underestimating ∑PBDEs.
Since these samples were not originally intended to be analyzed for PBDEs, the surrogate spike in this case was done post-extraction. We recognize that this is not ideal, however, Stable isotopes have now been used in biogeochemical applications for decades and are extremely useful tools in determining the trophic dynamics of a particular ecosystem. Carbon and nitrogen stable isotopes were measured in the majority of samples and are further presented in delta notation, δ, as parts per thousand different from a standard, or "per mil," ‰, using the following equation adopted from Peterson and Fry (1987): Where X denotes either 13 C or 15 N and R is the ratio of heavy to light isotopes in a sample (i.e. 13 C/ 12 C or 15 N/ 14 N) or standard, using a standard reference material of Vienna PeeDee Belemnite (PBD) or atmospheric nitrogen (N 2 , AIR) (Peterson & Fry 1987, Brault et al. 2013. δ 13 C and δ 15 N stable Isotopes for the majority of samples were determined via an elemental analyzer-isotope ratio mass spectrometer (EA-IRMS) at VIMS as described elsewhere (Brault et al. 2013). Some phytoplankton samples were analyzed at the University of California at Santa Cruz (UCSC) on a on a Carlo Erba EA 1108 elemental analyzer coupled to a Finnegan Delta-Plus isotope ratio mass spectrometer (EA-IRMS).
In this case, δ 13 C and δ 15 N values were averaged and standard deviation was ≤ 0.5, 77% of the time or greater.
Trophic levels were determined using the following equation modified from Hobson and Welch (1992): Where TL consumer is the trophic level of the consumer in question, D c is the δ 15 N of the consumer, 1.7 represents the average δ 15 N of phytoplankton (1.7 ± 0.27‰ standard error), which is assumed to be trophic level 1 and the baseline of the food web in this system, and 3.4 represents the average trophic enrichment factor as recommended in Borgå et al. (2012) (Hobson & Welch 1992, Hobson & Ambrose 1995, Borgå et al. 2012).
Trophic magnification factors (TMFs) were further determined following the procedure in Borgå et al. (2012) to add certainty to the argument of whether or not these chemicals are biomagnifying or not. TMFs were calculated by plotting a regression of trophic level (x) vs. log (concentration +1) (y) where TMF = 10 slope . On average, for a slope of 0 and TMF = 1, it can be assumed that the compound is not biomagnifying, for a TMF > 1, there is some level of biomagnification occurring, and for TMF <1, there is no biomagnification but instead, trophic dilution (Gobas et al. 2009, Borgå et al. 2012.

Statistical Analysis
Data was tested for normality using the Shapiro-Wilks test in RStudio and IBM

Plankton Trends
Linear regressions were run to determine if there were any trends between plankton concentration and δ 13 C, δ 15 N, and/or date. For both δ 13 C and δ 15 N, regressions were run for the whole sample set and then within years. Although it sometimes appears that there is a slightly negative trend between plankton and δ 13 C (decreasing concentration as δ 13 C becomes more enriched/less negative), only BDE-153 vs. δ 13 C for all years was significant (p = 0.045; r 2 = 0.136) (SI  Table 2.14).

Trends of Krill
No significant relationships were found between krill concentration and δ 13 C or samples, so we believe that a larger samples size combined with future improvements on sample collection and analysis will yield better results (Corsolini et al. 2006, Hale et al. 2008, Borghesi et al. 2008).
Average Antarctic silverfish δ 15 N was 10.86‰ ± 0.11 and δ 13 C was -21.44‰ ± 0.02. No statistical analyses could be performed on fish due to lack of detection in samples.

Trophic Levels
All plankton samples were assumed to be trophic level 1 and all other trophic levels were calculated from this baseline. No significant differences were found between the different trophic levels of each krill size class, and as such they were averaged together for a mean trophic level of 1.72 ± 0.05. Myctophid trophic level averaged 3.33 ± 0.06. Antarctic silverfish trophic level averaged 3.70 ± 0.03. Occasional slightly significant differences were found between some sampling years for fur seals trophic levels, likely due to the variability of isotopes, which reflect the variation in diet migratory species are subject to from year to year; thus, trophic levels for fur seals are 3.52 ± 0.07, and 3.76 ± 0.12, respectively.

Trophic Magnification Factors
TMFs were calculated for this food web using all samples with available isotope data from the one year of sampling (2010/11) in which samples were collected from all four groups. TMFs were calculated for a food web consisting of plankton, krill, fish, and fur seal milk, and separately for a scenario excluding fish in case there are discrepancies with the no detects. In both cases, all TMFs were found to be < 1 (range 0.33 -0.87) indicating that there is some level of biodilution or metabolic excretion processes happening within this Antarctic food web ( which had TMFs of 1.6 and 1.2, respectively, and all congeners showed an overall TMF range of 0.7 -1.6. Comparatively, their TMF values for PCBs ranged from 2.9 -11, demonstrating a much greater potential for food web biomagnification (Kelly et al. 2008).
Our TMF range (all samples) is slightly less and on a whole, lower than what was observed in the Arctic. These data provide further evidence that PBDE transport through food webs is species specific and illustrate the usefulness of TMFs inter-ecosystem comparison, in this case between the Arctic and Antarctic. In a time of increasing climate change studies and increased efforts to regulate organic contaminant production and use, it is critical to continue monitoring efforts of pollutants in regions such as Antarctica that are removed from production of synthetic contaminants, but not unaffected.                           Table 2.17 -Results for the two-sample t-test assuming unequal variances for phytoplankton concentration vs. 'mixed' plankton concentration based on the definition that mixed plankton is any plankton sample with a δ 15 N value >2.0‰. P-values presented are from the one-tailed test and a + icon indicates that the mean of phytoplankton was significantly greater than the mean of mixed plankton. The t-test was only run in the cases where both phyto and mixed plankton had > 30% detection of the congener in question.  Table 2.20 -Results for the two-sample t-test assuming unequal variances for krill concentrations in 2007/08 vs. 2010/11 (ng/g lipid). P-values presented are from the onetailed test and a + icon indicates that the mean from 2010/11 is greater than that from 2007/08. The t-test was only run in the cases where there was > 30% detection of the congener in question and the right-hand side is the test excluding sample "Kr24." Table 2.21 -Results for the two-sample t-test assuming unequal variances for juvenile vs. adult krill. P-values presented are from the one-tailed test and a + icon indicates that the mean of juveniles is greater than that of adults. The t-test was only run in the cases where there was > 30% detection of the congener in question and the right-hand side is the test excluding sample "Kr24."  Table 2.22 -Results for the two-sample t-test assuming unequal variances for adult vs. gravid krill. P-values presented are from the one-tailed test and a + icon indicates that the mean of adults is greater than that of gravid krill. The t-test was only run in the cases where there was > 30% detection of the congener in question and the right-hand side is the test excluding sample "Kr24." Table 2.23 -Results for the two-sample t-test assuming unequal variances for adult vs. Thysanoessa sp. P-values presented are from the one-tailed test and a -icon indicates that the mean of Thysanoessa sp. is greater than that of adult krill. The t-test was only run in the cases where there was > 30% detection of the congener in question.     Table 2.27 -Summary of analysis of seal milk concentration vs. breeding season. The first two columns (all years and 2000/01 -2010/11) are linear regressions and a positive, +, indicates a significantly or nearly significantly increasing concentration with time (i.e. breeding season) and a negative, -, indicates a significantly decreasing concentration with time. The third column (2011/12 -2012/13) is results from a two-sample t-test of means assuming unequal variances (one-tailed). In all cases, %detection was >30% and the negative, -, here represents a significant decline in the means between the two years. The asterisk, *, is indicative of the regression being run without a potential outlier to see if any differences were observed. In all cases, the trends remained the same, with some significance actually increasing.  Table 2.28 -Results for the two-sample t-test assuming unequal variances for perinatal vs. non-perinatal milk. P-values are presented and a + icon indicates the mean of nonperinatal milk is greater than the mean of perinatal milk; aicon indicated the mean of perinatal milk is greater than non-perinatal. The t-test was only run in the cases where there was > 30% detection of the congener in question. Sample size of perinatal milk in 2011/12 = 15 and in 2012/13 = 28. Sample size of non-perinatal milk in 2011/12 = 11 and in 2012/13 = 6.