SRS is located in west-central South Carolina, USA. The former nuclear production site encompasses 78,000 ha (778 km2) and has had restricted public access since 1951 (White and Gaines 2000). In 1972, the SRS was classified as the first U.S. National Environmental Research Park where the ecological impacts of anthropogenic manipulation of the environment could be studied (White and Gaines 2000). Approximately 90% of the SRS is undeveloped. The undeveloped area is comprised of managed pine timber production (54%), wetlands (23%), upland hardwood and mixed forest (11%), grasslands (9%), and upland scrub forest (3%) that are interspersed across the landscape (White and Gaines 2000). Between 1951 and 1988, industrial production activities on the SRS resulted in several releases of contaminants into the surrounding habitats on site. These perturbations included pollution with anthropogenic compounds, such as radionuclides, coal fly-ash runoff, and other trace elements into the surrounding ecosystems (White and Gaines 2000). As a result of restricted public access, the SRS has become a refuge for many wildlife species, including raccoons, which are known to occupy both contaminated and uncontaminated habitats on site (Gaines et al. 2002).
For this study, two discrete areas within the SRS were selected to collect raccoons in August and December 2013: an area known to be contaminated with trace elements and a reference site that was uncontaminated (Fig. 1). The first location—the D-Area ash basins—was the site of a coal-fired power plant that deposited sluiced fly ash into a series of basins from 1953 to 2012 while the plant was operational (U.S. Department of Energy 2012). These basins drained into Beaver Dam Creak, a tributary of the Savannah River, and the surrounding wetlands (Bryan et al. 2012; Gaines et al. 2002). The basins and creek watershed have been shown to have increased levels of aluminum (Al), As, Cd, Cr, Cu, iron (Fe), Hg, manganese (Mn), nickel (Ni), Se, and Zn in the water and biota (Cherry et al. 1979; Rowe et al. 1996). Thus, raccoons were collected in the forested areas surrounding the basins as well as riparian habitats adjacent to Beaver Dam Creek.
Upper Three Runs Creek (UTR) was selected as the control site for this study and is a natural creek originating from a spring-fed aquifer system that has not been impacted by industrial activities on the SRS (Bowers 1997). Beaver are common in the area and numerous beaver ponds and Carolina bays exist in the vicinity of UTR. Thus, a combination of wetland and creek habitats as well as a mixture of mixed hardwood and pine forest existed on each site. The similar habitat composition between the two sites allowed us to assume similar food availability to raccoons across the study system. Although some sublethal effects have been observed in some prey species exposed to coal fly ash, common prey items of raccoons are present in the ash basin utilized in this study (Hopkins et al. 1998, 1999). The area of UTR where raccoons were trapped was approximately 16 km from the D-Area ash basins, eliminating the possibility of individual raccoons traveling between the two areas, as raccoon home ranges generally are <2–4 km2, although home ranges up to 10 km in diameter have been reported (Gehrt and Fritzell 1997). UTR, referred to hereafter as the reference site, has been used as an undisturbed reference site in several publications (Burger et al. 2002; Morse et al. 1980).
To evaluate whether raccoons were sentinels of trace element contamination, we tested individual parameters (age, site, sampling season and site*sampling season interaction) as predictors of elevated concentration of eight trace elements in the liver of raccoons collected in contaminated versus reference sites. For trace elements that exhibited differences among contaminated and reference sites, we considered them as predictor variables to assess changes in morphometry, hematology, and parasite burden of individual raccoons across contaminated and reference sites.
Trapping and Necropsy Methods
In August and December 2013, raccoons were captured using steel wire box live traps (Tomahawk Live Trap, Hazelhurst, Wisconsin), baited with whole corn, fish meat, and/or plaster scent tabs. Traps were placed strategically along roads, water bodies, or in a grid within 500 m from the centroid of the reference site or the fly ash containment basins to target animals utilizing resources known to be uncontaminated or contaminated, respectively. After capture, animals were transported from the field to the Savannah River Ecology Lab animal care facility, where they were immobilized using Telazol (2.2 mg/kg IM) and subsequently euthanized via intracardiac injection of potassium chloride administered while the animal was under general anesthesia. All animal handling practices and euthanasia were performed in accordance with University of Georgia Animal Care and Use guidelines under protocol A2012 12-010-Y2-A3.
Twenty-six raccoons were trapped throughout the two trapping seasons, and sex and estimated age classes were recorded. Raccoons were grouped into three age classes: I (1–2 years), II (3–4 years), and III (5–6 years) based on tooth wear. We captured 11 raccoons in August [contaminated site n = 4 (all males); reference site n = 7 (6 males and 1 female)] and 15 raccoons in December [contaminated site n = 11 (5 males and 6 females); reference site n = 4 (all males)]. Seven raccoons were aged at 1–2 years (contaminated site n = 4; reference site n = 3), 12 aged at 3–4 years (contaminated site n = 7; reference site n = 5), and 7 aged at 5–6 years (contaminated site n = 4; reference site n = 3).
Raccoons were weighed and standard morphometric measurements were collected, from which body mass/nose-anus length ratio was calculated. Approximately 3 ml of blood was collected, two blood smears were made, and 2 ml of blood were transferred to ethylenediaminetetraacetic acid (EDTA) Vacutainer tubes (Becton, Dickinson and Co., Franklin Lakes, NJ). Tubes were refrigerated and shipped on ice overnight to the Georgia Veterinary Diagnostic and Investigational Laboratory (VDIL; University of Georgia, Tifton, GA) within 48 h of collection. Tissue samples were taken from the liver, spleen, kidneys, ileum, and colon and fixed in 10% neutral buffered formalin for histopathology analyses. Liver samples for trace element quantification were placed in Whirl–pak bags and frozen at −20 °C until analysis. The intestinal tract, including colon, was frozen and stored for helminth identification.
Trace Element Analysis
Trace element [Cr, Ni, Cu, Zn, As, Se, Cd, Pb, uranium (U), and total mercury (THg)] analysis was conducted on all liver samples. Wet weights of liver samples were recorded and tissues subsequently freeze-dried and homogenized. For trace element analysis, approximately 250 mg of dry sample was used for microwave digestion (MARSX Xpress, CEM Corporation, Matthews, NC) with 10.0 ml trace metal-grade nitric acid (70% HNO3). After digestion, samples were brought to a final volume of 15.0 ml with Milli-Q (18 MΩ) water before analysis was performed by inductively coupled plasma mass spectroscopy (Nexlon 300X ICP-MS; Perkin Elmer, Norwalk, CT) according to the Quality Assurance/Quality Control (QA/QC) protocols outlined in EPA Method 6020A (USEPA 2007). Minimum detection limits (MDL; ppm) for each element were: Cr (0.54), Ni (0.84), Cu (0.67), Zn (5.15), As (0.13), Se (0.18), Cd (0.15), Pb (0.05), and U (0.09). Element concentrations lower than MDL were substituted by half of the corresponding MDL as standard procedure. Included in the digestion analysis for quality control purposes a certified reference material (TORT-3 lobster hepatopancreas; National Research Council, Ottawa, ON, Canada), a blank, and a digestion replicate were each run in conjunction with every 20 samples. For elements in certified reference materials, mean percent recoveries ranged from 85 to 105%.
For THg analysis subsamples of the freeze-dried and homogenized liver tissue (30–50 mg) were analyzed using a DMA 80 Dual Cell Mercury Analyzer (Milestone, Shelton, CT) according to U.S. Environmental Protection Agency (EPA) method 7473, which has an instrument detection limit (IDL) for this method of 0.01 nanograms (ng) of total mercury. For quality assurance, we included with each set of 10 samples a replicate, a blank, and two standard reference materials (SRM; TORT-2 lobster hepatopancreas and PACS-2 marine sediment, National Research Council of Canada, Ottawa, ON). Solid SRMs were used to calibrate the instrument. Method detection limits (MDLs; threefold the standard deviation of procedural blanks) averaged 0.0233 parts per billion (ppb) dry mass. Mean percent recoveries of THg for the SRMs TORT-2 and PACS-2 were 102.1 ± 1.6 and 103.8 ± 3.9, respectively. All THg concentrations were converted to parts per million (ppm) presented on a dry mass basis to be comparable to the other trace element concentrations.
Complete blood cell counts were conducted for 24 of 26 raccoons, excluding the other two animals in which blood samples were clotted. Anticoagulant (EDTA) blood tubes were processed for red blood cell (RBC) and total white blood cell (WBC) counts, mean corpuscular volume (MCV), hematocrit (HCT), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), hemoglobin (HGB), platelet counts, mean platelet volume (MVP), plasma protein, and reticulocyte and absolute reticulocyte counts by automation using the Advia 2120i Hematology System (Siemens Biomedical Solutions Diagnostics, Tarrytown, NY). Blood smears were stained with Wright’s (Romanosky) and Giemsa Stain (counter stain) procedure (Poly Scientific, Bay Shore, NY), and examined for cellular morphology, WBC differential cell counts, and blood parasites.
Histopathologic analyses were conducted in liver, spleen, kidney, ileum, and colon samples collected from the 11 raccoons trapped in August 2013. These analyses were not conducted in raccoons from December 2013 because of logistical constraints. After fixation, sections of these tissues were processed routinely and embedded in paraffin. Sections that were 5-μm thick were stained with hematoxylin and eosin. Slides were examined by a single, blind observer (LF) for inflammation, fibrosis, degeneration/necrosis, pigment deposition, and other changes. A grading scheme was created and each tissue was graded subjectively for multiple parameters on a 0 (no changes), 1 (mild changes), 2 (moderate changes), and 3 (marked changes) scale for a total of 103 categorical variables. We included in the final analyses only 16 variables that were found to have substantial variation in the breadth of abnormalities and 3 binary categories (Table A.1; Online Resource 1).
Upon thawing, intestines were carefully opened along their length with scissors and contents were initially screened for helminths by hand. Intestines and digested material were rinsed through a 180-µm sieve to ensure all parasites were collected; the solid material left after rinsing was sorted again under a dissecting microscope. Parasites were preserved in absolute ethanol and identified to genus or species (when possible) using a compound microscope. Total number of helminths was determined for each raccoon using a dissecting scope.
Distributions of all variables were tested for normality (Shapiro–Wilkes test, p < 0.05), and all further statistical tests were conducted using log-transformed data. Based on a backward elimination procedure, stepwise regression models were used to test for differences in trace element exposure between raccoons from the contaminated and reference sites. Previous to their definitive inclusion in stepwise regression models, we tested for the univariate effect of site, age and sampling season on trace element accumulation of raccoons. We only included male data to test the univariate effect of age and sampling season because of the unbalanced sample size of sexes distributed among contaminated and reference sites. We broke down age by site and season to rule out the potential association of either site or season with animal age in the univariate analysis.
To determine the impact of trace element accumulation and sampling season on morphometry (body size variation—PC1), hematology (RBC, WBC, and platelet counts), and parasite burden (helminth abundances) of individual raccoons, we considered the spectrum of trace elements that were at higher concentrations in livers of raccoons from the contaminated site compared to the reference site as predictor variables. We used stepwise regression models with backward elimination, including morphometry, hematology, and parasitology responses in separate models. To test the impact of trace elements on raccoon morphometry and hematology, we considered all raccoons from contaminated and reference sites. Before the inclusion of trace element concentrations in the stepwise regression models, Pearson correlation coefficients were calculated to determine collinearity between the predictor variables. To reduce the number of variables analyzed, we conducted principal component analyses (PCA) independently on the morphometric dataset. Principal components that accounted for at least 70% of the total variance and whose eigenvalues were >1 were retained for interpretation and incorporation into multivariate regression models.
To test for relationships between the degree of histological lesions and trace element concentrations in raccoon livers from contaminated and reference sites in August 2013, we used multinomial logistic regression models only for 16 of the 103 categorical histological variables (16 variables which had substantial variation in degree of abnormalities). To test whether binary response variables (presence/absence of hepatic, ileum and colon granulomas, and blood parasites) correlated to contaminant exposure, simple logistic regression models were performed.
Nonmetric multidimensional scaling (NMDS) was used to evaluate visually the variation in community composition and abundance of intestinal helminth species harbored by all raccoons from contaminated and reference sites. We used abundance data for each helminth species from individual raccoons to calculate the Sorensen (Bray–Curtis) distance metric among raccoons from both sites. Two-dimensional analysis was chosen by minimizing stress, a goodness-of-fit measure of the mismatch between the rank-order of Euclidean distance in the ordination space, and the distance among species indicated by the Bray–Curtis dissimilarity matrix. Relationships between trace element concentrations and the ordination scores were graphically explored.
Stepwise regression models with backward elimination were used to test for the influence of trace element concentrations and sampling season on (non-log-transformed) parasite abundances of raccoons collected at both contaminated and reference sites. Because male and female raccoons have previously been found to have different parasite abundances (Cole and Shoop 1987; Kresta et al. 2009), and we had an unbalanced sample size between sexes in contaminated (9 males vs. 6 females) and reference (10 males vs. 1 female) sites, we only included male data in the stepwise regression models. In further stepwise regression models, we included WBC counts as an additional predictor of parasite abundance to test how biomarkers of immune response and trace element concentrations may influence parasite infestation in raccoons.
Statistical tests were conducted in the software R v. 2.15.2 (R Core Team 2012) and statistical significance was set at α = 0.05. NMDS ordination and vector fitting were conducted using the “ecodist” package in R (Goslee and Urban 2007).