Reproductive Failure in UK Harbour Porpoises Phocoena phocoena: Legacy of Pollutant Exposure?

Reproductive failure in marine mammals has been associated with exposure to persistent organic pollutants (POP), the effects of which were reported to occur at many stages of the reproductive cycle including implantation failure [ 1 ] and foetal death (abortion) in seals [ 2 ], and increased first-born calf mortality in bottlenose dolphins ( Tursiops truncatus ) [ 3 ]. Additionally, sterility caused by stenosis, occlusions and leiomyomas were reported in Baltic seals [ 2 , 4 , 5 ], the presence of luteinised cysts with the potential to impede ovulation were observed on the ovaries of Mediterranean striped dolphins ( Stenella coeruleoalba ) [ 6 ], and severe reproductive dysfunction through the development of cancer and hermaphroditism was documented in St Lawrence Estuary beluga whales ( Delphinapterus leucas ) [ 7 – 9 ]; again all associated with high levels of exposure to POPs, specifically organochlorines (OCs) such as polychlorinated biphenyls (PCBs) and dichlorodiphenyltrichloroethane (DDT). However, in some cases there have been difficulties in linking these proposed toxicological endpoints of reproductive dysfunction to high POP burdens due to the presence of confounding factors. For example, it is not known if the presence of ovarian luteinised cysts and lower fecundity, manifested in a high number of abortions, in Mediterranean striped dolphins that died during a morbillivirus epizootic were caused by high PCB levels, the morbillivirus infection or a combination of both [ 6 ]. In the early seventies, stillbirths and premature pupping in California sea lions ( Zalophus californianus californianus ) were attributed to the contamination by tDDT (total DDT) [ 10 ]. This relationship though was confounded by the presence of an enzootic bacterial disease (leptospirosis) capable of inducing abortions and, in other cases, prematurity was associated with nutritional stress caused by environmental variability [ 11 ]. While the levels of tDDT in California sea lions declined by one order of magnitude from 1970 to 2000 [ 12 ], marine biotoxins such as domoic acid, a naturally occurring algal toxin, has more recently been linked to (contributing to) reproductive failure in sea lions in this region [ 13 ]. The effects of PCBs and DDT on reproduction in the genetically distinct North-east Atlantic continuous system harbour porpoise ( Phocoena phocoena ) population, that ranges from France to Norway (and found primarily on continental shelf waters) including the North Sea [ 14 ], is a particular concern. Although the use and production of PCBs in Europe was phased out in the 1980s, diffuse inputs into the marine environment continue and environmental levels in marine biota (fish and mussels) are either declining slowly, or there is no general improvement [ 15 ]. Harbour porpoises, like other marine mammals feeding at higher trophic levels and with large lipid storage, will have increased risk of individual and population level toxicity from exposure to OCs due to limited capacity to metabolise and excrete these compounds [ 16 ]. Although an initial decline in blubber Σ25CBs concentrations was observed in UK porpoises sampled in the mid-1990s, this plateaued off after 1998 [ 17 ]. In contrast, levels of blubber ΣDDT significantly declined since the early 1990s, reflecting a lack of fresh inputs of this agricultural chemical [ 17 ]. Harbour porpoises are small cetaceans with high parasitic exposure, and individual health status often present energetic “knife-edge” conditions. General health status may be further exacerbated by adding additional health complications arising from high pollutant burdens. This may be reflected in the relationship between PCB exposure and infectious disease mortality in UK harbour porpoises, with significantly higher blubber PCB levels reported in individuals that died from infectious disease compared to physical trauma [ 18 ]. A low estimated pregnancy rate compared to other geographic areas was recently reported for harbour porpoises inhabiting Scottish waters (30–40%, sampling period 1990–2005 [ 19 ]; and for porpoises in western European waters (42%; sampling period 2001–2003; [ 20 ], which possibly suggests reproductive failure/impairment in the population. As the vast majority of individuals sampled within these studies however were stranded animals that died as a result of numerous pathological reasons, such as generalized bacterial infection, pneumonia and starvation, in addition to trauma, it is not known if they are really representative samples of the extant population. In order to investigate this further, the current study assesses pregnancy rates and evidence of reproductive failure in female harbour porpoises, using case (infectious disease) and control (trauma) cause of death groups, and investigates their association with exposure to PCBs and DDT. Confounding factors, such as age, nutritional status and exposure to infectious disease, lead to difficulties in specifically attributing the presence of higher OCs to reproductive disorders in previous studies on marine mammals [ 21 ]. These factors will be controlled for within the current study.

Female harbour porpoises were collected between 1990 and 2012 by the UK Cetacean Strandings Investigation Programme and the UK Bycatch Observer Scheme in England (n = 123), Scotland (n = 117) and Wales (n = 89). The dataset comprised of 329 individuals for which blubber tissue was assessed for quantification of hexane extractable lipid and 25 individual lipid normalised polychlorinated biphenyls. A sub-set, 175 females, was assessed for three lipid normalised organochlorine pesticides. Within the whole dataset, females died from a range of infectious diseases (n = 134) including generalised bacterial infection, (meningo)encephalitis, pneumonia (parasitic, mycotic and bacterial) and gastritis and/or enteritis. In addition to this there were 146 cases of death due to physical trauma and 49 “other” COD cases, mostly being starvation (n = 22) and live stranding (n = 10). The sample was composed of 190 sexually immature and 130 sexually mature females, and for nine individuals maturity status was not determined. Sexually immature females ranged in length and age from 70 to 157 cm (n = 187) and 0 to 6 years (n = 161), respectively, and sexually mature females from 138 to 191 cm (n = 130) and 3 to 21 years (n = 90), respectively (see Fig 1 ). Mean ages of sexually immature and mature females were 1.34 and 8.4 years, respectively ( Table 1 ). Using all the dataset, the average age of females at attainment of sexual maturity was 4.73 years (standard error (SE) = 0.03, n = 250), compared to 4.63 years (SE = 0.08, n = 99) within the infectious disease group and 4.92 years (SE = 0.02, n = 114) within the trauma group. 1 Includes one female that was not assigned to a mature reproductive status category. 2 Includes three individuals that were not assigned to a mature reproductive status category. 3 Includes two females that were not assigned to a mature reproductive status category. Using data obtained outside the conception period for the species in UK waters, pregnancy rates in the sample ranged from 26.5% for females that died from infectious disease to 50% for “healthy” individuals that died of trauma (e.g. bycatch, bottlenose dolphin kills or ship-strike) (see Table 2 ). However, there was no significant difference in the proportion of pregnant individuals between the infectious disease and trauma COD groups (χ 2 = 3.520, df = 1, p = 0.0.061, n = 69). Cases of dystocia or stillbirth (cause of death) were not included in the trauma group within this analysis though for comparisons with other studies, females presenting with cases of “foetal death” were regarded as pregnant. PR was also estimated using data that was obtained outside the conception period (May to September [ 63 ]; see methods ). Using number of corpora scars as an index of reproductive activity (i.e. number of ovulations), a significant positive relationship was observed between (natural logarithms transformed) age and number of corpora scars (p = 0.000, n = 225). Large-scale individual variation was observed in corpora scar number at a given age, and three females in particular presented with higher than expected corpora scar number for their age (see Fig 2 ). These individuals died of starvation (SW1997/178, lactating and mild multifocal eosinophilic endometritis, 4 years of age and presented with 14 corpora scars), dystocia or stillbirth (SW1997/94, lactating and pregnant, 5 years and 15 corpora scars) and infectious disease (SW2003/219a, pregnant, 6 years and 17 corpora scars). Dashed box highlights three females that presented with higher than expected number of corpora scars. Within the whole sexually mature dataset (n = 127; for three individuals the reproductive tracts were not fully assessed), 19.7% showed evidence of reproductive failure. These include cases of dystocia and stillbirth (n = 7), foetal death (n = 4), recent abortion (n = 12), and either the female recently aborted or her calf did not survive (n = 2) (see Table 3 for an outline of definitions for these categories). Cases of infectious disease of the reproductive tract and mammary gland tissue (12 of 127, 9.5% of the total mature group) included acute suppurative endometritis in a recently aborted animal and a Brucella ceti infection of the mammary glands (mastitis) in another female that either aborted or whose calf died. Additionally, cases of reproductive tract and mammary gland infections included endometriosis (bacterial, mild multifocal eosinophilic; n = 2), vaginitis (n = 1), lesions of the vagina, uterus and clitoris (n = 3), a uterine tear secondary to peritonitis (n = 1), pyometra (n = 1), mastitis (n = 1), and finally one 11-year old female porpoise presented with vaginal (possibly leiomyoma) tumours, purulent metritis and reduced uterine lumen. Tumours of the reproductive tract constituted 10.2% (13 of 127) of the total sample. Three malignant tumours (2.4% of the total mature sample) confirmed by histopathology included a squamous cell carcinoma of the cervix in a recently aborted animal (n = 1), a uterine adenocarcinoma (n = 1) and a metastasising adenocarcinoma (n = 1). Benign tumours (n = 10) included clitoral or vaginal papilloma-like lesions of suspected but unconfirmed viral cause (n = 6), vaginal epithelial plaques (n = 2), vaginal wall leiomyoma (n = 1), and, as noted above, an 11-year old female presented with vaginal (possibly leiomyoma) tumours (n = 1). Reproductive dysfunction was also observed in 25% of individuals that should be more representative of the extant population. In the “healthy” mature female group that died as a result of trauma (n = 20) and exhibited a pregnancy rate of 50%, one female had recently aborted while foetal death was observed in another individual; and cases of uterine adenocarcinoma, papillomatous lesion of the clitoral epithelium, and vaginal epithelial plaques possibly of a viral aetiology were reported in three other individuals. Levels of ΣPCBs in sexually immature and mature females ranged from 0.48 to 159.68 mg/kg (n = 190) and 0.40 to 138.83 mg/kg (n = 130), respectively. Mean levels in sexually immature (nulliparous) and mature females were 14.03 and 13.26 mg/kg, respectively ( Table 1 ). Females presenting with one corpora scar (corpus albicans or lutem) ranged from 6 to 10 years (n = 4) in age ( Fig 2 ) and 144–161 cm (n = 7) in length. Mean ΣPCBs in primiparous females, including one recently pregnant and lactating female and two pregnant females, was 18.18 mg/kg (range: 8.4–24.76 mg/kg) ( Fig 3a ). Four resting mature females presenting with one corpus albicans scar ranged from 10.98 to 51.59 mg/kg in ΣPCBs, and where the uteri were assessed for evidence of previous gravidity (n = 3), all were previously gravid. A significant negative relationship was observed between ΣPCBs and corpora scar number in sexually mature females (p = 0.000, n = 77) ( Fig 3a ). Graphs exclude two neonates ranging between 86.9–159.7 mg/kg in ΣPCBs. Blue line represents the threshold level (9 mg/kg ΣPCBs mg/kg lipid) for adverse health effects. Of the aged sample, 98% were females ≤15 years and a significant negative relationship was observed between ΣPCBs and age (p = 0.003, n = 254; Fig 3b ). This suggests that after offloading their pollutant burdens, female harbour porpoises do not re-accumulate PCB burdens similar to that of their pre-parturient levels in later years. All three individuals aged between 20 and 21 years were either currently or recently reproductively active, and ΣPCBs ranged from 2.65–5.15 mg/kg. The relationship between male blubber PCB concentrations and age is shown in Fig 4 . The least-squares linear regression equation suggests an annual accumulation rate of 1.11 mg/kg: ΣPCBs (mg/kg) = 1.11 × age (years) + 13.3, p<0.0001, R 2 = 0.09, n = 257. The SE of the slope was 0.22. Fitted regression line and 95% CI values are shown, n = 257. Annual ΣPCB accumulation rate is 1.11 mg/kg. Within the whole female porpoise sample, highest mean values for ΣPCBs were reported in resting mature (18.5 mg/kg) females followed by sexually immature (14.03 mg/kg), lactating (7.49 mg/kg) and pregnant (6 mg/kg) individuals (see Fig 5a ; Table 1 ). Using a general linear model with Tukey’s post-hoc test, a significant difference in mean ΣPCBs was observed among reproductive status categories (p = 0.000, n = 317), with the resting mature and sexually immature females having significantly higher concentrations compared to the other two groups (p≤0.009). There was no significant difference in mean ΣPCBs between sexually immature and resting females (p = 0.977). This suggests that some resting mature females had not previously offloaded as females were either nulliparous, infertile or suffered from reproductive failure. Where corpora scar data were available and excluding individuals that presented with only one corpora scar, i.e. individuals that may suffer from first-time reproductive failure, the mean values for ΣPCBs in resting mature females was only reduced to 17.1 mg/kg. Of the whole female contaminant sample 47% had ΣPCBs burdens above the threshold for adverse health effects in marine mammals (see Fig 3 ), and the majority of these were sexually immature (68%) and resting mature females (21%). Within the different female reproductive status categories, 52% and 53% of sexually immature and resting mature females had ΣPCBs burdens above this threshold, respectively. The dark horizontal line indicates the median, and the dashed blue line the mean. Outliers are highlighted by blue circles. The average ΣPCBs concentration in sexually immature three and four year olds (i.e. females approaching sexual maturity) was calculated. If sexually mature female harbour porpoises successfully offloaded at least 70% of their PCB pollutant burden to their first born offspring via gestation and primarily lactation [ 29 , 30 ], ΣPCBs concentrations in these reproductively successful females could be as low as 3.3 mg/kg based on an average burden of 11 mg/kg (range 2.53–37.70 mg/kg, n = 22) in females approaching sexual maturity. Using an annual ΣPCBs accumulation rate of 1.1 mg/kg ( Fig 4 ), it would take approximately 7 years before a female who successfully offloaded (approx. 70%) would re-accumulate levels above the average burden in females approaching sexual maturity. As such a large variation was observed in ΣPCBs in sexually immature females (see Figs 3 and 5 ), it is proposed that if sexually mature females had ΣPCBs levels above the 11 mg/kg average concentration in females approaching sexual maturity, then these mature individuals should be regarded as nulliparous, or suspected of infertility or reproductive failure. Within the whole mature sample, 42 of 127 (33%) presented with ΣPCBs levels above 11 mg/kg. These consisted of 6 lactating, 7 pregnant and 29 resting mature female harbour porpoises. All lactating female porpoises died between March and August, i.e. the calving period, and thus did not have sufficient time to offload their ΣPCBs pollutant burden to their nursing calves. Ovarian corpora scar number in lactating females ranged from 1–7 (n = 5), and where data were available animals were 7 to 10 years (n = 3) in age. One lactating female had recently aborted and another presented with a vaginal epithelial plaque of possibly viral aetiology. Of the seven pregnant females with ΣPCBs concentrations above 11 mg/kg, three were single cases of foetal death, dystocia or stillbirth and possible clitoral papilloma-like lesions of suspected but unconfirmed viral cause. All seven pregnant females ranged from 1–17 in ovarian corpora scar number and were 6–15 years old. The oldest pregnant female was a foetal death case with a ΣPCBs concentration of 11.6 mg/kg ( Fig 3b ). The individual with the highest ovarian corpora scar number was only six years in age, had a ΣPCBs concentration of 20.9 mg/kg and died of pneumonia. Although all these seven females were pregnant at the time of death, ΣPCBs concentrations suggest that they were never reproductively successful in the past. Within the resting group 29 of 60 (48%) had ΣPCBs levels above 11 mg/kg. Of these, three recently aborted. A fourth either recently aborted or its newborn calf did not survive and also had a Brucella ceti infection of the mammary gland. Six other resting females (above the 11 mg/kg level) had a range of lesions including a benign leiomyoma in the vaginal wall, bacterial endometritis, vaginal papilloma-like lesions, a vaginal epithelial plaque of possibly viral aetiology, a vaginal wall lesion, and vaginitis. Finally one female (SW1990/94), who died from pneumonia, had vaginal tumours (suspected leiomyomas) in addition to purulent metritis and reduced uterine lumen. Although corpora scar number and age ranged from 1–12 and 3–14 years, respectively, the high PCB burdens (>11 mg/kg) suggest that these 29 resting females never successfully reproduced. Within the whole resting mature sample, 36 of 60 (60%) were assessed for evidence of previous gravidity on necropsy. All showed evidence of previous gravidity (100%). Ovarian corpora scar number ranged from 1–11 (n = 24). These 36 females ranged in ΣPCBs concentrations from 0.95–138.83 mg/kg and 55% of the sample was above the 9 mg/kg threshold—for these individuals corpora scar number ranged from 1–11. This suggests that high contaminant burdens, above the threshold for adverse health effects in marine mammals, did not inhibit ovulation, conception or implantation in female harbour porpoises. Additionally, as 50% of the previously gravid sample had ΣPCBs concentrations above 11 mg/kg level (mean level of sample = 39.29 mg/kg), it can be concluded that these females probably did not offload their contaminant burdens due to either foetal or newborn mortality. Females with ΣPCBs concentrations above 11 mg/kg (n = 18) ranged in ovarian corpora scar number and age from 1–11 (n = 12) and 1–10 years (n = 10), respectively. Within the whole previously gravid sample, 4 of 36 (11.1%) cases were either recent abortion (n = 3), or early calf mortality or recent abortion (n = 1). The highest ΣPCBs concentrations in the whole female sample (159.68 mg/kg) was measured in a neonate (SW1995/102) that died of starvation/hypothermia and stranded on the east coast of England in 1995 ( Table 1 ). All female porpoises with pollutant burdens above 30 mg/kg (n = 33) died from either infectious disease or other non-trauma causes of death. These females ranged from 0–14 years in age (n = 26), though 73% of individuals were ≤ 5 years; of which 83% were sexually immature. The animal (SW1991/14) with the highest ΣPCBs concentration (138.83 mg/kg) within the sexually mature group was 14 years old and died from starvation. Analysis of the data for the three COD groups showed higher mean ΣPCBs concentrations were obtained for females that died as a result of “other” causes (20.7 mg/kg, range = 0.4–159.7 mg/kg), compared to infectious disease (16.62 mg/kg, range = 0.95–138.8 mg/kg) and trauma (8.45 mg/kg, range = 0.48–44.15 mg/kg), though median values were similar between the infectious disease (11.88 mg/kg) and the “other” (11.41 mg/kg) groups (see Fig 5b ). Using a general linear model with Tukey’s post-hoc test, a significant difference in mean ΣPCBs was observed among COD groups (p = 0.000, n = 329), with the trauma group having significantly lower concentrations compared to the infectious disease and the “other” groups (p≤0.001). Individuals that died from starvation constituted 22 of 49 (45%) of the “other” group. Since these females had mobilised their blubber fat reserves prior to death, this probably concentrated some of the less easily mobilised fat soluble PCB congeners [ 39 ]. Only females that died of either trauma or infectious disease were retained within the subsequent analysis, due to the small sample size of the “other” group. Prior to undertaking regression analysis, data were checked to see if they violated assumptions of multinomial logistic regressions. Both ΣPCBs and ΣDDTs showed strong multicollinearity (VIF >10), and thus the variable ΣDDTs was removed from the analysis. The remaining variables did not violate assumptions of sample size, multicollinearity or outliers. All collinearity tolerances were low >0.55 and correlation coefficients were <0.70. The standard deviation of residuals ranged between -2 to 2, and maximum Mahalanobis distances were not greater than the critical value of χ2 for P < 0.001. ΣPCBs was a significant factor in predicting female reproductive status (resting mature, pregnant and lactating). A multinomial logistic regression was statistically significant against the null model (χ2 = 17.558, df = 2, p = 0.000, n = 117), with 15.9% of the variance in the dependent variable explained. Analysis of predicted accuracies misclassified all lactating females as either resting or pregnant. Resting females were more prone to have higher levels of ΣPCBs than pregnant (Odds Ratio (OR) = 0.914, 95% CI = 0.859–0.971, p = 0.004) or lactating females (OR = 0.947, 95% CI = 0.901–0.955, p = 0.033). Results suggested that if the ΣPCBs were to increase by 1 mg/kg then the multinomial log-odds of a female being in a pregnant or lactating state as opposed to resting would be expected to decrease by 8.6% and 5.3%, respectively. A backward stepwise multinomial logistic regression was employed to assess the effects of all other factors, in addition to ΣPCBs, on female reproductive status. There was evidence of effects from nutritional status (length-to-girth ratio), health status and season. Parameters such as region (i.e. country), year, reproductive activity (proxied by number of corpora scars) and other proxies for nutritional status (%lipid and ventral blubber thickness) were not found to be significant. Overall, nutritional status (length-to-girth ratio) was the most important factor in predicting female reproductive status, followed by health status (proxied by cause of death), ΣPCBs and season ( Table 4 ). The model only including these four variables was statistically significant against the null model (χ 2 = 63.53, p = 0.000, df = 8, n = 113), and proved a better fit with 50% of the variance in the dependent variable explained. There was no significant interaction among variables. After adjusting for the effects of confounding variables, ΣPCBs remained a significant predictor of female reproductive status (p = 0.019). Analysis of predicted accuracies suggested that the model correctly classified 66% of the sample, with the “pregnant” group being amongst the highest percentage correctly classification at 74%, though lactating females were still misclassified by 57.7%. Length-to-girth (LN:G) ratio (proxy for nutritional stress), cause of death (COD) class (infectious disease and trauma; proxy for health status), ΣPCBs (mg/kg), and season (season 1 = April to September, and season 2 = October to March). Main effects were observed for nutritional status, with pregnant females more likely to have a lower length-to-girth ratio, i.e. “fatter”, relative to both lactating (regression coefficient B = -13.232, SE = 3.02, p = 0.000) and resting females (regression coefficient B = -9.768, SE = 2.60, p = 0.000). No significant difference was observed among lactating or resting females in nutritional status. Results further suggested that females that died of infectious disease were less likely than trauma cases to be in a lactating state as opposed to resting (regression coefficient B = -1.837, SE = 0.667, p = 0.006) or pregnant (regression coefficient B = -1.685, SE = 0.774, p = 0.03). There was no significant difference among resting or pregnant females in this parameter. Season was significant as lactating females were more likely to be sampled during season 1 (April to September), i.e. during the breeding period, than Season 2. The relationship between ΣPCBs in resting females compared to pregnant (p = 0.065) and lactating (p = 0.066) females was almost significant, having controlled for other factors. With every 1 mg/kg increase in ΣPCBs the multinomial log-odds of a female being in a resting state as opposed to pregnant or lactating increased by 6.8% and 5.7%, respectively—though in both cases the 95% CI for the odds ratio included the null (1.00). Including age in the model reduced the available sample to 79. Age was not found to be a significant predictor of female reproductive status (p = 0.252) and also, the final model did not include cause of death (proxy for health status). However, this may be a reflection of a reduction in sample size. Nutritional status remained a highly significant predictor (p = 0.000), followed by ΣPCBs (p = 0.013) and season (p = 0.022). The backward stepwise multinomial logistic regression model was run again, substituting ΣDDTs for ΣPCBs, and including variables nutritional status (length-to-girth ratio), health status and season. The model was significant (p = 0.000), though as the sample size was reduced to 64 females this violated the assumptions of logistic regression. Three variables were retained in the model, including nutritional status (p = 0.001), health status (p = 0.009) and ΣDDTs (p = 0.042).

We assessed for evidence of reproductive failure both directly, e.g. through observations of foetal death, and indirectly by using individual PCB burdens. Using these data, results suggested that reproductive failure could have occurred in up to 39% or more of mature females sampled. Above all, female’s health status played an important factor in the occurrence of reproductive failure. However, within the group of “healthy” UK harbour porpoises that died as a result of trauma we still observed a lower reproductive output compared to other (less contaminated) populations; with a lower pregnancy rate (almost half the rate) and a higher average age at attainment of sexual maturity observed. PCBs may have negatively impacted foetal or newborn survival as previously gravid resting female harbour porpoises had not successfully offloaded their pollutant burdens, via gestation and primarily lactation. Though ultimately difficulties arose in showing casual associations between cases of reproductive dysfunction and ΣPCBs concentrations due to female’s capabilities in offloading pollutant burdens. Nevertheless, when all ages were considered 47% of individuals in the current study had ΣPCBs concentrations above the threshold for adverse health effects in marine mammals, which included 52% of sexually immature females and 53% of resting mature females. Whether or not the stranded harbour porpoise sample is reflective of the extant population will continue to be debated. Even though previous research reported that bycaught harbour porpoises in the North Sea were in a poorer health status than their more northern counterparts [ 44 ], which may explain the high occurrence of diseased stranded harbour porpoises in this region. Interestingly, the higher pollutant levels in some resting females provided information on their lifetime reproductive success, and as a result samples from necropsied animals provided more information than a prevue of reproductive status at the time of death. Preliminary results suggest that reproductive dysfunction in UK porpoises may be related to PCB exposure occurring either through endocrine disrupting effects or via immunosuppression and increased disease risk. Declines of major organochlorine concentrations in biota have been slow due to global cycling and long-half lives of pollutants [ 116 ], and as of 2005 1.1 million ton of PCB containing equipment, corresponding to 350,000 ton of PCB containing liquid, still required disposal by EU Member States [ 117 ]. Taking this into consideration, as well as inherited maternal pollutant burdens in first born offspring and generational epigenetics effects, raises concerns about the current and future population-level effects of PCBs on the continuous-system North-east Atlantic harbour porpoise population.

The UK Cetacean Stranding Investigation Programme conducts work on UK strandings under contract to the UK Department of Environment, Food and Rural Affairs. It has appropriate licenses from the relevant authorities (Natural England, Scottish Natural Heritage and Natural Resources Wales) to allow it to collect and hold carcasses and samples from European Protected Species, in line with UK legislation enacted under the EU Habitats Directive. No animals were killed or harmed for the purposes of this study. Gross necropsy and tissue sampling protocols followed European Cetacean Society guidelines [ 22 ]. Condition of individuals was fresh-to-moderate decomposition on necropsy, with the majority in fresh (68%) or slight (27%) decomposition. Basic data collected from each animal included stranding location, date, sex, total length, ventral blubber thickness (along the ventral abdomen in front of the dorsal fin) and girth (in front of the dorsal fin). Causes of death (COD) were determined by specific diagnostic criteria (see [ 18 , 23 ], and individuals were categories in to three COD groups: infectious disease, trauma (bycatch, boat/ship strike, bottlenose dolphin attacks and dystocia) and others (live stranding, starvation, neoplasia and not established). During gross examinations, the reproductive tracts of sexually mature individuals were examined for the presence of a foetus, evidence of female aborting a calf, dystocia and foetal death. Additionally in mature individuals, reproductive tracts were assessed for cases of sterility and infertility from stenosis, occlusions, leiomyoma and endometriosis, and other reproductive tract abnormalities. Where possible, occurrences of disease and infection were confirmed by bacteriology, virology and histopathology assessments. In order to assess previous gravidity, the development of the blood vessels in the broad ligaments and the presence of striations within the smooth musculature of the uterine wall were noted, as well as the degree of uterine involution. Mammary glands were examined for evidence of lactation via gross examination and, in some individuals, histological assessment of mammary tissue. Frozen blubber and teeth samples (-20°C) and 10% neutral buffered formalin fixed reproductive material were retained for further investigations. Blubber samples were analysed by the Centre for Environment, Fisheries, and Aquaculture Science Laboratory for quantification of hexane extractable lipid and wet weight concentrations (mg/kg) of 25 individual chlorobiphenyl (IUPAC numbers: 18, 28, 31, 44, 47, 49, 52, 66, 101, 105, 110, 118, 128, 138, 141, 149, 151, 153, 156, 158, 170, 180, 183, 187, 194) and a range of organochlorine pesticides (p,p’-DDE, p,p’-TDE (also known as p,p’-DDD) and p,p’-DDT) using previously established and validated protocols based on international standardised methodologies (see [ 17 , 18 ]). Concentrations were converted to a lipid weight basis (mg/kg lipid) using the proportion of hexane extractable lipid in each individual blubber sample, and the sum of the concentrations of the 25 PCB congeners (ΣPCBs mg/kg lipid) and three DDT congeners (ΣDDT mg/kg lipid) were determined. For all analyses, appropriate quality control materials (certified or laboratory reference materials) were analysed within each sample batch in order that the day-to-day performance of the methods could be monitored [ 24 ]. The straightest and least worn teeth were selected for age estimation from the middle section of the lower jaw. Age was determined by analysing growth layer groups (GLGs) in the dentine tissue of dental samples, following Lockyer [ 25 ]. Teeth were decalcified, sectioned and the most central and complete sections (including the whole pulp cavity) were selected from each tooth, stained, mounted on glass slides, and allowed to dry. GLGs were counted under a binocular microscope and on enhanced computer images of the sections. All readings were initially made blind (with no access to other data on the animals) and replicate counts were made by at least two readers. In cases where there was disagreement (>1 year disparity), teeth were re-examined by readers and an age and/or an age range was agreed. As ages were recorded by a number of different researchers, cross-calibration exercises were carried out. Fixed ovarian samples were assessed externally for the presence of corpora scars (corpus luteum and albicans) and then hand-sectioned into 0.5–2 mm slices and examined internally under a binocular microscope for the presence of additional corpora scars. Females were considered sexually mature if their ovaries contained at least one corpus luteum or albicans. Total numbers of corpora scars (number of ovulations) were counted, which was used an index of reproductive activity (after Murphy et al. [ 26 ]). Assessment of female reproductive status follows the procedures and terminology recommended by the International Whaling Commission [ 27 ]. Females were classified into five reproductive states: (1) sexually immature, (2) pregnant (foetus present), (3) pregnant and lactating, (4) sexually mature and lactating and (5) resting mature (not pregnant or lactating) [ 26 ]. The pregnancy rate (PR) was estimated by calculating the proportion of pregnant females in the sexually mature sample. Only females sampled outside the conception period for the species in UK waters (May-September; [ 19 ]) were included in the analysis, due to the increased possibility that embryos or small foetuses were not detected during early stages of gestation. Average age at sexual maturity (ASM) and its variance was estimated using the sum of fraction of immature algorithm [ 28 ]. ASM =  j + ∑ i = j K p i Variance  ( s 2 ) =  ∑ p i q i N i − 1 Where, if I i ≠ N i , p i = I i / N i , and q i = (M i ) /N i; if I i = N i , p i = (I i - ½)/N i , and q i = (M i + ½)/N i , and if M i = N i , p i = (I i + ½)/N i , and q i = (M i - ½)/N i . Confidence interval (at p = 0.05) = ASM ± 1.96 √s 2 j = the first indeterminate age class k = the last indeterminate age class p i = fraction of immature specimens in age class i q i = fraction of mature specimens in age class i (p i + q i = 1) I i = number of immature specimens in age class i M i = number of mature specimens in age class i N i = number of specimens in age class i (N i = I i + M i ) Primiparous and multiparous female cetaceans can offload their organochlorine pollutant burden through transplacental transfer and also via lactation as OCs are mobilised from blubber to lipid rich milk [ 3 ]. Individual studies reported that in bottlenose dolphins, the majority (c. 80% of OCs) of a female’s contaminant burden is believed to be transferred to first born calves during early lactation [ 29 ]. A similar estimation was reported for striped dolphins where 72–91% of total body burdens were offloaded during lactation, whereas only 4–9% of total body burdens wereoffloaded during gestation ([ 30 ] reported in [ 31 ]). Further, depending on age and female reproductive history, OC transfer estimates of 60–100% during lactation and 4–10% during gestation were reported in long-finned pilot whales ( Globicephala melas ) [ 32 ]. In order to assess offloading of pollutant burdens in mature female harbour porpoises, linear regression analysis was undertaken with ΣPCBs (mg/kg) as the dependent variable and corpora scar number and age (years) as independent variables. Pattern of offloading was compared between females in different reproductive status categories. Natural log transformations were undertaken prior to regression analysis so as to stabilise the variance. Accumulation of PCBs in harbour porpoises was assessed using available data from male porpoises that stranded along UK coastlines between 1990 and 2012 (n = 263), as males do not offload their contaminant load during their lifetime. A linear model was applied to the relationship between blubber PCB concentrations (mg/kg) and age (years) (after Hall et al. [ 33 ]) and, because of large-scale individual variation in ΣPCBs at a given age, outliers more than three standard residuals from the fitted line were removed from the analysis. General linear models were undertaken with Tukey’s post-hoc tests to compare mean (LN transformed) ΣPCBs mg/kg concentrations among female reproductive status categories and also COD groups. Blubber “total PCBs” toxicity threshold concentration for the onset of physiological endpoints in marine mammals of 17mg/kg lipid (as Aroclor 1254) [ 34 ] was converted to the equivalent value of 9 mg/kg ΣPCBs mg/kg lipid (see [ 35 ]). The threshold was derived by Kannan et al. [ 34 ] and is based on experimental studies of both immunological and reproductive effects in seals, otters, and mink. Using thresholds in this way warrants caution owing to possible differences in species sensitivities, though it should provide a benchmark for interpreting whether associations between reproductive activity and PCB exposure are biologically significant [ 18 ]. It should be noted though that threshold levels from which effects are to be expected would be lower in embryos and calves than in adults [ 36 ]. To further investigate the effects of ΣPCBs on reproduction, we used a multinomial logistic regression model with mature female reproductive status (1 = resting mature, 2 = pregnant, 3 = lactating) as the dependent variable and ΣPCBs (mg/kg lipid) as a covariate predictor. In addition, various other predictors or confounding factors, including both categorical and continuous, were included in the model. These comprised of ΣDDT (mg/kg lipid), region (1 = England, 2 = Wales, 3 = Scotland), year, season (season 1 = April-September, season 2 = October-March), age (years), corpora scar number (index of reproductive activity), cause of death (0 = infectious disease, 1 = trauma; proxy for health status), and three proxies of nutritional status. These included percentage of hexane extractable lipid (%lipid), ventral blubber thickness (mm), and body length-to-girth ratio—the latter was reported to be a significant predictor of survival in post-released (live-stranded) common dolphins [ 37 ]. All data were tested to establish if they violated assumptions of the multinomial logistic regression; though this type of regression does not assume normality, linearity, or homoscedasticity. A backward stepwise regression method was employed with interactive effects as oppose to a forward stepwise, as the latter is more likely to exclude predictors involved in suppressor effects [ 38 ]. The likelihood ratio test statistic was used to compare the fit of two models, one of which was the null model. Statistical analyses were performed using SPSS version 22.0 and SigmaPlot version 10.0 (SPSS Inc., Chicago, IL, USA).

(DOC) Click here for additional data file.