LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date If requested. DATE DUE DATE DUE DATE DUE MAR {$372608 DEC 1 0 2088 —-O-.9l1—1—q 6/07 p:/ClRC/DateDue.indd-p.1 _ ,. _r“_-~—«_——¥ AN ECOLOGICAL RISK ASSESSMENT FOR GREAT HORNED OWLS AND EAGLES EXPOSED TO POLYCHLORINATED BIPHENYLS AND TOTAL DDT AT THE KALAMAZOO RIVER SUPERFUND SITE, MICHIGAN By Karl Daniel Strause A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Zoology 2007 ABSTRACT AN ECOLOGICAL RISK ASSESSMENTFOR GREAT HORNED OWLS AND EAGLES EXPOSED To POLYCHLORINATED BIPHENYLS AND TOTAL DDT AT THE KALAMAZOO RIVER SUPERFUND SITE, MICHIGAN By Karl Daniel Strause Selection Of receptors is a key element of effective risk and natural resource damage assessments. This is especially critical when site-specific field studies are employed. The great horned owl (Bubo virginianus; GHO) has advantages over other species as a key tertiary terrestrial receptor that can be used as an integrated measure of exposure to residues in a multiple-lines-of—evidence approach. The methods described herein exploit attributes of the GHO including its propensity to nest in artificial nesting platforms, which allows for better control of experimental conditions than normally experienced in studies of wildlife. The GHO was used in a multiple-lines-of-evidence study of polychlorinated biphenyls (PCBS) and p,p ’-dichlorodiphenyltrichloroethane (DDT) exposures at the Kalamazoo River SUperfund Site (KRSS) in Kalamazoo and Allegan Counties, Michigan. Over the course of five yrs, 48 artificial and six natural GHO nests, Covering approximately 14 active territories along approximately 38 km of river floodplain, were monitored for activity at the KRSS. The study examined risks from total PCBS, including 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQWHO-AVm), and total DDTS (sum of DDT/DDE/DDD; ZDDT) by measuring concentrations in eggs and nestling blood plasma. Dietary modeling was also completed to estimate potential ingested dose for the contaminants of concern (COCS) from site foraging activities. An ecological risk assessment compared concentrations of the (COCS) in eggs, plasma, or diet to toxicity reference values to generate hazard quotient values descriptive of potential risk to resident raptor populations. Productivity/relative abundance measures for KRSS GHOS were compared with other GHO pOpulationS. Egg shell thickness was measured to assess effects of p, p ’—dichlorodiphenyldichloroethylene (p, p ’-DDE) on egg viability. Tissues data, dietary exposUre estimates, and productivity/relative abundance measures indicated no population level effects were present at the Upper reach of the Kalamazoo River Superfund Site, closest to the sources of PCB contamination to the River. Bald eagles (Haliaeetus Ieucocephalus) residing at the Kalamazoo River Site were also studied using the multiple-lines-of—evidence approach. Observations of reduced productivity and elevated contaminant concentrations in eagle eggs and nestling plasma collected from the site indicated that cOntaminant exposures were likely at the threshold for adverse population effects for resident bald eagle populations. Additionally, data bases describing the concentrations of total PCBS in eggs and nestling plasma of great horned owls and total PCBS and p, p ’-DDE in eggs and nestling plasma of bald eagles from the Great Lakes region were used to develop a relationship to predict concentrations of PCBS and DDE in eggs from measure concentration in nestling plasma. An accurate conversion factor to translate nondestructive plasma-based contaminant concentrations to comparable egg-based concentrations will prove valuable to risk assessors investigating the potential effects of chemical exposures to raptors. For My Mother — Marie Theresa (Feairheller/Slawecki) Strause she always knew I had it in me iv ACKNOWLEDGEMENTS I have many people to thank for this degree. I am indebted to Drs. Matthew Zwiernik and John Giesy for providing the opportunity to study at Michigan State University and the world famous Aquatic Toxicology Laboratory. My good fiiend and mentor Dr. David Hohreiter prepared me for this journey, and as is his forte, Dave gave me a dead-on, incredibly accurate portrayal of what I was getting myself into. The Kalamazoo River Study Group provided generous financing for this research project. I benefited greatly from the combined expertise and tutelage provide by the members of my steering committee; Drs. Richard Balander, Steven Bursian, Thomas Burton and James Sikarskie. Simon Hollamby and Heather Simmons provided crucial expertise during my first field season. I had the most dedicated and talented technical assistants one could ask for in Pemela Moseley and William Claflin. Donna Pieracini, my #1 Volunteer, played an instrumental role in helping me complete my field work. Michelle Moffat provided expertise and many appreciated hours toward photographic and video documentation of the field methods used in this study. I received logistical support from the Kellogg Biological Station and I will be eternally grateful to Nina Consolotti, Barbara Baker, Kim Shupp, Joe Johnson and Philip Goodall for the help and advice that they were always willing to provide. My good friends Cyrus Park, Jeremy Moore, Mick (top-10) and Sarah Kramer, and Herr Dr. Markus and Uli Hecker kept me going when things got rough. Dr. Sook Hyeon Im shared her expertise in the extraction lab and made a chemist out of me, despite my attempts to wreck havoc at the Farm. Dr. Eva-Maria Muecke supported me through many agonizing days of statistical analyses with compassionate instruction and fresh baked scones. I would like to commend the staff of the Center for Food Safety and Toxicology, Department of Zoology, and Center for Integrative Toxicology for their professionalism, expertise and unwavering support throughout my Six years in East Lansing. Most notably Jennifer Sysak, Susan Dies, Margaret Nicholas, Lisa Crafi, Alice Ellis, and Amy Swagart helped my hold things together when they seemed ready to fall apart at any moment. Drs. Dan Villeneuve and Katie Coady gave generously of their time and helped me more than they realize. I wish to thank Professors Richard Merritt and Stephen Boyd for their compassion and the helping hand they extended when I was up to my neck in an academic morass. Dr. William Bowerman shared his climbing expertise and tips on how to stay out of Coast Guard rescue helicopters. I benefited greatly from the genuine interest and accumulated years of knowledge freely shared with me by my new friends Mr. Jack Holt and Mr. Sergej Postupalsky. My colleagues at the ATL were always willing to provide technical and emotional support and I thank Patrick Bradley, Dr. Denise Kay, Dr. Alan Blankenship, Dr. Paul Jones and the one and only Dr. John Newsted for every minute of their time they gave to me during my efforts to get things nailed down straight. Dr. Shufang Shi provided inspiration and insight into what is really important in life. My parents gave their love and support through the good, the bad, and the ugly. I love them very much and know how lucky I am to have them. Thank you all, I couldn’t have done it alone! vi TABLE OF CONTENTS KEY TO ABBREVIATIONS AND UNITS OF MEASURE ........................... xvi CHAPTER 1 SITE-SPECIFIC ASSESSMENTS OF ENVIRONMENTAL RISK AND NATURAL RESOURCE DAMAGE BASED ON GREAT HORNED OWLS Introductron4 Methods9 ResultsZO Dlscussmn26 References Cited ....................................................................... 33 CHAPTER2 ‘ PLASMA TO EGG CONVERSION FACTOR FOR EVALUATING PCB AND DDT EXPOSURES IN GREAT HORNED OWLS AND BALD EAGLES Introductlon41 Methods43 ResultsSO D13cussron74 References Cited ....................................................................... 86 CHAPTER3 RISK ASSESSMENT OF GREAT HORNED OWLS (BUBO VIRGINIAN US) EXPOSED TO POLYCHLORINATED BIPHENYLS (PCBS) AND DDT ALONG THE KALAMAZOO RIVER, MICHIGAN Introduction94 Methods96 Resultle8 Discu331on127 References Cited ..................................................................... 136 CHAPTER4 EVALUATION OF RISK ASSESSMENT METHODOLOGIES FOR EXPOSURE OF GREAT HORNED OWLS (BUBO VIRGINIAN US) TO PCBS ON THE KALAMAZOO RIVER, MICHIGAN Introduction144 Methodsl46 Resultsl64 Dlscussmnl83 References Cited ...................................................................... 200 vii ICIIILAII?7IIEEIZ.ES RISK ASSESSMENT OF BALD EAGLES (HALIAEET US LE UCOCEPIML US) EXPOSED TO POLYCHLORINATED BIPHENYLS (PCBS) AND DDT ALONG THE KALAMAZOO RIVER, MICHIGAN Innoducfionn.u.u.H.H.u.n.n.n.n."uuuunuunuunu"NHHNHHAHNHHHHHHHZOS Ivlketlrr3 100 g), individual time points and collection sties were examined together to reconcile the frequency of occurrence of larger prey species when remains of the same prey item were present in multiple samples. Multiple prey item identification keys were utilized for comparative identification including owl pellet identification keys (Carolina Biological Supply Company, Burlington, NC, USA) and the vertebrate Skeletal collection from the MSU 15 museum. Avian remains were identified with the aid of MSU Kellogg Biological Station bird sanctuary personnel. Dietary composition was based on the frequency of occurrence of all identifiable prey items and compiled on the basis of absolute (%) frequency of occurrence and relative (%) composition of biomass. Prey Item Sampling Once identified as a principal component of owl diet, prey species were collected from the most contaminated GHO foraging areas and a reference location. Species selection and sample sizes were determined based on sensitivity and power analyses of preliminary data and expected contribution to GHO dietary exposures. For this study, a total of 171 small mammals including meadow voles (Microtus pennsylvanicus), white-footed mice (Peromyscus leucopus), deer mice (Peromyscus maniculatus), meadow jumping mice (Zapus hudsonius), eastern chipmunks (T amias striatus), short tail shrew (Blarina brevicauda) and masked shrews (Sorex cinereus) were sampled from six locations at two time points. Also sampled from these locations were arthropods, including four orders each of terrestrial and aquatic invertebrates. Larger mammals such as red squirrels (Tamiasciurus hudsonicus), fox squirrels (Sciurus niger), eastern cottontails (Sylvilagus foridanus), muskrats (Ondatra zibethicus), and mink (Mustella vison) as well as passerine species including the American robin (Turdus migratorius), house wren (T roglodytes aedon) and tree swallows (T achycineta bicolor) were sampled opportunistically throughout the study area. Sampling techniques varied depending on target species. 16 Hazard Evaluation Here we provide methodologies for site-specific assessment of the hazard of chemicals in soils to GHOS based on' a multiple—lines-of—evidence approach that could enable well- informed decisions regarding potential remedial actions and determination of natural resource damages (EPA 1997; Fairbrother 2003). Such an approach has been applied at other contaminated sites for wildlife species such as mink (Bursian et al. 2003) or tree swallows (Custer et al. 2005). However, to our knowledge, this is the first case in which field studies and multiple-lines-of—evidence have been utilized to assess potential risks of PCBS and ZDDT to GHOS. The multiple-lines-of—evidence included several methods of estimating exposure. Direct observations of population densities and reproductive success were made and compared to the results of the hazard assessment. Exposure of GHOS to these compounds was characterized in two ways. Concentrations of PCBS in the diet were calculated fiom the site—specific dietary composition and concentrations in prey items, as well as measured concentrations of PCBS and XDDT in eggs and blood plasma of nestling GHOS. Each measure of exposure was compared to the threshold for a toxic effect determined from the literature and expressed as a toxicity reference value (TRV) (U SEPA 1998). An ecological risk assessment (ERA) was conducted by calculating hazard quotients (HQS). Hazard quotients were determined by dividing the measured or predicted concentration in the diet, egg or nestling blood plasma by the appropriate TRV. Toxicity reference values for the GHO hazard assessment were selected following criteria outlined by the USEPA (1995) and were derived from chronic toxicity studies in which a dose-response relationship was observed in the species of concern, or alternatively a 17 closely related species (Strause et al. 2007a,b). Toxicity reference values were selected from studies that examined effects of PCBS and EDDT in owls and eagles, however, there were no suitable studies that used the GHO as the test species. The PCB TRVS were based on a feeding study with screech owls (0tus asio) that examined productivity endpoints (McLane and Hughes 1980). The ZDDT TRVS were based on field studies that examined productivity and eggshell thinning endpoints in bald eagles (Haliaeetus leucocephalus) (Elliott and Harris 2002), and shell thinning endpoints in the barn owl (T yto alba) (Klaas et al. 1978). Aside from applying an uncertainty factor of 3 to derive the PCB lowest observable adverse effect concentration from a validly determined no observable adverse effect concentration, application of additional uncertainty or extrapolation factors to our selected TRVS was not necessary. Comparisons were made between, within, and among, sites, individuals and prey species. The multiple-lineS-of- evidence approach can be optimized, based on information needed, level of effort available, and site-specific criteria and characteristics. Population Density and Reproductive Success The final line of evidence included measurements of population health. Health of the GHO population was asSessed through the evaluation of productivity including nest success, number of nestlings per nest, fledging success, and nestling age and growth measurements as well as GHO abundance. Much of the information on population dynamics was acquired in conjunction with the owl banding and nest monitoring tasks described above. However, an additional effort was made to evaluate GHO population health using vocalization surveys. 18 Vocalization surveys The results of vocalization surveys and triangulation were used to identify active breeding territories, locations of nests, Site use, relative abundance and confirmation of fledging success. A combination of vocalization survey methods were used including an active method in which GHO hoots were broadcast to provoke responses (Frank 1997; Brenner and Karwoski 1985), and a passive or silent survey method during sensitive lifestage events and the periods when GHOS were most active (e.g., just before and during mating) (Rohner and Doyle 1992). Relative abundance determinations were made based on the number of individuals responding on a per survey basis. Pair vocalization responses and post survey observations were evaluated and referenced to literature-based foraging areas to delineate active territories. Nestling fledge success was determined by nestling vocalizations post banding and/or subsequent visual confirmation. All positive responses and non-responses were recorded. For the positive responses, sex and age (adult or juvenile) and global positioning system coordinates of river location, and approximate azimuth values (compass readings) of response origin were recorded for the purpose of location by triangulation. Post surveys, targeted areas of 150 m radius were Searched systematically by foot for signs of GHO activity (whitewash and castings) to determine roost sites and to locate nests. Active or potentially active roost sites and nest lOcations were recorded using a GPS receiver. To minimize disturbance to incubating birds, ground activity during the months of February and March was limited to 0Ccupancy identification of previously located nests and known nesting platforms. 1\Testing activity was confirmed visually by spotting scope from predetermined l9 monitoring locations no less than 50 m from the nest or by overhead flights using fixed wing aircraft. RESULTS Over the course of five years, monitoring efforts were completed at 48 artificial and six natural nests covering approximately 14 active territories and approximately 38 km of river. Nesting activity was observed at 20 individual nests and resulted in 25 nesting attempts. Of the 20 nests utilized by GHO, five were natural nests, including four appropriated nests of other avian species and a tree cavity nest. Artificial nesting Platforms were successful in attracting GHOS to preferred study areas in the floodplain. In fact, nesting activity did not occur in natural nests in those tenitories for which artificial nesting platforms were in place. Reuse of artificial nesting platforms over IIlultiple seasons allowed for the minimization of temporal and spatial variability and allowed easy access for researchers. The robust owl population was ideal for evaluating tl'le multiple-lines-of-evidence at both the target and reference Sites. Detailed methods and results for contaminant analysis and exposure assessments Eire provided in separate papers (Strause et al. 2007a,b). The results are summarized here to illustrate the effectiveness of the methods and as an example of sample sizes that may be necessary to detect differences between the study and reference locations. Great l'1()rned owl exposures were assessed by collecting both fresh eggs (destructive) and l”lestling blood plasma (non-destructive). Sample availability varied among years and locations (Table 1.1). 20 Table 1.1. Sampling scope and blood plasma and egg summary. Description of sampling effort by year and location. Note that for 2000 - 2002 the Fort Custer and Trowbridge sampling areas were monitored for productivity, thus only addled eggs were collected. Reference Sample Sites Target Sample Sites Trowbridge Allegan Year Ceresco Fort Custer Impoundments SGA 2000 Active Nests 0 1 2 Plasma 0 1 0 Eggs , 0 0 3 \ Data Targetedl NM P, NP, RA P, NP, RA E 2 ()01 Active Nests l 1 2 1 Plasma 0 1 4 0 Eggs 1 0 0 2 Data Targetedl E P, E, NP, RA P, E, NP, RA E 2 002 Active Nests 2 0 4 2 Plasma 1 0 3 2 Eggs 2 0 1 5 \ Data Targetedl E, NP P, E, NP, RA P, E, NP, RA E, NP 2003 Active Nests 1 l 2 1 Plasma 1 0 1 2 Eggs 1 . 1 3 0 \ Data Targetedl E, NP E, NP E, NP E, NP 2004 Active Nests 0 0 2 2 Plasma 0 0 0 0 Eggs 0 0 3 2 \ Data Targetedl E, NP E, NP E, NP E, NP NM=not monitored; P=productivity; E=egg sampling; NP=nestling plasma sampling; =relative abundance 21 A total of 40 egg and blood plasma samples were collected. Of the 24 eggs collected during the study, five were from the reference areas and 19 were from the target areas. Blood plasma was collected from 16 individual nestlings, this included four samples from the reference areas, and 12 samples from the target areas. Statistically significant differences in concentrations of total PCBS were observed among locations (reference versus target) for the predicted dietary exposure and for total PCB and EDDT concentrations in GHO eggs and blood plasma (Strause et al. 2007a,b). PCB concentrations in eggs were significantly greater at the Allegan SGA compared to the Trowbridge and reference sites (p<0.05), and total DDT concentrations were Significantly different among each of the Allegan SGA, Trowbridge and reference sites (p<0.03). These differences were the result of exposures to mean PCB concentrations in floodplain soil of approximately 0.17 mg PCBS/kg, dw (dry weight) in reference areas and approximately 15 mg PCBS/kg, dw in the target areas. Differences in dietary composition between the reference and target areas also were observed (Figure 1.3). Differences between predicted dietary exposures (average potential daily dose) were largely the result of significant differences in concentrations in the prey items (Table 1.2), and were not a product of differences in dietary prey item composition. Concentrations of P CBS and ZDDT in eggs were significantly different between reference and target areas (Figure 1.4). Diet-based HQ values calculated from geometric mean total PCB _ concentrations in prey animals collected from the most contaminated areas of the KRSS floodplain were less than 1.0 at the target locations. Tissue-based HQS calculated from the geometric mean concentrations of total PCBS and ZDDT in eggs were 5 1.5 at all target locations (Strause et al. 2007a). In addition, a well defined relationship was 22 .mooméoom 88 mEEnam a??? $3: 268 Set ooqobdooo mo 5:26on 60809 as 33%on 8mm .mmmbwnm ESE >89 98 8:09 an wogfloe 3.030 mo 22:89:00 EH05 .mA 03E 8% u 5 $2526; 335 202822 exec omumsv<>aofi8u 3 32% “some / Asa: :Bonxcgfibmobohbctommmm aaogpseaamssfim Ame u 5 .820 an. A80 5% See :Bocxcguahscménnmm , E; c 20283 my»? a». .000 9 V... o o o r0000; .00.... v o o o» 099) Ag w v vnououoo “00900.0; owa=c<>30fiufi O O O 0 «an». #03... . as. a e HESE esogbhflbmotflxoctummmm 23 Table 1.2. PCB concentrations in GHO dietary items sampled from proximal foraging areas. Waterfowl were not sampled based on sensitivity analysis. Trowbridge Fort Custer Mean total PCBS Mean total PCBS (i std dev ) (i std dev) Dietary items N (mg/kg)2 N (mg/kg)2 Small mammall 21 *0.13 i 0.16 18 0.021 i 0.042 House wren adult 6 *3.57 i 2.30 5 0.09 i 0.032 American robin 8 *1.14 i 1.44 4 0.091 d: 0.65 Tree swallow 5 *11.46 1 11.90 2 1.49 i 0.15 Shrew 17 *1.31 i- 0.94 16 0.009 i 0.005 Muskrat 7 *0.07 i- 0.03 4 ‘ 0.01 i 0.01 1 . . . . Includes; white-footed mouse, deer mouse, jumping mouse, meadow vole, red squrrrel, and eastern chipmunk. 2 On a wet-weight basis. * Indicates a significant difference between sites at p<0.05. established for total concentrations of PCB in eggs and those in nestling blood plasma (Strause et al. 2007c). The statistical power of the tests were such that statistically Significant differences (Type I error (or) of 0.05 and Type 11 error (0) < 0.20) in exposure could have been detected with as few as 4 eggs or 12 samples of nestling blood plasma per area. Relative abundance of GHOS per river km was significantly different between the referrence and target areas of the Kalamazoo River, but reproductive productivity per Clefended territory (number of nestlings fledged per active nest) was not Significantly different between study sites. During the three-year period (2000 — 2002) in which abundance measurements were completed at the KRSS, significant differences in the 24 g 31 ~ ‘3. it PCBS .52 3) I DDT E 16- 5 14. 3 2 u 12- c 8 c 10» o O 8, ’7 ‘0. a 6' 5 4- '8 s 2' a.“ Ceresco Fort Custer Trowbridge Allegan SGA Sample Site Figure 1.4. Concentrations of PCBS and ZDDTS in GHO tissues (egg). Median concentrations and associated one standard deviation of samples collected at four locations. Sampling locations presented from upstream to downstream (left to right) with the two reference sites upstream of point sources (Ceresco and Fort Custer), and two target sites downstream of point sources (Trowbridge and Allegan State Game Area). 25 number of adult, juvenile and paired responses of GHOS were observed, with the Trowbridge impoundment (target area) having greater numbers of each response compared to Ft. Custer (reference) (Table 1.3). The Trowbridge impoundment had a greater number of active nests (6 versus 1) and greater overall recruitment to floodplain populations with six successful fledglings compared to one successful fledgling at Ft. Custer, however, the mean rates of productivity for the two sites were identical at 1.0 fledgling per active nest (Table 1.3). DISCUSSION Use of the GHO as a key receptor species in ERAS is predicated on its relatively great exposure potential, broad applicability among geographic regions and ecosystems, and ease of study. While the first two characteristics have been well documented for the GHO, its nocturnal nature and aggressive disposition may have previously dissuaded researchers from using the species in previous ERAS. For this study, the GHO proved to be a relatively easy and effective receptor species with applicability to both screening level and site-Specific baseline ERAS. The single most important outcome of this study Was our ability to induce breeding pairs Of GHOS to occupy nesting sites centrally located Within areas of interest and reuse those nesting sites over multiple years. This provided for conservative and worst case exposure assessment evaluations and risk chalracterizations. These behavioral attributes of the GHO offered significant advantages over other top terrestrial food web receptors including all other large resident raptors. 26 Table 1.3. Relative abundance and reproductive productivity of GHOS. Abundance based on adult, juvenile and pair responses to great horned owl calls broadcast from predetermined locations throughout sampling areas. Ft. Custer Trowbridge Relative Abundance1 N2=24 N2=22 Adults Mean Response Rate3 Total4 1.31 2.76 Foraging5 0.85 l .64 Paired6 0.47 1.13 Juveniles Response Frequency7 11, (%) Fledgling 1 (4%) 8 (36%) Productivity Active Nests l 6 F ledglings l 6 Fledglings/N est 1 1 .—I Derived from hoot call/response surveys completed at dawn and dusk. N=number of complete surveys. Mean response rate is averaged across N completed surveys. Includes discrete responses from both individual and paired owls. Includes responses from unpaired individuals only. GUI-huh) Includes responses from paired (male + female) owls only. Response frequency of fledgling owls, n=number of surveys with at least one fledgling begging call response, (%)= (n) / number of surveys (N2). The strategy of conducting initial surveys to identify occupied GHO territories, followed by reconnaissance of active owl territories within the areas of interest was effective for locating existing owl territories. However, successful location of optimally lOcated natural nests (in relation to contaminated floodplain foraging habitats) was rare. Site-Specific characteristics indicate that this may have been due to an absence of aVailable nests in the floodplain of the study area because other nest building species are more limited in nesting habitat and prefer upland areas. Artificial nesting platforms were placed inside the perimeter of defended territories and centrally located within the areas 27 of interest. The density of nesting platform placement ranged from one to three nests per territory at 500 to 1000 m intervals. Over the five-year study period, nesting activity was identified in 81% of the territories containing nesting platforms. Nesting activity occurred in 90% of territories in which paired owls were identified. Both relative abundance and pair response were useful predictors of nesting potential. Nesting activity in natural nests was never observed in those territories in which artificial nesting platforms were placed at least 6 mo prior to expected egg drop. Platforms were utilized preferentially even in instances where appropriate natural nests were available and/or when natural nests were utilized the year prior to artificial nest placement. Breeding and reproductive success of nesting pairs utilizing artificial nesting platforms was comparable to natural nest-based reproductive studies. Of the territories in which a platform had been placed, GHOS initiated incubation 65% of the time. In a 28-yr study, in a proximal geographic-a1 area of similar characteristics, it was found that 62% of owls in occupied territories initiated incubation. For that study, the resulting annual mean productivity expressed as the number of young/occupied territory varied moderately from 0.5 to 1.1 and the number of fledglings/successful nest was a consistent value of 1.7 (Holt 1996). In this study, reproductive productivity in both the reference and target was similar to the Holt study. The annual mean number of juveniles fledged per occupied territory ranged from 0.5 to 1.6 and the number of fledglings/successful nest was 1.4. Post-fledge survival was successfully monitored in all territories in which active surveys were systematically performed and nestlings were banded. Monitoring of survival of juveniles by their begging responses to broadcasted adult hoot calls was possible as late as 24 wk 28 post-banding. An option for longer-term monitoring of juvenile survivorship includes the temporary attachment of a radio transmitter prior to fledging from the nest. In this study, several methods of estimating exposure were applied. For select nests, fresh eggs were sampled shortly after identification of incubation activity, while other territories were monitored for productivity. Nestlings from these territories as well as nestlings from egg sampling territories (re-nesters or completed incubation of partial clutch sampling) were banded, 7 ml of blood collected, general health determined, and select morphological measurements taken. Prey remains (including pellets) were collected from active nests, the base of the nest tree, and beneath nearby feeding perches for all nests in which fledgling productivity was monitored. Pellet and prey remains analyses identified 285 individual prey items. In order to determine which type of egg sampling would have the least effect on territorial and Site-wide productivity, fresh eggs were collected using two different sampling approaches. Either the entire clutch was collected to induce re-nesting or a single egg was left to induce continued incubation of the remaining egg(s). When the total clutch sampling approach was used, two of four pairs re-nested and produced three young. For nests in which the most recently laid eggs were left for continued incubation, three of seven pairs continued incubation. Two of the nests each produced one nestling and the third nest was destroyed by severe weather. In all, 24 eggs were sampled from 12 different territories. Over the five-year study, the territories targeted for egg collection varied. Overall, the egg sampling effort targeted 24 territory-years in which 15 tenitory-years contained incubation activity. A territory-year is a level of effort term defined as any one 29 territory monitored over one year. Thus, a single territory monitored over four years or four territories monitored over a single year both involve the same level of effort, 4 territory-years. The cumulative number of sampled eggs would have increased to 30 eggs had the entire clutch been collected for all territories targeted for fresh egg collections. Conclusions resulting from each of the lines of evidence examined in this study were consistent between and among sites. Contaminant exposure based on both dietary- and tissue-based methodologies produced similar results of significantly different COC exposures for the downstream target vs. the upstream reference area. However, the results of the hazard assessments indicated that GHO populations residing in the floodplain were not at risk for effects induced by total PCBS or ZDDT in contaminated soils. Maximum HQ values of <10 (diet exposures) and <15 (egg exposures) indicated that exposure of GHOS to the COCS in Kalamazoo River were below or near the threshold for effects (Strause et al. 2007a,b). Confidence in the risk conclusions was further strengthened by site-specific measurements of productivity, abundance and nestling grth and success. Population parameters for target area owls were not significantly different from the upstream reference areas, and were similar to those expected in a healthy environment. The mean rate of 1.0 fledgling per active nest observed at both locations was consistent with productivity measures for healthy mid- westem GHO populations residing in upland habitats (Holt 1996). Additionally, measures of site-use (abundance) indicated the target area populations at Trowbridge were near the carrying capacity for undisturbed GHO habitats (Houston et al. 1998). This consistency across each of the multiple lines of evidence for both measurement and 30 assessment endpoints combined with the relative certainty of each measurement, the minimal impact on the receptor and environment, and the level of effort expended, highlights the utility of the GHO as a receptor in this and possibly other ERAS and natural resource damage assessment investigations. Here we have provided an overview of the advantages of the GHO as a site- specific surrogate species for the determination of potential risk of contaminants in terrestrial ecosystems. We have given a short overview of a case history. The space available here was limited. For that reason, neither the methods nor the results could be fully described. Detailed methods in the form of standard operating procedures are available from the authors. In addition, detailed results of the asseSSments are published elsewhere (Strause et al. 2007a, b, c). Acknowledgements The authors would like to thank the students and staff of the Michigan State University Aquatic Toxicology Laboratory, including Paul Jones, Dan Villeneuve, Patrick Bradley, Pam Moseley, William Claflin, Donna Pieracini, Ryan Holem, and Scott Kissman as well as personnel of the Michigan State University Veterinary Clinic, including Jim Sikarskie, Simon Hollemby and Heather Simmons, personnel of the Michigan State University, Kellogg Biological Station, most notably Nina Consolotti, Joe Johnson, and Michael Klug. We would like to recognize the assistance provided by the US Fish and Wildlife Service, East Lansing Field Office, most notably Lisa Williams, Dave Best and Charles Mehne as well as the staff of the MI Department of Natural Resources at Fort Custer and the Allegan State Game Area including William Kosmider, John Lerg, and Mike Baily. 31 We thank Ray Adams of the Kalamazoo Nature Center. Funding for this project was provided by the Kalamazoo River Study Group through an unrestricted research grant to J. P. Giesy (PI) of Michigan State University. Animal Use All aspects of the study that involved the use of animals were conducted in the most human way possible. Toward this end, all aspects of the study were conducted using standard operating procedures that had been approved (AUF #S 02/10-030-00; 03 /04- 044-00; 08/03-105-00) by the MSU All University Committee on Animal Use and Care (AUCAUC). All of the necessary state and federal approvals and permits obtained for the project (Michigan Department of Natural Resources (MDNR) Land Use Permit #A- 02-09; MDNR Scientific Collecting Permit #SC1220; USFWS Migratory Bird Scientific Collection Permit #MB081272-l) are on file at the MSU-ATL. 32 REFERENCES Adamcik RS, Todd AW, Keith LB. 1978. Demographic and dietary responses of great horned owls during a snowshoe hare cycle. The Canadian Field-Naturalist 92: 156-166 American Omithologists Union. 1983. Checklist of North American Birds, 6th ed. Washington, DC, USA Austing GR. 1964. The World of the Red Tailed Hawk. Living World Books, JB Lippincott CO., Philadelphia, PA, USA Austing GR, Holt JB Jr. 1966. The World of the Great Horned Owl. Living World Books, JB Lippincott Co., Philadelphia, PA, USA Baumgartner FM. 193 8. Courtship and nestling of the great horned owl. The Wilson Bulletin 50: 274-285 Blankenship AL, Zwiemik MJ, Coady KK, Kay DP, Newsted J L, Strause K, Park C, Bradley PW, Neigh AM, Millsap SD, Jones PD, Giesy, JP. 2005. Differential accumulation of PCB congeners in the terrestrial food web of the Kalamazoo River superfund site, Michigan, USA. Environ Sci Technol 39:5954-5963 Bohm RT. 1985. Use of artificial nests by great grey owls, great horned owls, and red- tailed hawks in northeastern Minnesota. The Loon 57:150-152 Brenner FJ, Karwoski D. 1985. Vocalizations as an index of abundance for the great horned owl (Bubo virginianus). Proc P Acad Sci 59: 129-130 Bursian SJ, Aulerich RJ, Yamini B, Tillitt DE. 2003. Dietary Exposure of Mink to Fish From the Housatonic River: Effects on Reproduction and Survival. Final Report. Weston Solutions, West Chester, PA, USA Burton JA. 1984. Owls of the World. Tanager Books, Dover, NH, USA Council on Environmental Quality. 1972. Development of a Continuing Program to Provide Indicators and Indices of Wildlife and the Natural Environment. Ecology Program, Smithsonian Institution, Final Report. Washington, DC, USA Craighead JJ, Craighead FC Jr. 1956. Hawks, Owls and Wildlife. Stackpole Publishing Co. Harrisburg , PA, USA Custer CM, Custer TW, Rosiu CJ, Melancon MJ, Bickham J W, Matson CW. 2005. Exposure and effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin in tree swallows 33 (Tachycineta bicolor) nesting along the Woonasquatucket River, Rhode Island,USA. Environ Toxicol Chem 24: 93-109 Elliott JE, Harris ML. 2002. An ecotoxicological assessment of chlorinated hydrocarbon effects on bald eagle populations. Rev Toxicol 4: 1-60 Errington PL. 1930. The pellet analysis method of raptor food habits study. The Condor 32: 176-186 Fairbrother A. 2003. Lines of evidence in wildlife risk assessment. Hum Ecol Risk Assess 9: 1475-1491 Frank RA. 1997. Effects of Dieldrin on Great Horned Owls at the Rocky Mountain Arsenal National Wildlife Refuge. MS. Thesis, University of Wisconsin, Madison, WI, USA Hayward JL, Galusha JG, Frias G. 1993. Analysis of great horned owl pellets with rhinoceros auklet remains. The Auk 110:133-135 Henderson CL. 1992. Woodworking for Wildlife, Homes for Birds and Mammals. Minnesota’s Bookstore. St. Paul, MN, USA Henny CJ. 1977. Birds of prey, DDT, and tussock moths in Pacific Norhtwest. Trans North Am Wildl Nat Resour Conf 42-: 397-411 Henny CJ, Meeker DL. 1981. An evaluation of blood plasma for monitoring DDE in birds of prey. Environ Pollut (Series A) 25: 291-304 Hoffman DJ. 1995. Wildlife Toxicity Testing. In: Hoffman DJ, Rattner BA, Burton GA Jr. and Cairns J Jr. (eds), Handbook of Toxicology, pp 47-69. Lewis Publications, Boca Raton, FL, USA Holt JB Jr. 1996. A banding study of Cincinnati area Great Horned Owls. The Journal of Raptor Research 30: 194-197 Houston CS, Francis CM. 1995. Survival of great horned owls in relation to snowshoe hare cycle. The Auk 112: 44-59 Houston CS, Smith DG, Rohner C. 1998. Great Horned Owl. In: Poole A and Gill F (eds), Birds of North America No. 372. pp 1-27. The Birds of North America Inc., Philadelphia, PA, USA International Joint Commission. 1991. Report of the Ecological Committee’s Biological Effects Subcommittee to the Great Lakes Science Advisory Board. Proceedings of the “Expert Consultation Meeting on Bald Eagles”. Windsor, ON, Canada 34 Johnsgard PA. 1988. North American Owls Biology and Natural History. Smithsonian Institution Press, Washington, DC, USA ' Kay DP, Blankenship AL, Coady KK, Neigh AM, Zwiemik MJ, Millsap SD, Strause K, Park C, Bradley P, Newsted JL, Jones PD, Giesy JP. 2005. Differential accumulation of polychlorinated biphenyls in the aquatic food web at the Kalamazoo River superfund site, Michigan. Environ Sci Technol. 39:5964-5974 Klaas EE, Wiemeyer SN, Ohlendorf HM, Swineford DW. 1978. Organochlorine residues, eggshell thickness, and nest success in barn owls from the Chesapeake Bay. Estuaries 1: 46-53 Marti C. 1974. Feeding ecology of four sympatric owls. The Condor 76: 45-61 Marti CD. 1987. Raptor food habits studies. National Wildlife Federation, Scientific and Technical Series. 10: 67-80 Marti CD, Kochert MN. 1996. Dietary and trophic characteristics of great horned owls in southwestern Idaho. Journal of Field Ornithology 67: 499-506 Mauro L. 1987. Raptor physiology — hematology and blood chemistry. In: Pendleton BG, Millsap BA, Kline KW and Bird DA (eds), Raptor Management Techniques Manual. Natl Wildl Fed Sci Tech Ser No. 10, National Wildlife Federation, Washington, DC, USA McLane MAR, Hughes DL. 1980. Reproductive success of screech owls fed Aroclor 1248. Arch Environ Contam Toxicol 9: 661-665 Millsap SD, Blankenship AL, Bradley PW, Jones PD, Kay DP, Neigh A, Park C, Strause KD, Zwiemik MJ, Giesy JP. 2004. Comparison of risk assessment methodologies for exposure of mink to PCBS on the Kalamazoo River, Michigan. Environ Sci Technol 38: 6451-6459 Mehne C. 1994. Unpublished Data on Contaminant Concentrations in Bird Eggs Collected From the Kalamazoo River Basin. From Dr. C. Mehne, Kalamazoo, MI, USA Neigh AM, Zwiemik MJ, Blankenship AL, Bradley PW, Kay DP, MacCarroll MA, Park CS, Jones PD, Millsap SD, Newsted J W, Giesy JP. 2006a. Exposure and multiple lines of evidence assessment of risk for PCBS found in the diets of passerine birds at the Kalamazoo River Superfund Site, Michigan. Hum Ecol Risk Assess 69: 395-415 Neigh AM, Zwiemik MJ, Bradley PW, Kay DP, Park CS, Jones PD, Newsted JL, Blankenship AL, Giesy JG. 2006b. Tree swallow (Tachycineta bicolor) exposures 35 to polychlorinated biphenyls at the KalamazOo River Superfund Site, Michigan, USA. Environ Toxicol Chem 25: 428-437 Nero RW. 1992. New great horned owl longevity record. The Blue Jay 50: 91-92 Rohner C. 1996. The numerical response of great horned owls to the snowshoe hare cycle: consequences of non-territorial ‘floaters’ on demography. The Journal of Animal Ecology 65: 359-370 Rohner C, Doyle F I. 1992. Methods of locating great horned owl nests in the boreal forest. The Journal of Raptor Research 26: 33-35 Rusch DH, Meslow EC, Keith LB, Doerr PD. 1972. Response of great horned owl populations to changing prey densities. J. Wildl Manage 36: 282-296 Sheffield SR. 1997. Owls as Biomarkers of Environmental Contamination. In: Second International Symposium, Biology and Conservation of Owls of the Northern Hemisphere. In: Duncan JR, Johnson DH and Nicholls TH (eds), Winnipeg MB, Canada, USDA Forest Service 1997 General Technical Report NC-190 Stickel J, Wiemeyer SN, Blus LJ. 1973. Pesticide residues in eggs of wild birds: adjustment for loss of moisture and lipid. Bull Environ Contam Toxicol 9: 193- 196 Strause KD. 2006. An Ecological Risk Assessment for Great Horned Owls (Bubo virginianus) and Bald Eagles (Haliaeetus leucocephalus) Exposed to Polychlorinated Biphenyls and total DDTS at the Kalamazoo River Superfund Site, Michigan. PhD. Thesis. Michigan State University, East Lansing, MI, USA Strause KD, Zwiemik MJ, Im SH, Bradley PW, Moseley PP, Kay DP, Park CP, Jones PD, Blankenship AL, Newsted J L, Giesy JP. 2007a. Risk assessment of great horned owls exposed to polychlorinated biphenyls (PCBS) and DDT along the Kalamazoo River, Michigan. Environ Toxicol Chem (Submitted) Strause KD, Zwiemik MJ, Newsted J L, Neigh A, Millsap SD, Park CS, Moseley PP, Kay DP, Bradley PW, Jones PD, Blankenship AL, Giesy JP. 2007b. Comparison of risk assessment methodologies for exposure of great horned owls (Bubo virginanus) to PCBS on the Kalamazoo River. Integrated Environmental Assessment and Management (Submitted) Strause KD, Zwiemik MJ, Irn SH, Newsted J L, Kay DP, Bradely PW, Williams L, Blankenship AL, Giesy JP. 2007c. Plasma to egg conversion factor for evaluating PCB and DDE exposures in great horned owls and bald eagles. Environ Toxicol Chem (submitted) 36 Sundlof SF, Forrester DJ, Thompson NP, Collopy MW. 1986. Residues of chlorinated hydrocarbons in tissues of raptors in Florida. J Wildl Dis 22: 71-82 Tabaka CS, Ullrey DE, Sikarskie JG, DeBar SR, Ku PK. 1996. Diet, cast composition, and energy and nutrient intake of red-tailed hawks (Buteojamaicensis), great horned owls (Bubo virginianus), and turkey vultures (Cathartes aura). Journal of Zoo and Wildlife Medicine 27: 187-196 USEPA (U .S. Environmental Protection Agency). 1993. Wildlife Exposure Factors Handbook, Volume I of II. EPA-600-R-93-187a. Office of Health and Environmental Assessment, Washington, DC, USA USEPA (U .8. Environmental Protection Agency). 1994. ECO Update: Field Studies for Ecological Risk Assessment. EPA-540-F-94-014. Office of Solid Waste and Emergency Response, Washington, DC, USA USEPA (U .8. Environmental Protection Agency). 1995. Great Lakes Water Quality Initiative Criteria Documents for the Protection of Wildlife: DDT, Mercury, 2,3,7,8-TCDD, PCBS. EPA-820-B-95-0082. Washington, DC, USA USEPA (U .8. Environmental Protection Agency). 1997. Ecological Risk Assessment Guidance for Superfirnd: Process for Designing and Conducting Ecological Risk Assessments - Interim Final. EPA-540-R-97-006. Office of Solid Waste and Emergency Response, Washington, DC, USA USEPA (U .8. Environmental Protection Agency). 1998. Guidelines for Ecological Risk Assessment. EPA-630/R-95/002F. Risk Assessment Forum, Washington, DC, USA USEPA (U .S. Environmental Protection Agency). 1999. Issuance of Final Guidance: Ecological Risk Assessment and Risk Management Principles for Superfund Sites. OSWER Directive 9285 .7-28 P, Washington, DC, USA Voous KH. 1988. Owls of the Northern Hemisphere. MIT Press, Cambridge, MA, USA 37 Chapter 2 Plasma to egg conversion factor for evaluating PCB and DDT exposures in great horned owls and bald eagles. Authors: Karl D. Strause:r Matthew J. Zwiemik? Sook Hyeon Irn,I John L. Newsted,§ Denise P. Kay,§ Patrick W. Bradley,l Alan L. glankenship, § Lisa L. Williams ll, John P. Giesy, lZoology Department, Center for Integrative Toxicology, National Food Safety and Toxicology Center, Michigan State University, E. Lansing, MI 48824 IDK Science R&D Center, DK B/D, 27-4 Gongduck-dong, Mapogu, Seoul, Korea §ENTRIx, Inc., Okemos, MI 48823 " United States Fish and Wildlife Service East Lansing Field Office, 2651 Coolidge Road, East Lansing, MI 48823 #Biology and Chemistry Department, City University of Hong Kong, Kowloon, Hong Kong, SAR, China. 38 ABSTRACT The benefits of nondestructive sampling techniques, such as plasma sampling, to directly measure contaminant exposure levels in at-risk or protected raptor populations are many. However, such assays are generally inconsistent with the most certain source of toxicity reference values that are based on feeding studies and quantified as dietary or in ovo (egg-based) concentrations. An accurate conversion factor to translate nondestructive plasma-based contaminant concentrations to comparable egg-based concentrations will prove valuable to risk assessors investigating the potential effects of chemical exposures to raptors. We used data bases describing the concentrations of total polychlorinated biphenyls (PCBS) in great horned owls (GHO; Bubo virginianus) and total PCBS and p, p ’-dichlorodiphenyldichloroethylene (p, p ’-DDE) in bald eagles (Haliaeetus leucocephalus) from the Great Lakes region to develop a relationship to predict concentrations of PCBS and DDE in eggs. To develop a robust predictive relationship, all of the source data included concentrations of both total PCBS and/or DDE for nestling blood plasma and egg samples collected from within discrete active nesting territories and, in most instances, the same nest. The key characteristics (slope and elevation) of each relationship were tested for differences related to species and geographic region. Predicted variability of relationships were examined and compared to variability associated with natural systems. The results of statistical testing indicate that applying the conversion factors between species (GHO to bald eagle) and among geographic regions yields predicted egg concentrations that are not statistically dissimilar and are within the 39 natural variability observed for residue concentrations among eggs of raptors within species and region. Keywords: Raptors, PCBS, Plasma, Egg, Non-destructive sampling 40 INTRODUCTION Because raptors are at the top of the food chain, they are maximally exposed to many persistent and bioaccumulative residues [1]. This, combined with the fact that they are susceptible to the toxic effects of many contaminants of concern, means that raptors can be used as effective and sensitive biological monitors for contaminant exposures and assessment of environmental effects [2]. Raptors also are often used as enviromnental sentinels for monitoring of contaminants [3] or as primary or surrogate receptor species in ecological risk assessments [4]. Raptors are particularly useful, because they are often territorial and long lived, reproducing in the same territory over long periods of time. Thus, extensive data bases of historical contaminant exposures are often available. Historically, contaminant monitoring programs utilizing raptors have primarily used eggs because of the several advantages of using them for assessing contaminant exposure and effects. These include ease of collection and the fact that the proximal exposure of the developing embryo to the chemicals gives a direct measure of one of the most sensitive endpoints, embryo lethality [5]. Eggs are relatively easy to transport and store, and egg samples from wild bird populations are available independent of egg fertility. In addition, since lipophilic compounds tend to accumulate in lipids of eggs, quantification of residues is facilitated. Furthermore, controlled laboratory studies, including feeding and egg-injection studies, offer direct comparisons of concentrations of residues in eggs with effects such as survival, eggshell integrity and developmental deformities [6-9]. Egg-based contaminant exposure measurements have also been correlated with temporal and spatial effects. Nevertheless, egg sampling has some serious limitations when used in site-specific and long-terrn investigations of potential 41 ecological risk. These include the destructive nature of the sample, a high level of nest disturbance that significantly increases the frequency of nest abandonment, high levels of uncertainty for assigning spatial origin to the observed exposure concentration, and narrow temporal limits to the “window” of monitored exposure [10,11,(Frank, 1997, Master’s thesis, University of Wisconsin, Madison, WI, USA]. Egg sampling efforts also may be limited by the gender-restricted nature of the sample. When the disadvantages of determining egg-based exposure data outweigh the advantages, residues are measured in blood plasma [12-14]. This approach also has several advantages. These include the ability to collect blood without destroying the individual, the ability to collect samples from the same individual over time and the ability to collect samples from nestlings [14]. Because nestlings are sedentary and most . residues in their blood are accumulated from food; nestling blood plasma is an integrated measure of concentrations of residues in the area proximal to the nest site, much more so than are concentrations of residues in eggs or adult plasma samples [11,15]. For most raptor species, collection of blood plasma from nestlings reduces the risk of injury to the bird and minimizes abandonment or nest relocation by adult birds. Also, blood samples need not be gender- or age-specific. Use of blood plasma has been further advanced by development of more sensitive methods of residue analyses that has lessened the mass of analyte required for quantification. Limitations of plasma contaminant data are primarily in interpreting the effects of residues in blood plasma. There is relatively little information relating the concentrations of Specific residues in blood plasma of nestling raptors to adverse outcomes, while there is more information on the effects of concentrations of residues in eggs on effects on both 42 individuals and populations. Long term monitoring of residues in eggs, blood and raptor populations have demonstrated that trends in concentrations of residues are similar for eggs and blood plasma [15]. Thus, the use of blood plasma for monitoring populations for‘adverse effects would be facilitated by predicting concentrations of residues in blood to concentrations in eggs. We used synoptic sampling of blood plasma from nestling great horned owls (Bubo virginianus) and bald eagles (Haliaeetus leucocephalus) from the Great Lakes region to develop a relationship to predict concentrations of polychlorinated biphenyls (PCBS) and p, p ’-dichlorodiphenyldichloroethylene (p, p ’-DDE) in eggs. We compared these relationships to those previously published for bald eagles from other regional sub- populations and assessed the variability of predicting total PCB concentrations in eggs from those in blood plasma. MATERIALS AND METHODS Collection of Great Horned Owl Blood Plasma and Eggs Blood plasma from nestlings and fresh or addled eggs of great horned owls (GHO) were collected from the Kalamazoo River Superfund Site (KRSS). Collections were made between April 2000 and April 2004 along a 190 km stretch of the river’s floodplain between the cities of Marshall and Saugatuck, MI (Figure 2.1). Collections were made from both naturally occurring nests and artificial nesting platforms. This location represented a gradient of concentrations of both PCBS and ZDDT (dichlorodiphenltrichloroethane (DDT) and its metabolites 43 I LOWER KRSS / / UPPER KRSS / LLEGAN COUNTY . 0,8, 1..an Km I KALAMAZOO ' CALHOUN COUNTY COUNTY LAKE MICHIGAN Figure 2.1. Map of area of the Kalamazoo River, indicating the location in southern Michigan as well as the three reaches across which a gradient of polychlorinated biphenyl (PCB) concentrations was observed. 44 dichlorodiphenyldichloroethylene (DDE)/dichlorodiphenyldichloroethane (DDD)) that ranged from local “background concentrations” to relatively great concentrations of PCBS and EDDT [16]. Matched egg and nestling plasma samples were collected from nests of the same mated pair occupying the same nest in the same reproductive year. In other instances, matched egg and nestling plasma were collected from the same nest over a period of two or more years. In cases where nesting pairs selected a new nest site, samples were collected from alternate nests within the same territory over two or more consecutive years. To encourage re-nesting after collection of eggs, fi'esh eggs were collected as soon as possible following confirmation of incubation. Addled eggs were obtained when nests were abandoned. Eggs were labeled, transported back to the laboratory, and stored at 4 °C until processing. Length, width, and whole-egg weight and water volume were measured prior to removal of contents. Egg contents were stored in solvent-rinsed glass jars at -20 °C until measurement of PCBS and ZDDT. Nestling blood samples were collected when nestlings were approximately 4 to 6 wk of age and had attained a minimum body weight of 0.75 kg. A sample of 5 to 7 ml was withdrawn from the brachial vein with a heparinized disposable syringe (25-gauge hypodermic needle) and sterile technique. Blood was transferred to a heparinized VacutainerTM tubes and labeled. VacutainersTM containing whole blood were centrifirged at 1200 rpm for 10 min within 48 h of field sampling. Plasma (supernatant) was transferred to a new VacutainerTM apprOpriately labeled and stored upright at -20 °C until measurement of PCBS and ZDDT. 45 Whole egg homogenates and nestling plasma samples were processed and analyzed for congener-specific total PCBS and ZDDT using methods described previously [17]. All chemical concentrations in eggs were corrected for moisture loss [18]. Collection of Bald Eagle Blood Plasma and Eggs The values used to develop the egg to plasma relationships for bald eagles in the Great lakes region were compiled from studies conducted by several State and Federal agencies as well as public and private research institutes, the majority of which were completed between 1996 and 2002. With a few exceptions, most egg samples included in this data base originated from the United States Fish and Wildlife Service (U SFWS) environmental contaminants program using addled egg collection [19]. Most measurements of residues in blood plasma were from the Michigan wildlife contaminant trend monitoring program administered by the Michigan Department of Environmental Quality (MDEQ), Office of Surface Water Quality [20-23]. Additional data were also used [17,24,25, (Bowerman, 1991, Master’s thesis, Northern Michigan University, Marquette, MI, USA)]. All studies reported concentrations of total PCB, and/or p,p’-DDE for blood plasma of nestlings and/or egg samples collected from discrete active eagle nesting territories. From these reports, concentrations of individually paired blood plasma and egg samples were assembled according to the following three general guidelines: 1) Plasma samples collected from 1996 forward were paired with egg samples collected \Kiithin a 5-year window of sampling for the two media (eg., egg (‘97) paired with plasma 46 (‘01)); 2) Samples collected prior to 1996 were paired only for the same or two consecutive collection years; 3) For either grouping, a third or fourth sample was included in instances where two consecutive collections of plasma or egg were made (eg., egg(‘86/’87), plasma(‘87/’88)), in which case the geometric mean concentration of the two combined samples was used. The selection of a 5-year maximum window for pairing the most recent samples was based on the trend monitoring data for PCB and j, p ’-DDE concentrations in eggs that were not significantly different within sub- populations, between years fi'om 1996 onward [19]. Sample collection and processing for these studies are consistent with the methods described for Kalamazoo River GHOS, but methods of chemical analyses varied to some degree. USFWS analyses of p,p’-DDE and total PCBS in addled eggs were completed by the USFWS Patuxent Analytical Control Facility using gas-liquid chromatography. Nominal lower limits of detection were 10 ng/ g, w for DDE and 50 ng/g for total PCBS. Egg concentrations were corrected for moisture loss [19]. MDEQ analyses of p,p’-DDE and total PCBS (sum of 20 PCB congeners) in nestling plasma were completed at the Clemson Institute of Environmental Toxicology using capillary gas chromatography with electron capture device following U. S. Environmental Protection Agency approved methods. All reported results were confirmed by dual column analyses. Quantification levels for both compounds were 2 ng/ g [20-23]. For use in this assessment, the MDEQ PCB plasma concentrations for the 20 quantified PCB congeners were converted to a total PCB equivalent using the relationship: Total PCBS = i.57(sum 20 PCB congeners, ng/g w) + 0.98 [24]. Analyses of p,p’-DDE and total P CBS in nestling plasma for the Green Bay and Fox River samples [25] were completed 47 at Michigan State University and included the use of gas chromatography with electron capture detection and confirmation with mass spectrometry. Detection limits were 2.5 ng/g for DDE and 5 ng/g for total PCBS. Analytical methods for additional egg and plasma samples from Green Bay and Fox River [26] are provided by the authors. Statistical Analyses Sample sets were analyzed for normality by the Kolrnogrov-Smirnov, one-sample test with Lilliefors transformation. Concentration data were log-normally distributed and after log-transforrnation satisfied assumptions of normality. To evaluate the plasma to egg relationship for each PCB and p, p ’-DDE data base, a Pearson product-moment correlation analysis was performed on the log-transformed values. Paired blood plasma and egg concentrations of residues were plotted as a firnction of the blood plasma values and the line of best fit for each sample set was derived through simple regression and residuals analyses. Normality and correlation analyses were completed using the Statistica (Version 6.1) statistical package (Statsoft, Tulsa, OK, USA.). Regression residuals were calculated using EXCEL (Microsoft® Windows PE, 2002; Microsoft, Redmond, WA, USA). To assess the robustness of the relationships developed for the GHO, predictions were compared with measured values for bald eagles at other locations available in the literature. Tests for homogeneity of regression coefficients and elevation used analysis of covariance methods (ANCOVA) [27]. Multiple comparisons among elevations were made by use of Tukey’s HSD for unequal sample size [27]. The criterion for Significance 48 used in all tests was p<0.05. Comparisons of conversion factor predictive variability were made by calculating the relative percent difference (RPD). To test the similarity between egg to nestling plasma relationships between bald eagles of the Pacific Coast versus bald eagles of the Great Lakes, an AN COVA was used to test for equality of the population regression coefficients (slope) and elevation. Tests of elevation may be considered to be the same as asking whether the two population Y intercepts are different. However, one must be cautious of misleading interpretations from such a characterization if discussion of the Y intercepts would require a risky extrapolation of the regression lines far below the range of X for which data were obtained [27,28]. The use of elevations instead of Y intercepts also assures comparison of relationships over only the range of plasma/egg concentrations measured in each study and eliminates the potentially confounding effects that analytical detection limits may contribute to a test of Y intercepts. If either test identified a statistically Significant difference within the pool of data being evaluated, additional pair-wise comparisons were completed using Tukeys HSD to identify which of the population Slopes/elevations differed from one another. 49 RESULTS Great Horned Owls The fidelity of GHO to established home territories and preferred nesting Sites resulted in multiple instances of re-nesting that provided samples of both nestling blood plasma and eggs for the same breeding pair in the same nesting territory. A total of 14 paired GHO nestling blood plasma and egg samples were assembled from a total of 16 blood plasma and 17 egg samples. This included four nests where the samples of blood plasma and eggs were collected in the same nest during the same year and three nests where the paired samples were collected from the same nest but during different years. In seven instances, samples were collected from two different nests within an active nesting territory but during different years. Relationships between concentrations of both total PCBS and ZDDT were investigated. Total PCB concentrations for these 14 data points (Table 2.1) exhibited a gradient among the three reaches of the KRSS and there was a significant positive correlation (r=0.766, p=0.001) between log-normalized nestling blood plasma and egg PCB concentrations (Figure 2.2). The narrow range of ZDDT concentrations detected in KRSS GHO plasma and egg did not exhibit such a gradient and concentrations of ZDDT in'eggs were negatively correlated with those in blood plasma (r=-0.735, p=0.003). Thus, a EDDT egg to blood plasma predictive relationship was not developed for GHOS. 50 Table 2.1. Great Lakes (Kalamazoo River) great horned owl plasma to egg polychlorinated biphenyl (PCB) conversion factor sample pairing. Plasma Egg PCB Sample Location (ng PCB/ml) (pg PCB/g ww) (sample year: plasma/egg) Lower KRSS (02/00) 147 12.2 Lower KRSS (02/01) 147 19.8 Lower KRSS (02/02) 147 25.7 Lower KRSS (02/04) 147 2.09 Upper KRSS (02/03) 80.4 1.61 Upper KRSS (00/02) 60.2 2.74 Upper KRSS (01/03) 43.7 1.61 Upper KRSS (03/03) 31.3 0.61 Lower KRSS (03/04) 31.0 2.78 Reference (03/01) 25.9 0.31 Reference (03/02) 25.9 0.17 Reference (03/03) 25.9 0.22 Upper KRSS (01/02) 24.1 2.74 Reference (02/02) 14.0 0.21 51 Log (1J9 PCB/9 e99 ww) 0 Same nest/Same year 2 ' I Same nest/Different year 15 _ A Same territory/Different year 1 .. 0.5 _ 0 I I L. l O 0.5 1 2.5 -0.5 - x6 0" A -1 _ _1 5 _ Log (ng PCB/ml plasma) Figure 2.2. Concentration of total polychlorinated biphenyls (PCBS) in eggs of Great Horned Owls as a function of PCBS in blood plasma of nestling Great Horned Owls along the Kalamazoo River, Michigan. Regression line with 95% confidence limits of the predicted line (loglo (pg PCB egg/g, w) = l.647[loglo ng PCBpjasma/mlfl - 2.578) (p<0.001, r2=0.666). 52 Great Lakes Bald Eagles A total of‘30 (total PCBS) and 31 (p,p’—DDE) paired nestling plasma and egg samples were assembled from the available individual samples from the Great Lakes region (Table 2.2). Pairings originate from a single nesting territory and are combined following the pairing guidelines discussed previously. Great Lakes bald eagles exhibited Significant positive correlations between total PCB and p,p’-DDE concentrations in nestling blood plasma and egg samples (PCBS; r=0.789, p<0.001; p,p’-DDE: r=0.569, p=0.001). Log- normalized PCB and p, p ’-DDE concentrations were plotted along with the line of best fit and 95% confidence interval for the regression line (Figure 2.3, Figure 2.4). 53 3898-3 83 anagram .02 3 of 9% 8.3828483 83 8328mm 8698c ASA 3&8 o2 3; one 2: ”fleece 88m 5&8 889% as; B on on on a: 88893 as; B sesame ease beau am Ea ca 3: N2 Basses .822 band mam seems 83 Begum one as of m: essence 83 assaaaam 38.38 83 885 NE 3 9% at 888058 83 8886 seems as; “5 of ca 0.2 8m @3838 88:5 “5 89358 d sages: one one 0.: 3a 5 899358 .m 8852 ES 83058 a 8352 one a: 3m QR a $8238 .m 83%: 28m aaaze eooemea 3 cam Ea SN a. seen: 28850 $638 83 coasaom mom as man a m Seats 93 868m oaeeae .6 53m a 8858M 3.” S; a: E easesae ..6 55 .m 8882mm 8693 age soaosmsaxm o? 92 0.2 an Sesame .2 aasmsqnm geese Ema: asaeo a 88253 c; QR an. E wage as: 8380 a 8882mm 8898-03 .5 ap-sm-oz\.fi €02 can . a: c2 5 88828-3 .LO snagged .z 8.8893 cacao on: 2 8:8. on” ca 03 qoaaaae 820 8:2 2 oafimae x86 £2 S 93 em» 3 8w flange 0.25 8:2 S sesame sax Canaan 5.: em 98 as finesse sax 828a Awwohfimflm ”SDUEEBV >65 33 AwwoREmmE Heme» oafimmv @moo we canon me 838 me 2858 we 5:804 2.95m man? Rd wwm «Emma wwm «885 5:83 038% mom madam 295mm 88$ ceasing Ammfl- Q43 0:233oHoEBEanmGoSEoG- as. new Amonc 38nt uBmEoEobom $0 8 «Emma 2mg 23 EMS 320 ..m N 2an AVE Brno—2 use «.5330 Soc 305955028 30 and «SEE _. .SE OmQE Bob 338.5588 gamma . a QmD «HE .8»?ng .bmmuo>ED gwflog EoEoZ £85 9.89.32 .33 .cmanomv Bob mqoewbnooqoo gamma . P: .M omsmbm Eon mcoumbcoocoo wwo c228 «ESE .22 no OmQZ Soc mcoumbqoocoo «833 OmQE Soc 2233880 «Emma . .82 33am: assumes moa om anaemtmmoa nae ”amamqouflouofi wfimn 829:st mom :39 a 8 395280 2a mcocowcoo mom om c8 323.5880 @833 mom 094C204: "O .33 OmQE Sow mqouabeooaoo «SEE 0 Range «5.830 Bob mcoembcooeoo wwo new 952m n .3: @5ng see 823588 Em A tease we3 cam 3.0 3 8333 E32 0888203 2.0 3. as 0.2 3.380: 38 33 $5 sesame wagon 0385 82m $6 3 was 0.9 me 908 Esau em oaseooae 83 e52 eaaofi ”no 0.4 98 me .8935 sex can 8828-3 83 8: So on m2 3: ceases e 83 30582285 Easesae 23 seem 8.0 em 03 0.8 8.38% 0-08 23 Sam Baggage 23 Sam 8.0 an 92 0.2 3.38% 0-08 23 seam 3838 House agoann 89m 2 .o 0.2 S .3 0.8 48838 ago eS033 83H masons 33 33 8.9 3: as 9.8 seasons 83 sea 889% Ben sag 05:8 Rd 0.2 :3 3a 8.38058 22$ 83 custom masons 93 no: . So as 8.0 3% 38838 83 8: 8898-3 cam Eon beam 02 0.2 Sam o. 3 38898-3 cam 38 see 55 Log (pg PCB/gags WW) O o 3 o ..4 1.5 ’ 0 1 0.5 O O 0 . . : .' J. . . . o 0.5 1 / ,.-"15 2 2.5 3 3.5 O. o o -o.5 . Log (ng PCB/ml plasma) -1 Figure 2.3. Concentration of total polychlorinated biphenyls (PCBS) in eggs of bald eagles as a function of PCBS in blood plasma of nestling bald eagles in the Great Lakes region. Regression line with 95% confidence limits of the predicted line (loglo pg PCBcgg/g, w) = 0.905[log10 ng PCBplasma/mlfl — 1.193) (p<0.001, r2=0.623). 56 1.5 0.5 -0.5 Log (ug p.p’-DDE/9 699 WW) C) -1_5 _ Log (ng p,p’-DDEIml plasma) Figure 2.4. Concentration of total p,p’-dichlorodiphenyldichloroethylene (p,p’-DDE) in eggs of bald eagles as a function of p,p’-DDE in blood plasma of nestling bald eagles in the Great Lakes region. Regression line with 95% confidence limits of the predicted line (10 10 pg p,p’-DDEegg/g, w) = O.676[log10 ng p,p’-DDEp1asma/ml)] — 0.578) (p<0.001, =0.324). 57 Confirmation of Concept To test if the generality of relationships for the prediction of concentrations of residues in eggs from those in blood plasma of nestlings are consistent between owls and eagles, and among geographically distinct sub-populations, ANCOVA for slope and elevation were conducted. The relationship between residue concentrations in nestling blood plasma and egg has not been previously investigated for GHOS, but Elliott and Norstrom [13] and Elliot and Harris [29] have examined the distribution of PCBS and p,p’-DDE in bald eagle nestling plasma and egg samples from the Pacific Coast of Canada and the U.S.. We have presented a comparison of the key characteristics of each of the four PCB data bases (one GHO, three bald eagle) and three bald eagle p,p’-DDE data bases for which nestling plasma and egg contaminant relationships are now available (Table 2.3, Table 2.4). A more robust egg to blood plasma relationship is obtained by using individual samples from the same nest or a much larger number of summary mean concentrations [29]. 58 av :5an as . I madam 932 0508006 3 a 62 .80 2 an 0 mm a $3- Eng 8:46 @845 a: clam u A a. 4 o» o u. a _ mg m Sova A8 a: . n man go o 088 58 2 an 0 mm a memo- 3.3 €245 873v $3 «a w .m 2 .E o A o e. a n m: m . 80.ng 83 P: 8:- mono G 8 I S 8 com I m3 m8 oumm 05am §2€5 933 320V ”gnaw 2mm . a so ov $3 $3- $3 98:2: E: 43 89c" m E: N omhwm 33383 @823 389 :50 @0802 $80 $3 $QO 3 oaaom we a a a omfim owamm m wgm 038% 30:80on a“ QBBE 90 2m 2998 mwm 038mm «.835 m5 cobm Acosflzaoansmv 86on mQEmcozflmm mom 9 9:22.“. mod .mogomfiooEQmoEHofiwEm €020 AmmUB £3235 wopmquofiobom 138 8m. SUV £93m 868280 $0 3 «88$ .md 2an 59 .83 @5505 838 822on a £6? @0238 8:898 mo o>ufinom2n8 qufiwm 038% mwo 98 «Emma @363 538 080880 6 . .3888 wagon 8E3 2E8 d8 GEE» 30338 03898 mo 0333883“ mwfihmm 29:8 wwo was «892m magma 3:239: .0 A33 .mE: n b noumbdooqoo mom 38 mwu 20:3 @083)?on 98 053 £qu u xv Gouda—28:8 mUm 138 «883 magmas 50253 mquouflou wowwofi a oaflOmow m 033. E @063on $3? EoBBEk can 3on nommmouwom .n .Eofimmomms ma: 8. 39:2: mumbfia 98 8me 33 06 How 33.56 08 3:3, magma»: =< .353qu 338%»: some 5 @088on mm ”n 3 Oman 8% 08 entomoc mafia“: maougwm m0 no.“ 833/ .5 “won mo on: 0:... Sm 329, 8:3meme can mm ..m 6258 .2 23 6O .0080 86008 88808 8080808 0 8883 08850800 0.800808 mo 38080008808 38.808 03800 30 880 08003 85008 8008 8.808000 .8 . 808.808 88008 88808 08800 80 8888588088800 0800880 .8 38080008808 $8.808 08800 $0 880 08008 M88008 8088388 .0 A33 .33 n b 88008800800 mom 38 mwo 0883 8020803008 880 CE,» 4888 n 5 80808800800 mom A088 08003 w88.m08800._,fi00r 880888.808 wofinwofi 0 0880008 0 030B 8 8080008 00305 80080884» 880 0820 88000838 .0, o 808000000 08... 8 008808 0003080 880 000.08' 808 08 898 808808 080 00808 w880808 =< 88.800808 08008008 8000 8 80800088 8 8 00 0008 0808 08 0880008 EN m: 08880885 mo 80.8 0030.? .5 0.008 .8 08A 088 8.8 0080> 00808880 880 mm .0 Ae 88.98 as . I mhdm 5002 3.508000 mu m use 30 08 Gm 0 mm m 280.. 80.0 A213 A873 03 oImm 0 A 8 8 o: 0 0. a 8 82 m 8.on A3 . R: 80.0- 20.0 A; I 88 A2: I 3 08 A708 0808 802 080888 A080 2.0280 2000 208 , 80.on A8 a 82- 20.0 A88 I 8.8 Asm I e 8.: Em ouum 0808 832308 A303 0200 2000 208 A?» 0888 03 A8888 03 90 88008088 0805 m n n 0 80m 080m . m a 3080.80.83: 088$ wmm 0A880m 088E 08m808m 8.808 0_ 80m 8080380883 800% mgcmcozflmm mam 9 0E00_n_ monuqd 0088880208 8080 Amman- $3 08038080808380886082808.. 8488 8.8 SUV 8800.8 8280800 36 8 08003 .0. N 0308. 61 There were no significant differences between the slopes of regression lines for concentrations of PCB in eggs and nestling blood plasma (ANCOVA, p>0.05) for the three bald eagle groups. A graphical representation of the PCB and p,p’-DDE egg to nestling blood plasma relationships between the four PCB sample groups and three p, p ’- DDE sample groups is provided (Figure 2.5, Figure 2.6). There were some statistically significant differences among elevations for these three groups (ANCOVA, p<0.003). The elevation of the Great Lakes bald eagle group was significantly different from the two Pacific Coast groups (Tukey’s Test, p<0.03). These findings indicate that the three population samples are all estimates of the comrhon population regression coefficient and are approximately parallel, but with differing elevations and differing values for a 62 +Great Horned Owl - Great Lakes ' 1.674(log plasma)-2.578 4:}— Bald Eagle - Pacific Coast 0.734(Iog plasma)-O.409 —l— Bald Eagle - Pacific Coast/Great Lakes 0.824(log plasma)-O.583 2 ' + Bald Eagle - Great Lakes 0.905(log plasma)-1.193 1.5 " a 1 ' U) 8, I 2’ 0.5 m E a) 0 l 3 0 3 8’ ..J -0.5 ' -1 - -1.5 ‘ Log (ng PCB/ml plasma) Figure 2.5. Concentration of total polychlorinated biphenyls (PCBS) in eggs of GHO and bald eagles as a function of PCBS in blood plasma of nestlings, of the same species, respectively. Regression lines are plotted and predictive equations are given in the figure legend. 63 —B— Bald Eagle - Pacific Coast 0.676(log plasma)—O.220 —I— Bald Eagle - Pacific Coast/Great Lakes 2 - 0.680(log plasma)-O.318 + Bald Eagle - Great Lakes 1 5 0.676(log plasma)-O.578 g 1 . d.) 2’ UJ D C? .Q. 0.5 ' O. O) 3- C) 3 o . . 0 0.5 3 .05 . Log (ng p,p’-DDE/ml plasma) Figure 2.6. Concentration of p,p’-dichlorodiphenyldichloroethylene (p,p’-DDE) in eggs of bald eagles as a function of p,p’-DDE in blood plasma of nestlings, of the same species, respectively. Regression lines are plotted and predictive equations are given in the figure legend. 64 predicted Y (egg concentration). The SIOpe of the relationship for the GHO from the Kalamazoo River was not different from that of the bald eagles (ANCOVA, p>0.05), but the elevation was significantly different (Tukey’s Test, p<0.001). The elevation of the relationship for GHO was significantly different from that of the two Pacific Coast bald ' eagle groups (Tukey’s Test, p<0.03) but not significantly different (Tukey’s Test, p>0.05) than the Great Lakes bald eagle group. Taken together, the results of the ANCOVA and Tukey’s Tests of differences in elevation for the PCB egg to plasma relationship indicate that the observed differences among the four groups is unlikely to be related to differences between species. For the p, p ’-DDE data set, there were no statistically significant differences among either slopes or elevations of the relationship between concentrations in eggs and nestling blood plasma for the three bald eagle groups (AN COVA, p>0.05). This indicates that the three sample groups could have been drawn at random as sub-populations from the same population. Egg Predictions To examine the predictive variability of results obtained from the various relationships, measured nestling blood plasma concentrations from the Great Lakes bald eagle data base were used in each of the four PCB conversion factor equations and three DDE conversion factor equations to predict concentrations of PCBS and DDE in eggs. The predicted PCB/DDE egg concentration was then compared to the measured egg PCB/DDE concentration comprising the plasma/egg pair from the same Great Lakes bald eagle data base. This approach allowed predictions made from each relationship to be compared to measured PCB/DDE concentrations in eggs. Differences between the predicted and 65 measured egg PCB/DDE concentrations were assessed by calculating the RPD between the two values. [RPD = ((lmeasuredcgg-predictedeggl) + (measuredegg + predictedegg) / 2)) * 100]. The analysis was conducted with three randomly selected sub-samples fi'om the Great Lakes eagle data base. Using a random number generator, three sets of 10 paired plasma/egg samples were selected from the 30 PCB samples and 31 DDE samples comprising the Great Lakes bald eagle data base. A second and more restrictive evaluation included calculating the mean predicted versus measured egg RPD for a subset of the Great Lakes eagle plasma/egg samples with a restricted range of plasma concentrations that matches or falls within the range of plasma concentrations used to establish each specific egg to plasma relationship. Predictive variability was assessed in each of the three randomly selected subsets and the single restricted subset. Results of the RPD analysis are expressed as mean RPD. (Table 2.5, Table 2.6). For the randomly selected subsets and PCB relationships, predicted egg PCB concentrations from the Great Lakes eagle data base produced the lowest range of mean RPD (73 to 78%), which would be expected since this relationship was derived from the same plasma and egg samples used in the comparison. Using the GHO conversion factor, mean PCB RPD values ranged from 74 to 97%. Mean PCB RPD values for the two Pacific Coast eagle conversion factors had the widest ranges from 73 to 97% and 70 to 100%. Predictive variability for the restricted PCB subsets show that the GHO conversion factor produced the lowest mean RPD (55%) among the four PCB conversion factors. For p, p ’-DDE, the predicted egg concentrations from the Great Lakes eagle data base again produced the lowest range of mean RPDA(52—81%) for the randomly selected subsets and the restricted sample (77%), as expected. Mean DDE RPD ranges 66 for the two Pacific Coast eagle conversion. factors included 89 to 93% and 95 to 97%, and restricted mean RPDs were 99% and 105%. 67 .quQ 38 2mm». 2.3 8di $86 2: Bob 58% toaoo owouoEmB -mfimma 08. was STE mtonoo @3828 3889a 00% $0.88. maoumbqooaoo 50% 35:99. BEESEQEOQ 30 83308 323 @806on How 30838.8 935 «855% 68th 36 @0803 32m was 0360 2mm .m.N 053. r 68 89 a: as 2 s 3 a? .S a we an: .2 I we @m: .sm I we an: .2 l we 9E mama we owns wwm can: .8? 8 ea? .8: 4: can: 681...: an? $43 9558 we owauaaaa 3. mm 3 52 035mm wouoEmormSmmE an: d. i s 3. mm P 2 came .8” n 8: a. 1.1111 9&1 .2 I 68 P E 8 8 can: .3. u 8: ma an: .2 .. c . w t. 8 8 2: can: .8” I 8: a. s E: E: as R: E; wag me 85E mwm 828m 28 383 new omo v.83 sew 28m .m 4.0338 .3 28m 23 888 saga 2258 we 85m «835 _ a 033m 885 QmMDm<§ EOE AHXD MUZmMmEEQ HZmUMmm mE OmHUHQmam .83 8% 880m 05 388 E Eon. mo on: 05 mo 5338 036:8on 06:35 2: no mHkuHHoHom 8:898 2&0on .05 98 32538.» 0&8me (Ho 38me wuuabmsa .momofia 56388 By 32:2: 83 8% Emma 28. 8wa H85 .0 =83 wwo can «Emma 8:3 383239: 8% N 2an 8m .w .333 H888 8w .oaammbsm uoHofimoHImEmflm 06. 9. ofioomm moNHm oEEwm 3:8:qu mEH H 358%“. 8me 8% 88¢.“ donHoZHoo mQQIQd c5 mom £8on :80 95% 8 wow: mHHoHHHwHHHHooHHoo «Emma (Ho owdaH. 2: E65» warn 33 .83 2mg 33 $wa 330 06. HHH $809 88 3:3 36568 =H~ 32:05 295%. nHHm UoHoEmeanHmmE .o .83 98 2mg 23 383 H86 2: a $58 3% Hogan nausea gamma fine Hmnz Ea Eve omIz 88H 838% 3&3 wwoREmflm Sun 828 8 HonHoHHow 8983 88qu mam: @3352 oaasmbsm @8023 ~288qu .n $0 88308 ME? 30 @306on H new 321.. .H._. ..B _..-.H_.. Em .w @288 2 033. 70 .oommp Saw Ewan Ema. moxmq H35 2: 8on 85me 0:38 onoEmoHHNEmEQ 05 98 EYE nmtonoo @3023 .280ng 025 3.8%.. maoumbqoocoo Samsmg 022.3680EomExcofiwonoEoww -. use. $0 @8338 323 BquE .8.“ Haoammommm REVS $858wa H583 Ewan 2mm .©.N 033.. 71 as as . G: 25 @w: .8 I23 . a? .2 I23 3&1 .2 I 2.8 9:, mama wé ems: 3m as? .28 I a can; .2 I u :8? .2 I t 2858 m5 owes «535 E. mm 2: 038mm nowombmouéfiwflm a? .2 I we 8 a 8 can: .23 I 3 2. 231.248 mm a a ea? .22 I a 9. an; .2 I 2.8 5 a 3 can: .2 I a z. E: as R: @we 0332 wwm opium 23 823 :35 2mg 28 W83 .938 228.2 03mm 2mm .38 2:22 €35 23.2 2.35 a 295m 825mm ammamfiaz 293 g mozmammbo 9595.2 8::qu 89055.2 72m: ‘ mzoflgmozoo mo? “E @2852 .m> $5595 72 .023 222020 002228 0222 302022 22.2 2.202 20 0:22 0222 .20 5020000 3220200222 0222222222 0222. 20 2022228022 02250920 02.220092- 022m 202222 3222922220.» 02283 .20 2.00.220 $502222: $3022.22 2202222322200 202 2202020222 023 02020 2mg 2220. 30202 2.820 .0 .3020 $0 20220 «22222222 200228 3202220232022 202 m 0302. 00m .20 .3023 2002022 0.20 magmbsm 20802228200233 022.2 20.2 .2320 02922.8 2222022220222 v.32 222 200222888 3me .8020 202002 2202202200 mQQIRQ 202222 mom 2,2200% 22000 02222022 02 20822 02022022228228 2222200222 .20 omqfi 0222 22222223 @223 0920. 8.020 2mg 202.8. 802.3 20020 0222 22 2.222022 82220 2002222 22226222220222 22.0 82020222 020800-226 20802220020833 .0 .023 222020 Emma 2023 3020.2 2820 0222 222 2222022 8020 . 20228 22320322222 m02222222 2208 2muz 205,220.20 cmuz 822 m2.282.... 20222.22 2202.89.22 o T22 20 20023 02 202020220w 202222022 802022222 92223 2002222220202 022280.253 20020028 222220222232 .n 30 200222220222 H22». mwo 200202200222 n 2% m AENITKNMV _. -.2_ A 822 9222 .6 220.2802 0.2 2222.2 73 DISCUSSION Because nestlings can be captured and handled more easily than adults, it is less likely to cause harm to individuals or the population [13,14]. The use of blood plasma from nestlings eliminates the potentially confounding influence of adult exposures during migrations or on their wintering grounds or in instances where resident, non-migratory species or individuals shift or greatly expand winter foraging territories [30]. Because most of the residues in the blood plasma of nestlings is accumulated from the area proximate to the nest, these results can be more easily compared with the results of diet studies to identify significant contributing sources of environmental contaminants in the food web [31]. Also, because the embryo and nestling are the most sensitive life stages for population-level effects of chlorinated hydrocarbon contaminants, field monitoring of nestling plasma combined with laboratory feeding and egg injection studies often provides the best opportunities to integrate laboratory and field studies in efforts to derive and verify field-based toxicity reference values for plasma-based endpoints [32,33]. Predictive Accuracy of the Plasma to Egg Conversion Factor Focused studies to specifically examine the predictive accuracy of the plasma to egg relationship are not available in the literature. Nevertheless, the relationship has been used in several studies as a tool for further interpretation of adult and nestling plasma data. The earliest efforts to predict concentrations of a residue in eggs fi'om measurements in blood plasma was for ZDDT concentrations in adult falcons and accipiter hawks. In wild American kestrels (Falco sparverius) of the Pacific Northwest, 74 ZDDT residues in adult female plasma closely paralleled ZDDT residues in eggs laid by the same birds [34]. Concentrations of DDE in eggs (lipid-basis) corresponded well with concentrations of DDE in adult European sparrow hawks (Accipiter nisus) [35]. A significant decrease in total concentrations of DDE in the bodies of females due to translocation to eggs was also observed. In both laboratory and field studies, concentrations of DDE in blood plasma of American kestrels correlated with exposure in the diet. Concentrations of ZDDT in adult female blood plasma in populations of three accipiters (goshawk (Accipiter gentilz's), coopers hawk (Accz'piter cooperiz), sharp- shinned hawk (Accipiter striatus» were correlated with EDDT concentrations in eggs [12]. Those authors also described a large decline in female kestrel plasma EDDT concentrations due to egg laying/transfer to eggs, similar to that observed for sparrow hawks. Studies of ZDDT exposure were conducted during a period when researchers were investigating the egg-thinning effects of DDT and its metabolites DDD/DDE, and the maternal transfer of contaminates to eggs was of primary interest to researchers who were deciphering the mechanisms of action for this compound. Separate adult plasma (female) to egg conversion factors for ZDDT were developed for wild, nesting kestrels (F alconidae) and accipiter hawks (Accipitridae) [12]. The parameters describing the two relationships were not statistically different for the two species. Therefore, those authors suggested use of a pooled regression to predict egg concentrations from concentrations of DDT in adult blood plasma for these two families. Use of the pooled data set in the log- log relationship for the two species resulted in a relationship that could be used to make comparisons among species. While these relationships provide useful background 75 information, the DDT relationships were not compared to the relationships deveIOped in this paper because they did not meet the selection criteria in this study. The pooled relationship developed for kestrel and accipiter hawks in one region [12] has been used to predict concentrations of ZDDT and PCBS in eggs from concentrations in blood plasmafrom wintering adult eagles in Colorado and Missouri [36]. The pooled relationship was used to predict concentrations of EDDT in eggs from plasma and to assess the potential for DDE-caused eggshell thinning and potential impacts to reproductive success of individual eagles. The egg to adult plasma ZDDT relationship also was used to estimate the potential hazard to reproductive success of migrating peregrine falcons exposed to DDE on their wintering grounds [37]. The same relationship was used to predict concentrations of DDE in egg from concentrations of DDE in blood plasma of adult bald eagles so that the exposure of those eagles could be compared to addled eggs collected from the same sub-region [3 8]. In that study predicted concentrations of DDE in eggs corresponded very well with measured values (RPDs<10%). However, concentrations of DDE in blood plasma were predicted from concentrations of DDE in whole-blood that had been adjusted for loss of DDE during storage and dilution. The relationship developed for kestrels and accipiter hawks also was used in two separate studies to predict concentrations of ZDDT in eggs of several South African raptors, including the greater kestrel (Falco repicoloides) and the lanner falcon (Falco biarmicus), and pied ' kingfishers (Ceryle rudis). In both studies a screening-level hazard assessment was conducted by comparing the predicted concentration of ZDDT to a toxicity reference value based on concentrations of DDE in the eggs of sensitive raptor and piscivorous species [39-42]. More recently, the USFWS 76 and MDEQ [26] derived toxicity reference values based on concentrations of PCBS and DDE in blood plasma of nestling bald eagles using plasma to egg relationships [29] and field-based benchmarks (concentrations in eggs that were associated with reduced productivity) [43,44]. The major obstacle to assessing the predictive accuracy of the plasma to egg relationship is the absence of an extensive data base containing individually matched plasma and egg concentrations from the same nest. The data base assembled from the USFWS addled egg monitoring [19] and the MDEQ plasma sampling [20-23] that is presented here provided a basis for this evaluation. The predictive variability RPD results indicate that applying the conversion factor between species (GHO to bald eagle) and among geographic regions (Pacific Coast to Great Lakes) yielded predicted egg concentrations that are within the natural variability observed for residue concentrations among eggs of raptors within a species and region. Within-clutch RPD values for total organochlorine pesticide concentrations in non-migratory sparrow hawks in Great Britain have been observed to be as great as 32%, while mean concentration (clutch means) between-clutch RPD values from the same female on the same territory are as great as 63% [45]. Relative percent difference values as great as 171% for PCBS (3—year interval) and 80% for DDE (4-year interval) have been seen for sparrow hawks from a sub- population of females nesting within the same geographic sub-region [46]. Studies of PCB and DDE concentrations in eggs sampled from the same population of eagles nesting in the vicinity of Green Bay exhibited between-clutch variability that ranged from 38% (one-year interval) to 125% (four-year interval) RPD for PCBS, and from 31% (one- year interval) to 133% (four-year interval) RPD for p, p ’-DDE [43]. 77 UTILITY AND APPLICATION There is variability inherent in predicting concentrations of PCBs and p, p ’-DDE in eggs using the egg to plasma relationships for GHO and bald eagle presented in this assessment. The predicted values and widely-bounded 95% prediction intervals for a relevant range of nestling blood plasma concentrations illustrate the homogeneity of slopes and heterogeneity of elevations among the four PCB and three DDE relationships (Table 2.7, Table 2.8). The predicted PCB and DDE egg concentrations for the Pacific Coast bald eagles show a consistent difference from the Great Lakes eagle and GHO predicted values. Even at their greatest, the observed divergences represent only a three- fold difference in the range of predicted egg concentrations that would be significant for an ecological risk assessment (e.g., at or above the toxicity reference value threshold concentration). This indicates that for ecological risk assessment applications, the use of a plasma to egg conversion factor would introduce minimal levels of uncertainty to calculations of risk. 78 .80220222200200 «22222230225098 @2228: 200292 20202.2 .20 0m§2 222230202 22 22202.2 «2322 $3 4+2 wwo 22 2022022200200 50% 3220222222 208222220on2022 2002222222322 SN 03.0.2. 2. 79 80 9.23222 88 I228 s... a, I 23 22 on“ 8% I 2.8 8% I 8.8 232 I 8. 8 98 is $2 82 8.3 I 2.8 9.2.2. I 3.8 22 22 I 8.8 2.2 a.“ 2.2: 222 8.8 I 2.8 8.8 I «n8 24$ I 22.8 2.22 :. .27 2.2. can 82 9.: I 2.8 8.2 I 3.8 8.: I R8 32 I :8 83 3.2 and 8.2 om can I 2.8 an.“ I 2.8 22.21 8.8 28.2 I 8.8 2.2.2 22.“ «no 22 .o 222 23. 2.8.22 82%. XS 32.2. 822 2:3- 2.8.2 2220028222 2205 2220028222 02208 2200202222 02203 2220028222 30% 283 .0238 022282 238 0228.2 2325 22.80 2825 22.20 222222202 228 0222.22 222.2 _ 02222.2 2022.2 0282 202mm 230 .22 .0 8232222880 22222822 ammqgm 0222.2 2.2222 02.222020 2322225 0222.22 2<0n22>2022 9:23.722 no.2 2302.. 233 222202 28 22.202222222200280 mum 22202022202223 . 22.02.0222,. .. 02203200 02.2.2.2.» .. fimwwnmflwmm0.2220222222222022. 20. ..umfififi www.mmm.Mflfimmmflfiw 9.22222 ....._....2.02. 02.22.._Hm..2.222o\omm puma... 22022222222082.2222. .,.w.mo..20_m2w2m022m 4.2m» 5 . .0». 111 ..I . L. .233 .8202 28 mac 2028 9...: 282002 qu 2222222222 22 30222222502200 mom 28.02 20.2 22222220220202 w022w02 22 00.228022 3222? 2220028222.» 20222 0220? 2202205232 .0. .225. wmo 20020220022 20.2 29228222 220220220022 $3 20302 202222 2022223 .22 8.288 2 n02.2.2.2. 81 22282.8 22.2 2 222 822 I 8.8 222.2 82 . 22.2.2 I 22.8 at... 1...... . .. 23 02 . 0.2.2 I 8 2 . 282 I 2.08 022 022 2.2...“ 82 22.2 I 8.8 8.2 I 008 8.20 I 22.8 2.2 was 22 8 222.2. I 2.020 a? I .2022 8.8 I 23.8 2.2 20.0 02.2 2: $3. 228.22 83. 2.2.22 222.2- 08.22 2.2002822 0006 2220028222 02208 2220023222 02205 .93 2820 @2200 me 32.2 022.80 0022.2 :80 0023.2 flofigfioocoUuamflm 0222.2 2022.2 022.2 2023 0282 2.22.0 . 22222.2 2mm2222 22 0 55.95 mm 2222. ..2 . 2.2 2 22 22 2 m 22 02:. 2220.2 22 mom 2.2222222 .3936 82m 2222322223 .2952 005222222232 0222 22202.2 A2228 5022 2222022 222222222 2222.2 mwwo «2:262:23? .3 2250 2222.822 222on 222 22202282228228 22222: 222222. $228022 2520.2 22822222228228 Assiozadmb 222222223220 ExomeQnoudon€oSEowbo2 -w b m N Amwomv 2222220233 22022222222220.3022 22228 Aowqfiv 33 52.222 02.8886 ...22 m 03.3. 11_r 22 2 P- \\ \\ Geometric mean concentration (95% UCL) of PCBS, DDT in egg (pg/9 wet wt) Sample Site: ' A I Reference E] Upper KRSS Lower KRSS b 2L N=5 N=5 N=7 PCBs EDDT (P <0.05) (P <0.03) N=5 N=5 N=7 N=5 N=5 N=5 TEQWHO-Avlm (P <0.005) 200 4 180 ‘ 160 . 14o . 120 . 100 .80 .40 .20 Geometric mean concentration (95% UCL) of TEQWHMM in egg (pg/9. wet wt) Figure 3.2. Geometric mean (ww) total polychlorinated biphenyls (PCBS), total dichlorodiphenyltrichloroethane (ZDDT), and 2,3,7,8-tetrachlorodibenzo-p—dioxin equivalents (TEQWHQAV-ian) in great horned owl (B. virginianus) eggs collected from the Kalamazoo River superfund site (KRSS), error bars show the 95%UCL. 112 any of the study sites where samples were collected in at least three of the five years of sampling (Kruskal-Wallace, p>O.38). Predicted Egg Concentrations . A total of 14 paired GHO nestling plasma and egg samples were obtained in this study. Log-normalized GHO PCB data are plotted in Figure 3.3. The formula describing the egg-to-plasma relationship (conversion factor equation) for PCBS in GHOS is expressed on a log-basis as “log PCBegg (pg/g) = 1.647[log PCBNM. (ng/ml)] — 2.578” (R2=O.666, p<0.001). Predicted mean concentrations of PCBs in eggs (pg PCB/g egg, ww) and ranges are given along with corresponding values measured in eggs for each sample site (Table 3.5). Predicted and measured geometric mean concentrations of PCBS in eggs from the Reference location and UKRSS were not significantly different (Mann-Whitney U test, p=0.65 (Reference), p>O.9 (UKRSS)). The predicted geometric mean PCB concentration in eggs from the LKRSS was approximately one-third that measured in eggs. However, the difference between the predicted and measured concentrations was not statistically significant (Mann-Whitney U test, p=0.24). T E QWHOAw-an Concentrations Dioxin equivalent concentrations (TEQWHoAvian) were calculated solely from GHO egg samples since minimum achievable method detection limits for individual coplanar. PCB congeners in GHO plasma were limited by sample volume. Concentrations of TEQWHO- Avian in GHO eggs at the Reference and both KRSS sites were significantly correlated with concentrations of total PCBS (r =0.89, p<0.001). Concentrations of TEQwuamm 113 1.5 - 1 2 ”a" LIJ 0.5 — 2’ In 3.) O I I I 5:) 0 0.5 1 25 0:2 cs — O ..l -1 . ’,’ _1 5 _ Log1o (ng PCB/ml Plasma) Figure 3.3. Egg to plasma polychlorinated biphenyl (PCB) relationship for Kalamazoo River superfund site (KRSS) great horned owls (B. virginianus), including the 95% confidence limits on the line of best fit. 114 22222282238322 22222222 .I. 222222222 222 2:3 20202.2 22022292228 wmo 02 2222222222 2226 2222.822 2.222% 0222 mam: 2282222628 2.220222222228228 mwo 22222022222222 .2 22.2 222.22 22.22 22.22 22.22 22.2 22288802 022 222202 vmmfim I ovm www.m2 I o coca I omm 053 I 3% 00¢ - 22$ Nmm - mm 02 H 52222 “22222222222222V 222.22 222.2 222.2 22 2.2 222 2222 22222 82222522 222222922 $22.3 ommim 223.2 82.22 2.2m mwm 2222032 022022222222222 I 2222 222m :23 $222 22m mom 222222 22.220889 mmbnmm I we? mama: I own moiv I 0mm omwfi I wmm 2222 I m3 mom I om owned 22m . 2222, 22m 22m 22m 2222 220222222222 220202220222 222222222202 220202229222 2222222222022 2222202220222 ma ~2qu ma Mmmmb mUmemmmM .AmmMVO 022m 2222222222225 .2252 082282232 22222 222022 «22222222222322 .3 2226 22222022 220222 2220222202 222 220222 2222222202222 2222225222 6.922022 22022222228228 28,220 022962 0222222202220 022 228222 oEoEoow 222222222628 22222 8E3 2.220222222228228 30222 22222222222222 25222222039223.2302 wwo 228222220222 22222 202222222222 22202.22 wwo 220202229222 .m.m 22222.2. were greater downstream than upstream with the greatest concentrations calculated for eggs from LKRSS. Geometric mean concentrations of TEQWHOAvian were significantly different among all three sites (ANOVA w/Tukey’s, p<0.005) (Table 3.4, Figure 3.2). All four non-ortho-substituted PCBS (International Union of Pure and Applied Chemistry (IUPAC) congener numbers 77 ,81,126,169) and five of the eight mono-ortho-substituted PCBs (IUPAC numbers 105,118,156,157,167) were regularly detected in egg samples from the three study sites. Mono-ortho-substituted congeners 114, 123, and 189 were not detected. Together, the non-ortho-substituted PCB congeners contributed 73.4%, 67.4%, and 54.2% of total TEanomm at the Reference, UKRSS and LKRSS, respectively (Figure 3.4). At least one of the non-ortho-substituted congeners monitored in the study was not present at concentrations greater than the detection limit in 80% of the samples from the Reference site, 60% of the samples from the UKRSS, and 20% of the samples from the LKRSS. The rank order of the frequency of detection for both non-ortho- and mono-ortho-substituted PCBs in eggs was: Reference: detected in 100% of samples (105,118,167,169), 80% of samples (126,157), 60% of samples (156), 40% of samples (77,81); UKRSS: detected in 100% of samples (118,126,167,169), 80% of samples (77,105,156,157), 40% of samples (81); LKRSS: detected in 100% of samples (77,105,118,126,156,157,167,169), 80% of samples (81). Polychlorinated biphenyl congeners 81 and 126 have the greatest TEF WHO/Man values relative to other congeners. Congeners 81 and 126 were detected in 53% and 93%, respectively, of all egg samples. Together they comprised 63.1% of the total concentration of TEQwquvian in eggs from the Reference location, 61.8% at the UKRSS, and 42.1% at the LKRSS. 116 1 00% 80% D Other ES] PCB-156 60% I PCB-118 PCB-105 40% PCB-126 I PCB-81 Percent of total TEQ WHOM“, 20% 1111 PCB-77 0% Reference Upper KRSS Lower KRSS Sample Site Figure 3.4. Percent contribution of polychlorinated biphenyl (PCB) coplanar and mono- ortho-substituted congeners to total 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQWHO-Avian) in great horned owl (B. virginianus) egg samples at the Kalamazoo River superfimd site (KRSS). Relative Potency The relative contributions of non-ortho- and mono-ortho-substituted congeners can be evaluated by standardizing the TEQWH0-Avian to the total PCB concentration to obtain a relative potency value [34]. Relative potency values can be used to assess the degree of weathering and in evaluations of exposure and bioaccumulation between trophic levels of an impacted food web and resulting changes in toxic potency of the weathered mixture. Geometric mean relative potency values for TEQWHGAvian and total PCBS in GHO eggs are similar among the KRSS and Reference sites. The greatest geometric mean 117 concentration, (1.3 x101 ug/g, ww) was observed for eggs collected at LKRSS (Table 3.4). ZDDT concentrations/Eggshell measurements Total DDT was detected in all egg and plasma samples analyzed. p,p’- dichlorodiphenyldichloroethylene occurred at the greatest concentration of the measured DDT analytes and contributed roughly 98% and 95% to the EDDT in all egg and plasma samples, respectively. Total DDT concentrations in eggs were greater from the KRSS than were those from the Reference location, and concentrations in the UKRSS and LKRSS were approximately equal. Geometric mean concentrations of ZDDT were significantly different between the Reference site and both UKRSS and LKRSS (Kruskel-Wallace, p<0.03). Geometric Mean XDDT concentrations in eggs were not significantly different between UKRSS and LKRSS (Kruskel-Wallace, p=0.95) (Table 3.6, Figure 3.5). The spatial distribution of ZDDT concentrations in blood plasma was similar to that of concentrations in eggs, but there was no statistically significant difference among the three sites (ANOVA w/Tukeys p=0.22). Egg shell thickness and Ratcliffe index measurements displayed similar trends among the Reference site and KRSS (Table 6, Figure 5). The mean Radcliff index at LKRSS was slightly less than values observed at the UKRSS and Reference site. However this difference was not statistically significant and neither egg shell thickness, nor the Ratcliffe index were significantly different among the three sampling locations. (ANOVA w/Tukey’s, p=0.27, p=0.44, respectively). Additionally, eggshell thickness and Ratclffe index were not 118 222222222 2222022382022 2222.2 magma V2820 33m £2228 2.22222222032 2222220222 22222 22222222 mmMM 223012 .22 22220822232282 2222on 822.2 22222 Sum .2220 022.0220 6322223922. 0222 0222220222 20222 2222222222 mmMM 202225 .0 2220222302 22222220 .222 222222 822220 2222 2222220222 22222 222222.22 822222.232 .22 .QQQ 2292 man. 20222022222222 2222.2 HOD 222222222 22822222228228 PDQ 22220.2. .22 222.2 I 22.22 22.22.22 I 2.2.2.222 22222.2 I 2222 2222 I 222 2 22.2 2 _ 222.2 2 222.2 2 2.2 222252 2225.2 22 I 22. 22 222.22 I 222.222 2222.2. I 2222 2222 I 2222 2 22.2 2 222.22 2 222.2 2 222 22252 22222 222.2 I 22.22 2222.22 I 22.2.22 2222. I 22.22 2222 I 2.22 2 2.2 2 222.22 2 2.2 2 2 22. 28222222222 72 2 z . 88 z 2222. z 252222 22.222 222222.22 282222 0222202222 2222222202222. 220222wwm 2.2.QQN mwm 22.—haw 22222222222 22222222 .2222 22222222222 29.52 0022222222232 2222 22202.2 92220 HQDNV 22222222222 222222 wwwo «24222222222322 .3 2250 222222022 222on 202 222.222 22222222 0.2222832 222222 2228202222 nonmmmo A22202222222222228 Cbmwv 02222222820302.23022222228030222 222202 .Aowefiv 23 22>» 2222222 "2222222280 6.2.“ 2322.2. 119 Pre-1 947 mean eggsheu thickness [33] 3600‘ 1800' .3, 1600‘ .E E 1400- D N o 1200- C A s“ s .3 g 1000- § 23 3 g 800 - C (U E 600 - .2 g 400 - O 0 0 200 - 0 , Reference I ZDDT E] Eggshell Thickness * 0.44 ‘ 0.42 E .C 8. '- 0.4 a)? _ gs- 0 0.38 E g a g ‘6 s - 0.36 § 0 (D ‘ 0.34 ‘ 0.32 ' 0.3 Upper KRSS Lower KRSS Samulefiite Figure 3.5. Geometric mean (ww) total dichlorodiphenyltrichloroethane (ZDDT) concentrations in great horned owl (B. virginianus) eggs and eggshell thickness at the Kalamazoo River superfund site (KRSS), error bars show the 95%UCL. 120 significantly correlated with ZDDT concentrations in eggs (r=0.35, p=0.17 and r=0.04, p=0.8 respectively). Relative Abundance/Site Use Rates of hoot call responses for individual, pairs and juvenile birds did not vary by time of survey (am vs. pm) or season (Table 3.7). Significant differences in the distribution and frequency of responses of individual (Xz=16.79, dfiZ, p=0.001; X2=l6.6, df=l, p=0.001) and the frequency of responses of paired owls (X2=7.0, df—‘l, p=0.01) were observed between FC and TB, with TB having a greater relative abundance of both resident classes. Juvenile response frequencies were also significantly greater (X2=7.57, df=1, p=0.01) at TB. Productivity Over the period of 2000 to 2002, there was no discemable difference in productivity (fledglings/active nest) between the nest sites in the upstream (FC) and downstream (TB) study areas where fledgling success was monitored (active nests n=1,F C and n=6,TB; no statistical testing performed due to the small sample size at FC). For the three-year study period, the arithmetic mean rate of productivity was 1.0 successful fledglings per active nest at both locations (Table 3.7). There were more active nests (6 vs. 1) and fledglings at TB (6 vs. 1) than FC. At TB there were two nests that each produced two fledglings, and also two failed nesting attempts (both nests abandoned during incubation). The observed number of active nests at TB compared to PC was very similar to the results of 121 22222 I 22222 22822 2222 Noom 222cm ooom 2 2 2.2 2 2 2 2 2 23222222222222 2 2 2 2 2. 2 2 2 222222222222 2" T2 2" . 2n 2nz 2" T2 2u2z 22222202222922 . 2222222222 22222 @222 22222 . 2222 222222 22 222 2222 2222 2 2.222 .2 222 .22 22x22 .22 2.22 .2 22222222222222 2222222232 2822223222222 2222822232 2822222222822 0222022232 2822222222822 2222522232 E E . Q S m02fl®>flh . . . . . . . 222222222 22 2:2 232 22222 220 Re 2222 2 w . . . . . . . . . 2222222222 2222 mwo $2 bmo 222 So M22 22 .2 . 22.2 22.2 2 22.2 22.2 22.2 22.2 2.2 2222.2 8322222022232 222232 22232 2222022232 22822 2232 22222222232 2222222 2232 2228222222 22222222 22 22 22 22 222522.222 NNHZ VNHZ ”NZ 2 T2 2H2 mu 2H2 qu 92222222222222 0322232 . ow2222223022. 282220 .222 $222530; 282220 .E ow2222223o22. 232220 .222 ow22252622. 2222220 .222 .2222 22 2222 882 2222222 2222222 2222223222 22222 22222222222222 232220 .222 222 22222222222222.2222 .$ 22322 222222.822 2.20% 222222222222 20 22252202222822 222222 82222222222222 222.2232 .2. M 222222: 122 .2222: 22.2222 mo 2222222522HZ .22 @2222 25222 2222 .20 22228 2222222222222 2222 226 25.22.2232 2228 .22on wfiflfimm 2222220 22222223 2.22222 .2022 222.222 2:222 2222\mm2222w220u 2222 22.222 250.2 :2... 802.2 mwfi2wwom 222222822222 .20 202282222 28.02. 22.2.. 2 .3538 M25285. 232208222 2222222 2222.225 .20 22.592228 < .2 2.822 22.22 .226 2202222226 2228 2222222 22222292 22 no 2222222222228 8222022202 22.2.2 wfimwon wfi2mwuw 2220 2.222222 2.22 2229» 22222222222 22222022288 mo 0228.203 SLEEVE 2230 $222322 .20 mucoswofi 0222022202 ow222o>< .22 .2230 222.6 $22822 + 2222282 22822222 222022 2822022202 20222220222 .w 222.25 2222222223922 2202222222222 8022 222228222222 222222220222 ..2 . .2226 82.2802 + 22282 22222222 222222 222222322222 222022 88.2 2222202228 2222202222 202222222222 .22 2222222222 22222 mo 228222 22222 22 92222 228.2 22222 .2292 22822 20.2 2.23.2222 228222222228 Z 220222 222w222o>2 22 8.22 22282228 282232 .22 2.222.252 022222288 .20 202282222HZ .0 2222222 222222 2232222 22 22222222222228 2.29522 2222822222238 2822 2220.22 222522222 2822222222 8222222252222 232222232 .22 dam 2222222222223 .2252 82.282232 :2 222.2802 2.2 2222.2. 123 the call/and response surveys. Measures of relative abundance and site use obtained with the call/response surveys also indicated that there was a greater number of actively defended territories (or resident owls) in the floodplain at TB. Risk Assessment Measured 95% UCL (geometric mean) concentrations of PCBS in eggs collected from the UKRSS did not exceed the egg-based NOAEC TRV. The maximum HQNOAEC in the UKRSS was <1.0 (HQ=0.5). PCB concentrations in eggs from LKRSS included the four greatest individual PCB concentrations out of 12 eggs for the entire KRSS. In the. LKRSS, the 95% UCL (geometric mean) concentration of PCBS in eggs, resulted in HQS of 1.0 and 3.1 when compared to the LOAEC and NOAEC, respectively (Figure 3.6). Hazard quotient values for TEQWHO-Avian, based on both the LOAEC and NOAEC, were <1.0 for all individual egg samples from the Reference and UKRSS locations. The greatest HQNOAEC was 0.2 for the UKRSS 95% UCL (geometric mean). Dioxin equivalent concentrations (TEQWHO.Avian) in the LKRSS, which included the five greatest concentrations out of 10 eggs in the KRSS, resulted in HQS of 0.5 and 1.3, respectively, when the 95% UCL (geometric mean) concentration was compared with the LOAEC and NOAEC (Figure 3.6). Hazard quotient values for the 95% UCL (geometric mean) ZDDT concentrations, based on both the NOAEC and LOAEC were 51.0 for all three sites with a maximum HQNQAEC value of 1.0 at the UKRSS (Figure 3.6). Geometric mean eggshell thickness values at the Reference and UKRSS were not significantly different and 51% below the 124 Figure 3.6. Hazard quotients (HQ) for the effects of total polychlorinated biphenyls (PCBS), 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQWHO-Avian), and total dichlorodiphenyltrichloroethane (EDDT) for great horned owl (B. virginianus) eggs at the Kalamazoo River superfund site (KRSS) based on the no observable adverse effect concentration (NOAEC) and the lowest observable adverse effect concentration (LOAEC). Each box encompasses the 95%CI about the geometric mean concentration. 125 Hazard Quotient (95% LCL/UCL) Hazard Quotient (95% LCL/UCL) Hazard Quotient (95% LCL/UCL) 1.0 0.9 ' 0.8 ' 0.7 ' 0.6 ' 0.5 ' 0.4 ' 0.3 ' 0.2 ' 0.1 ' 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 —————————————————————————————————————————————— Reference HQS E Total PCBS - TEQWHO—Avian III 2: DDT 3.- 2-0 LOAEC Lower KRSS HQS 126 pre-1947 benchmark for GHO, but mean thickness at the LKRSS was 4% less than the pre-1947 value. Values for the Ratcliffe Index were 6% less at the Reference location than the pre-1947 values while values were 4% and 7% less at UKRSS and LKRSS, respectively. Hazard quotient values for predicted 95% UCL (geometric mean) concentrations of total PCBS in eggs at all three sites are less than values based on measured concentrations in eggs (Table 3.5). Use of the predicted concentrations of PCBS in eggs at LKRSS resulted in a geometric mean HQNOAEC of 0.37 compared to a value of HQNOAEC of 1.3 based on the comparable measured geometric mean concentration in eggs. DISCUSSION A behavioral attribute that favors use of GHOS in ecological studies is their preference to use nests built by other bird species. To our knowledge, this study is the first to successfully incorporate this behavior into a study designed to induce GHOS to occupy areas of maximum exposure potential, and provide for conservative and worst-case exposure assessment evaluations of the terrestrial food web in a site-specific baseline ecological risk assessment. Previously, Strigiforms have been used to determine ambient or baseline environmental conditions of avian exposure to DDE and other chlorinated hydrocarbon COCS [35]. A second class of investigations focused on local and acute poisoning episodes stemming from use of the acetylcholinesterase inhibiting organophosphate and carbamate pesticides [3 6]. 127 Comparison of total PCB concentrations to other locations Few studies have measured concentrations of PCBS in wild GHO eggs (Rosenberg BG, 1990, Master’s thesis, University of Manitoba, Winnipeg, MB, Canada) [37], and in. some instances the analytical results (e. g., PCB quantification on an Aroclor-basis) are not directly comparable to PCB concentrations generated from congener-specific analyses. Surveys of healthy GHO populations in Ohio [37] and Saskatchewan (Rosenberg BG, 1990, Master’s thesis, University of Manitoba, Winnipeg, MB, Canada) found arithmetic mean PCB concentrations of 3.1 x103 ng/g, ww and 3.3 x103 ng/g, w, in eggs, respectively. These concentrations are greater than the arithmetic mean PCB concentration observed in GHO eggs from the Upper KRSS (2.0 x103 ng/g egg, ww).' Although these studies were limited in scope, the results support the conclusions of the risk assessment which suggest that GHOS in the UKRSS are unlikely to be affected by exposure to PCB. PCB congener profiles The relative concentrations of PCB congeners used to calculate TEQWHOAWM. in KRSS GHO eggs were similar to those observed in eggs of barn owls [38] and eagles [39,40] in North America and Europe. Similarities among the patterns include the predominance of PCB126 as the maximum detected concentration expressed on a wet weight basis (ratio of PCBs 126277 ~2:1 or greater) among coplanar congeners, and as the greatest relative contributor to total TEanaAvian. This is consistent with observations that PCB 77 and 81 are more susceptible to metabolism than PCB 126 and 169 [41]. Among mono-ortho- 128 substituted congeners, PCB 118 occurred at the greatest concentrations and PCBS 105 and 15 6 contributed the greatest relative proportion to TEanoAvian, ZDDT/Eggshell measurements The ZDDT concentrations in GHO eggs from all regions of the KRSS were within the range of ZDDT concentrations reported for investigations of healthy GHO populations associated with non-point source exposures to ZDDT (Rosenberg, BG, 1990, Master’s thesis, University of Manitoba, Winnipeg, MB, Canada) [37,42]. The fact that the relatively small concentrations of ZDDT measured in GHO eggs from the KRSS were similar to those measured in eggs from other healthy GHO populations indicates that ZDDT concentrations in KRSS GHO eggs were not having an adverse effect on GHO in the KRSS. The spatial distributions of EDDT concentrations observed in this study indicate that there was relatively little historical use of this pesticide in the Reference area. Greater concentrations of ZDDT in the UKRSS and LKRSS may be related to historical agricultural use since both of these sites receive significant inflow from tributaries with drainage basins that contain agricultural development, including fruit production. Additionally for the LKRSS, the greater concentrations of EDDT bioavailability may be associated with exposures in Great Lakes influenced habitats where fish are known to have greater concentrations of ZDDT [43]. Toxicity Reference Values (TR Vs) Few studies meet all of our criteria for deriving a TRV for PCBs in GHOS. While birds, especially raptors, are generally considered to be some of the most exposed and sensitive 129 animals to the effects of chlorinated hydrocarbons, there is a wide range of sensitivities to PCB and other aryl hydrocarbon receptor-active chemicals among species [44,45]. Application of laboratory or field-derived TRVS among dissimilar avian orders, such as Galliformes and Strigiformes introduces uncertainty and associated conservative bias that can result in protective but unrealistic TRV values. Additionally, values for thresholds of effect, based on the results of acute studies are of little use when trying to establish TRVS for chronic effects in wildlife. Co-contaminants in test diets or from field studies can substantially confound the toxicity results relative to a single chemical or class of chemicals. In particular, assignment of causality can be problematic when elevated levels of co-contaminants are present. Similarly, complex mixtures such as PCBS, which are subject to environmental weathering, are dynamically changing in relative congener composition and toxic petency depending on the environment to which they are exposed. Quantifying the toxicity of neat mixtures or even weathered mixtures from different systems may not reflect the actual toxicity of the mixture of concern. To address any one of these uncertainties, risk assessors frequently apply an uncertainty factor to the published toxicological benchmark. Aside from applying an uncertainty factor of 3 to derive the PCB LOAEC from a validly determined NOAEC, application of additional uncertainty or extrapolation factors to our selected TRVS was not necessary. This is because the selected studies meet the key requirements as described above. Most specifically, the test species used were closely related wildlife species (a specific preference stated in the Great Lakes Water Quality Criteria documents) [25]. The selected studies also employed chronic exposures over sensitive life stages and measured ecologically relevant endpoints with minimal impact from co-contaminants. 130 Risk Assessment Studies of PCB accumulation patterns in the terrestrial food web at the KRSS [46] have found considerable variation in site-specific patterns of bioavailability, bioaccumulation, and biomagnification. Site-specific exposure potentials are altered by habitat and hydrologic conditions. Site characterization studies have shown the mean PCB concentration in floodplain soils to be greatest in UKRSS and more than twice the concentration observed in floodplain soils of the LKRSS (Table 3.1). Our measures of PCBs in GHO eggs and GHO nestling blood plasma did not parallel these spatial patterns of PCB exposure potential. Inconsistencies of this type underscore the importance of site-specific studies at sites as large and diverse as the Kalamazoo River. The greater exposure of GHOS to PCBs at the LKRSS may be due to exposure through trophic pathways that include both terrestrial and aquatic pathways including fish from Lake Michigan. For example, diet studies of GHO in this stretch of the KRSS may show that a large portion of the diet is comprised of Anseriform (waterfowl) or Charadriforrn (gulls) prey. In this study, use of either total concentrations of PCBs or TEanamim as measures of exposure in GHO eggs resulted in similar estimates of risk. A review of the 95% UCL (geometric mean) concentration ranges of HQ for total PCBS and total TEanGmm indicates a high degree of concordance between these two measures of exposure. Ranges of HQNOAEC and HQLOAEC based on either total PCBS or TEQWHO—Avian almost completely overlap and HQLOAEC were consistently 51.0 at each of the three study sites (Figure 3.6). 131 Concentrations of ZDDT in eggs of GHO at the KRSS were correlated with neither eggshell thickness (mm) nor Ratcliffe index. Changes in values relative to the pre—1947 values for mean eggshell thickness (maximum -4%) and Ratcliffe index (maximum -7%) in the KRSS were not close to threshold values of 15% to 20% associated with adverse effects on successfiil raptor reproduction and population maintenance [1 7,3 3]. The results of the hazard assessment suggest that GHO populations residing in the Reference and the UKRSS are not at risk for effects induced by PCBS, TEQWHO.Avian or ZDDT in floodplain soil. This conclusion is consistent with measurements of fledgling productivity. At the LKRSS, the HQNQAEC values for eggs are slightly greater than 1.0, with values as great as 3.1 and 1.3 for 95% UCL (geometric mean) concentrations of total PCBS and TEQWHO-Avian, respectively. These HQNOAEC values indicate that exposure of GHO to PCBS in this reach of the Kalamazoo River were near the threshold for effects. It is important to note that the true effect level for individuals lies somewhere between the NOAEC and LOAEC and, even conservatively, population effects have not been expected at a HQ of 10.0. HQNOAEC values near 1.0 are not likely to be associated with adverse reproductive effects in individual resident GHOS in the LKRSS. Multiple Lines of Evidence / Assessment of Population-level Effects This assessment employed a multiple lines of evidence approach to minimize uncertainty in assessment endpoints and to provide the best available information for remedial decision-making for later stages of site clean-up efforts. Included in the tissue-based “top-down” HQ approach [13], on which we report here, the potential effects of PCBs 132 and ZDDT on GHO productivity and relative abundance/site use were monitored in the F C and TB areas of the Reference and UKRSS sites. Mean productivity rates were similar among locations where exposures to PCBs were much different, with 1.0 successful fledglings/active nest at the two locations where reproductive success was monitored. The mean rate of 1.0 fledgling per active nest observed at both locations is consistent with productivity measures for healthy mid- westem [47] GHO populations residing in varied upland habitats. Measures of site-use indicate TB populations, as reflected by territory-holding nesting pairs, were near the carrying capacity (roughly one pair per 1600 ha and a total of three pairs in the TB floodplain) [12]. Furthermore, nest acceptance rates and nest fidelity of actively breeding GHOS across all nesting seasons included in the study were consistent with previous studies of artificial nest acceptance and habitat usage by Strigiforms in Midwestern forests [11,47,48]. For unknown reasons, the observed number of active nests at PC (1.0 active pair in any study year) was less than the carrying capacity of this study location. Adult mortality may have been a prime factor in the low density of active nesting pairs at F C, as we are aware of four confirmed adult deaths (three owl-car collisions, one owl- train collision) during the study period from 2000-2004. Unfortunately, we were not able to identify the sex of these dead owls. Owl population health data support the conclusion that TB populations are not suffering adverse effects to population maintenance. The “top-down” assessment of potential hazards to resident GHO populations completed in this investigation has employed intensive sampling effort of worst-case exposure, state-of-the art analytical techniques, multiple lines to estimate exposure to PCBS, an assessment of potentially confounding chemical stressors, valid statistical 133 methods, and conservative benchmarks of toxicity. The results of these studies suggest that current concentrations of PCBS, expressed either as total PCB concentrations or as TEanGmm are not sufficient to pose a significant risk to GHO populations in the UKRSS where large areas of former sediments are now exposed as floodplain soils. The data also indicate that risk in the LKRSS, while greater, is unlikely to be sufficient to cause adverse population-level effects. Results from this study concur with earlier investigations at the KRSS that indicated GHOS would effectively integrate exposures from primary environmental media through multiple trophic levels [46]. This study confirms that GHOS are a useful sentinel species for site-specific baseline ecological risk assessments employing a multiple lines of evidence approach that includes using “top-down” methodology to combine measured residues of COCs in tissues and counts of population and productivity. The study’s successful use of GHO provides a workable model that can be applied to other large sites with extensive areas of contaminated soils that require ecological investigations of risk or long term monitoring for potentially impacted terrestrial communities. In instances where elevated levels of contaminants cause concern for potential environmental effects, measures of owl chemical exposure, productivity, and abundance can serve as an index of overall ecosystem health. Acknowledgement The authors extend special thanks to three indispensable project volunteers, Donna Pieracini, Simon Hollamby and Heather Simmons, who provided technical support and logistic assistance. We also thank the students and technicians at the MSU Aquatic 134 Toxicology Laboratory, including Michael Kramer, Ryan Holem, Scott Kissman, and Theresa Gratton. We would like to recognize aid provided by USFWS personnel Lisa Williams and Dave Best, USFWS-contractor Charles Mehne, and staff of the MSU Kellogg Biological Station. We also thank Karen Smyth and Katie Coady. Funding for this study was provided by the Kalamazoo River Study Group. This study was conducted under USFWS permit #MB775945-O. This manuscript is dedicated to the memory of Mr. William Claflin who’s passion for birds and love of the field contributed greatly to the success of this study. 135 10. 11. REFERENCES BBL. 1993. Allied Paper, Inc. /Portage Creek/Kalamazoo River Superfund Site description of the current situation. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. US. Environmental Protection Agency. 1976. PCBs in the United States industrial use and environmental distribution: Task 1. EPA 560/6—76/005. Washington, DC. Bursian SJ, Sharma C, Aulerich RJ, Yamini B, Mitchell RR, Orazio CE, Moore DRJ, Svirsky S, Tillitt DE. 2006. Dietary exposure of mink (Mustela vison) to fish from the Housatonic River, Berkshire County, Massachusetts, USA: Effects on reproduction, kit growth, and survival. Environ T oxicol Chem 25:1533-1540. Grant DL, Moodie CA, Phillips WEJ. 1974. Toxicodynamics of Aroclor 1251 in the male rat. Environ Physiol Biochem 4:214—225. Giesy JP, Ludwig JP, Tillitt DE. 1994. Deformities in birds of the Great Lakes region: Assigning causality. Environ Sci Technol 28:128A-135A. McLane MAR, Hall LC. 1972. DDE thins screech owl eggshells. Bull Environ Contam Toxicol 8:65-68. BBL. 1994. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site Remedial Investigation/Feasibility Study. Technical memorandum 2: Results of phase I TBSA soil sampling. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. BBL. 1994. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site Remedial Investigation/Feasibility Study. Technical memorandum 3: Results of the floodplain soils investigation, volume 1 of 2. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. BBL. 1994. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site Remedial Investigation/Feasibility Study. Draft technical memorandum 12: Former impoundment sediment geochronological plot study, volume 1-4. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. International Joint Commission . 1991. Report of the ecological committee’s biological effects subcommittee to the Great Lakes science advisory board. Proceedings of the “Expert Consultation Meeting on Bald Eagles.” International Joint Commission, Windsor, Ont., Canada. February 12-13, 1990. p33. Council on Environmental Quality. 1972. Development of a continuing program, to provide indicators and indices of wildlife and the natural environment. Council 136 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. on Environmental Quality, Ecology Program, Smithsonian Institution, Washington DC. Final Report, April 15, 1972. Houston CS, Smith DG, Rohner C. 1998. Great Horned Owl (Bubo virginianus). In Poole A and Gill F, eds, The Birds of North America, No. 372. The Birds of North America, Inc., Philadelphia, PA. Fairbrother A. 2003. Lines of evidence in wildlife risk assessments. Hum Ecol Risk Assess 9: 1475-1491 . Brenner FJ, Karwoski D. 1985. Vocalizations as an index of abundance for the great horned owl (Bubo virginianus). Proc Pa Acad Sci 59: 129-130. Strause KD, Zwiemik MJ, Kay DP, Park CS, Blankenship AL, Giesy JP. 2007. Utilizing the great horned owl in site-specific assessments of environmental risk and natural resource damage. Hum Ecol Risk Assess (In Press). Houston CS, Crawford RD, Oliphant LW. 1993. Unhatched eggs in raptor nests in Saskatchewan. In Redig PT, Cooper LW, Remple JL, Hunter DB, Hahn T, eds, Raptor Biomedicine University of Minnesota Press, Minneapolis, Minnesota, pp.49-53. Ratcliffe DA. 1968. Changes attributable to pesticides in egg breakage frequency and eggshell thickness in some British birds. J Appl Ecol 7:67-105. Stickel J, Wiemeyer SN, Blus LJ. 1973. Pesticide residues in eggs of wild birds: Adjustment for loss of moisture and lipid. Bull Environ Contam Toxicol 9: 193- 196. Mauro L. 1987. Raptor physiology — hematology and blood chemistry. In Pendleton BG, Millsap BA, Kline KW, Bird DA, eds, Raptor Management Techniques Manual Natl Wildl Fed Sci Tech Ser No. 10. National Wildlife Federation, Washington, DC. Bortolotti GR. 1984. Criteria for determining age and sex of nestling bald eagles. J Field Ornithol 55:467-481. Strause KD, Zwiemik M, Im SH, Newsted JL, Kay DP, Bradley PW, Williams L, Blankenship AL, Giesy JP. 2007. Plasma to Egg Conversion factor for evaluating PCB and DDE exposures in Great Horned Owls and Bald Eagles. Environ T oxicol Chem (Submitted). Van den Berg M, Bimbaurn L, Bosveld AT, Brunstrom B, Cook P, Freeley M, Giesy JP, Hanberg A, Hasegawa R, Kennedy SW, Kubiak T, Larsen J C, Rolaf van Leeuwen FX, Djien Liem AK, Nolt C, Peterson RE, Poellinger L, Safe S, Schrenk D, Tillett D, Tysklind M, Younes M, Waem F, Zacharewski T. 1998. 137 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. Toxic equivalency fectors (TEFs) for PCBS, PCDDs, PCDFs for humans and wildlife. Environ Health Perspect 106:775-792. Neigh AM, Zwiemik MJ, Bradley PW, Kay DP, Park CS, Jones PD, Newsted JL, Blankenship AL, Giesy JP. 2006. Tree swallow (T achycineta bicolor) exposure to polychlorinated biphenyls at the Kalamazoo River Superfund Site, Michigan, USA. Environ T oxicol Chem 25:428-437. McLane MAR, Hughes DL. 1980. Reproductive success of screech owls fed Aroclor 1248. Arch Environ Contam Toxicol 9:661-665. US. Environmental Protection Agency. 1995. Great Lakes water quality initiative criteria documents for the protection of wildlife: DDT, Mercury, 2,3,7,8-TCDD, PCBS. EPA 820-B-95-0083. Washington, DC. Elliott JE, Norstrom RJ, Lorenzen A, Hart LE, Philibert H, Kennedy SW, Stegeman JJ, Bellward GD, Cheng KM. 1996. Biological effects of polychlorinated dibenzo-p-dioxins, dibenxofurans, and biphenyls in bald eagle (Haliaeetus leucocephalus) chicks. Environ T oxicol Chem 15:782-793. Elliott JE, Wilson LK, Henny CJ, Trudeau S, Leighton FA, Kennedy SW, Cheng KM. 2000. Biological effects of chlorinated hydrocarbons in osprey chicks. Environ T oxicol Chem 202866-879. Woodford JE, Karasov WH, Meyer MW, Chambers L. 1998. Impact of 2,3,7,8- TCDD exposure on survival, growth and behavior of ospreys breeding in Wisconsin, USA. Environ Toxicol Chem 17:1323-1331 . Wiemeyer SN, Bunck CM, Stafford CJ. 1993. Environmental contaminants in bald eagle eggs —1980-84 — and further interpretations of relationships to productivity and shell thickness. Arch Environ Contam Toxicol 24:213-227. Elliott JE, Harris ML. 2002. An ecotoxicological assessment of chlorinated hydrocarbon effects on bald eagle populations. Rev Toxicol 4: 1-60. Mendenhall VM, Klaas EE, McLane MAR. 1983. Breeding success of barn owls (Tyto alba) fed low levels of DDE and dieldrin. Arch Environ Contam Toxicol 12:235-240. Nisbet ICT, Risebrough RW. 1994. Relationship of DDE to productivity of bald eagles Haliaeetus leucocephalus in California and Arizona, USA. In Mayburg B- U, Chancellor RD, eds, Raptor Conservation Today, World working group on birds of prey, The Pica Press, Berlin, Germany, pp 771-773. 138 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. Anderson BW, Hickey JJ. 1972. Eggshell changes in certain North American birds. Proceedings of the International Ornithological Congress, The Hague, The Neatherlands. August 30-5 September, 1970. 15:514-540. Froese KL, Verbrugge DA, Ankley GT, Nieme GJ, Larsen CP, Giesy JP. 1998. Bioaccumulation of polychlorinated biphenyls from sediments to aquatic insects and tree swallow eggs and nestlings in Saginaw Bay, Michigan, USA. Environ T oxicol Chem 17:484-492. Sheffield SR. 1997. Owls as biomonitors of environmental contamination. United States Forest Service, North Central Research Station. General Technical Report NC-190. St. Paul, MN, USA. Blus LJ. 1996. Effects of pesticides on owls in North America. J Raptor Res 30:198-206. Springer MA. 1980. Pesticide levels, egg and eggshell parameters of great horned owl. Ohio J Sci 80:184-187. Wiesmuller T, Schlatterer B, Wuntke B, Schneider R. 1999. PCDDs/PCDFs, coplanar PCBs and PCBS in barn owl eggs fi'om different areas in the State of Brandenburg, Germany. Bull Environ Contam T oxicol 63:15-24. Helander B, Olsson A, Bignert A, Asplund L, Litzen K. 2002. The role of DDE, PCB, coplanar PCB and eggshell parameters for reproduction in the white-tailed sea eagle (Haliaeetus albicilla) in Sweden. Ambio 31:386-403. Eliott JE, Norstrom RJ, Smith GE. 1996. Patterns, trends, and toxicological significance of chlorinated hydrocarbon and mercury contaminants in bald eagle eggs from the Pacific coast of Canada, 1990-1994. Arch Environ Contam Toxicol 31 :354-367. Smith LM, Schwartz TR, Feltz K, Kubiak TJ. 1990. Determination and occurrence of AHH-active polychlorinated biphenyls, 2,3,7,8-tetrachloro-p-dioxin and 2,3,7,8-tetrachlorodibenzofi1ran in Lake Michigan sediment and biota. The question of their relative toxicological significance. Chemosphere 90:1063-1085. Seidensticker J C, IV, Reynolds HV, 111. 1971. The nesting, reproductive performance, and chlorinated hydrocarbon residues in the red-tailed hawk and great horned owl in south-central Montana. Wilson Bull 83:408-418. Giesy JP, Verbrugge DA, Othoudt RA, Bowerman WW, Mora MA, Jones PD, Newsted J L, Vandervoort C, Heaton SN, Aulerich RJ, Bursian SJ, Ludwig JP, Ludwig M, Dawson GA, Kubiak TJ, Best DA, Tillitt DE. 1994. Contaminants in fishes from Great Lakes-influenced sections and above dams of three Michigan rivers. 1: Concentrations of organic chlorine insecticides, polychlorinated 139 44. 45. 46. 47. 48. biphenyls, dioxin equivalents, and mercury. Arch Environ Contam T oxicol 27:202-212. Gasiewicz TA, Elferink CJ, Henry EC. 1991. Characterization of multiple forms of the Ah receptor: Recognition of 'a dioxin-responsive enhancer involves heteromer formation. Biochemistry 30:2909-2916. Karchner SI, Franks DG, Kennedy SW, Hahn ME. 2006. The molecular basis for differential dioxin sensitivity in birds: role of the aryl hydrocarbon receptor. Proc Natl Acad Sci USA 103:6252-6257. Blankenship AL, Zwiemik MJ, Coady KK, Kay DP, Newsted J L, Strause K, Park C, Bradley PW, Neigh AM, Millsap SD, Jones PD, Giesy JP. 2005. Differential accumulation of PCB congeners in the terrestrial food web of the Kalamazoo River superfund site, Michigan, USA. Environ Sci Technol 39:5954-5963. Holt JB. 1996. A banding study of Cincinnati area great horned owls. J Raptor Res 30:194-197. Bohm RT. 1985. Use of artificial nests by great gray owls, great horned owls, and red-tailed hawks in northeastern Minnesota. Loon 57:150-152. 140 Chapter 4 Evaluation of Risk Assessment Methodologies for Exposure of Great Horned Owls (Bubo virginianus) to PCBs on the Kalamazoo River, Michigan Authors: Karl D. Strausel, Matthew J. Zwiemik”, John L. Newsted”, Arianne M. Neighl, Stephanie D. Millsapl, Cyrus S. ParkT, Pamela P. Moseleyl, Denise P. Kayf, Patrick W. Bradleyl, Paul D. Jonesl, Alan L. Blankenship”, James G. Sikarskie§, John P. Giesy " # lZoology Department, Center for Integrative Toxicology, National Food Safety and Toxicology Center, Michigan State University, E. Lansing, MI 48824 IENTRIX Inc., Okemos, MI 48823 §College of Veterinary Medicine, Center for Integrative Toxicology, Wildlife Rehabilitation Center, Michigan State University, E. Lansing, MI 48824 H Department of Biomedical Veterinary Sciences and Toxicology Center, University of Saskatchewan, Saskatoon, Saskatchewan, Canada #Biology and Chemistry Department, City University of Hong Kong, Kowloon, Hong Kong, SAR, China. 141 ABSTRACT Dietary exposures of great horned owls (GHO; Bubo virginianus) at the Kalamazoo River, Michigan, were examined due to the presence of polychlorinated biphenyls (PCBS) in the terrestrial food web. Average potential daily doses (APDD) in GHO diets were 7- to 10-fold and 3-fold greater at a contaminated location than at a reference location for site-specific exposures quantified as total PCBS and 2,3,7,8- tetrachlorodibenzo-p-dioxin equivalents (TEQWH0-Avian), respectively. Wetland and aquatic prey species with links to the aquatic food web contributed significantly to PCB exposure and calculations of APDD. Measures of risk based on comparison of modeled dietary intake (APDD) to toxicity reference values (TRVS), using hazard quotients (HQS), varied with the selection of diet composition method (mass-basis vs. numeric- basis) with mass-basis compositions yielding greater HQs at both study sites. Risk associated with dietary exposures (“bottom-up” risk assessment methodology ) were below population-level effects benchmarks (HQ <1) for all dietary compositions, and were consistent with risk based on concentrations in tissues (“top-down” risk assessment methodology) and indicated PCBS at the site posed little risk to terrestrial raptor species. Co-located studies that evaluated reproductive health and relative abundance were consistent with results of the risk assessment. Measures of risk based on comparison to TRVS were consistent with direct measures of ecologically relevant endpoints of reproductive fitness, but uncertainty exists in the selection of threshold values for effects in GHO especially based on TEQWH0-Avian because of the absence of species-specific, 142 dose-response thresholds. Results of the evaluation indicated that the best estimate of risk is through application of a multiple-lines-of-evidence approach. Keywords: raptors, dietary exposure, bioaccumulation, 2,3,7,8-tetrachlorodibenzo-p- dioxin. 143 INTRODUCTION Great horned owls (GHO: Bubo virginianus) are a usefirl sentinel species for site-specific baseline ecological risk assessments (ERAS) at sites with large contiguous areas of contaminated environmental media. Their sensitivity to the toxic effects of some organic contaminants, such as organochlorine, organophosphate and carbamate pesticides and their relatively great exposure as apex predators makes them valuable as a surrogate species for estimating risk to raptors in the terrestrial food chain (Sheffield 1997). Great horned owls have been used successfully in site-specific estimates of risk posed by polychlorinated biphenyls (PCBS) at the Kalamazoo River Superfund Site (KRS S) in Kalamazoo and Allegan Counties, Michigan (Strause et al. 2007). This previous study utilized a “top-down” or “tissue-based” approach in which exposure was determined by measuring concentrations of PCBs in eggs and blood plasma of nestlings. The risks fiom exposure to PCBS were assessed in a multiple-lines-of-evidence ERA that included both comparing the measured concentrations of PCBs to toxicity reference values (TRVS) as well as concurrent measures of productivity and abundance in a “weight of evidence” to identify cause and effect linkages between the chemical stressor and any observed subOptimal p0pulation or community structure at the site (Fairbrother 2003). A second method for assessing risk to wildlife uses a predictive approach in which the exposure is inferred by measuring concentrations of the chemicals of concern (COCS), such as PCBs, in matrices other than the receptor of interest. This predictive approach is often referred to as the “bottom-up” or “dietary-based” approach. In this method the exposure-response is inferred from food chain modeling, based on site- 144 specific measures of concentrations of the chemical stressors (COCS) in wildlife food items or sediments or soils. In some cases, the actual dietary items can be identified via diet studies and quantified via forage base and food item sampling programs, while in other situations default or average values are used. Data specific to dietary composition and prey item COC concentrations are combined with receptor species’ ingestion rate and body weight parameters to compute an average potential daily dose (APDD). This estimated daily dose is then compared to a dietary TRV to assess potential risks at the site. This study described the site-specific dietary exposure pathways to PCBs for GHOS at the KRSS. Site-specific dietary exposures (expressed as APDD) were then compared to TRVS for adverse effects determined in controlled laboratory studies to calculate hazard quotients (HQS) based on predicted exposure through diet. Additionally, the site- specific method of estimating dietary exposures was compared to a literature-derived APDD. Due to limitations in the time and or resources available, few assessments apply both risk assessment methods simultaneously at the same location and time. Most assessments use the predictive method. While the two approaches are inherently linked, the accuracy and precision of the two methods are seldom compared. Therefore, the overall object of this study was to evaluate the results of the predictive assessment approach with actual measurements of exposure and population level effects at the same time at the same location. The two established methodologies of risk assessment (i.e., “top-down” vs. “bottom-up”) were compared to determine how similar the predictions of risk would be based on both total PCBs, and 2,3,7,8-tetrachlorodibenzo-p—dioxin equivalents (TEQWHO-Avian) calculated from aryl hydrocarbon receptor-active PCBs. 145 Meeting these objectives required completion of the following: 1) collection of GHO pellet and prey remains samples from active nest sites to identify dietary components and enumerate dietary composition; 2) collection of representative prey item samples for the categories of prey (e.g., passerine birds, mice/voles) that contributed most significantly to GHO diet; 3) determination of concentrations of PCBs and TEQWH0-Avian based on congener-specific measurements; 4) calculation and comparison of HQs based on total PCBs and total TEanoAvian between site-specific and literature-based diets; 5) comparisons of “bottom-up” and “top-down” estimates of risk based on total PCBS and TEQWHQAVim-l using the HQ methodology. Additionally, information on the PCB/ TEanomian concentrations between prey categories and food web sources (e.g., terrestrial vs. aquatic) was also assessed. METHODS AND MATERIALS Study Sites The KRSS includes 123 km of river extending from the city of Kalamazoo, M1 to Lake Michigan at Saugatuck, MI. The primary COCs are PCBS, including total 2,3,7,8- tetrachlorodibenzo-p-dioxin equivalents from non-ortho (coplanar) and mono-ortho PCB congeners. PCBs were used in the production of carbonless copy paper and paper inks for approximately 15 yr (USEPA 1976). During this period, recycling of paper, including some carbonless copy paper resulted in releases of PCBS to the Kalamazoo River. The Kalamazoo River was placed on the Superfund National Priorities List in August 1990 146 due to the presence of elevated PCB concentrations in fish, sediments, and floodplain soils (BBL 1993). Two sites within the KRSS were chosen for the GHO dietary study. These included the Fort Custer State Recreation Area (F C) and the former Trowbridge Impoundment (TB) (Figure 4.1). Characterizations of these sites have been provided in earlier assessments of GHO exposure at the KRSS (Strause et al. 2007). The FC site is an upstream reference area located approximately 7 km upstream of Morrow Pond Dam (the upstream limit of the KRSS) and 40 km upstream of TB. The FC site contains floodplain habitat similar to that present at TB and represents “current” regional background exposures in the watershed. The TB site is located in the Upper KRSS downstream of the point sources in the KRSS and it is one of three former impoundments in the Upper KRSS where removal of an in-stream dam to sill level has exposed former river sediments which are now heavily vegetated floodplain soils and riparian wetland habitat. The former TB impoundment includes the greatest area of contaminated soils (132 ha) and the greatest mean PCB soil concentrations (1.5 x 104 ng/g, dw) in the river floodplain. The FC and TB sites were selected to make direct comparisons between GHO responses on a “high potential exposure” vs. background “no elevated exposure” basis. The GHO populations studied at each site were restricted to mated-pairs occupying natural nests and artificial nesting platforms within the 100 yr floodplain. The propensity for GHOS to use artificial nesting platforms allowed for better experimental control compared to wildlife studies that rely exclusively on natural nests. Nest platforms were placed throughout the FC and TB sites, and at TB the artificial platforms were 147 I LOWER KRSS I I UPPER KRSS; I I REFERENCE I Trowbridge ; ; LAKE MICHIGAN 0 8 16 555% N Km T KALAMAZOO CALHOUN COUNTY COUNTY r‘ 1 Ill 2 Figure 4.1. Kalamazoo River great horned owl (B. virginianus) study sites including the Reference sampling location (Ft. Custer), the Upper Kalamazoo River Superfund Site (Trowbridge) and Lower Kalamazoo River Superfund Site (LKRSS) sampling locations. 148 placed to provide for a “worst-case exposure” by maximizing GHO foraging in the most expansive areas of the contaminated floodplain. Sample Collection The studies of GHO exposure to PCBS at the KRSS were part of a broader study to investigate PCB congener bioaccumulation and dynamics in the terrestrial and aquatic food webs of the Kalamazoo River floodplain that included representative samples from all tropic levels in resident terrestrial and aquatic communities (Blankenship et al., 2005; Kay et al. 2005; Millsap et al. 2004; Neigh et al. 2006a, 2006b, 2006c). All sample collections were completed within the 100-yr floodplain. Representative taxa included raptors (owls and eagles), passerine birds, aquatic and terrestrial mammals, fish, terrestrial and aquatic invertebrates, plants and co-located soil and sediment samples (Blankenship et al. 2005; Kay et a1. 2005). The principal components of the GHO food chain in the KRSS floodplain are likely to include terrestrial mammals and terrestrial passerine/aquatic birds, although limited numbers of aquatic invertebrates also may be eaten (Figure 4.2). Pellets and Prey Remains Site-specific studies of GHO diet were completed only at active nest sites. Pellets and prey remains were collected to determine the principal prey items that comprised the diet of nestlings. Diet investigations were undertaken in conjunction with other GHO study 149 swallows//v Small mammals & Passerine birds Waterfowl Aquatic emergent Muskrats Terrestrial insects /invertebrates Vegetation Sediments Soils Figure 4.2. Great horned owl (B. virginianus) food chain and exposure pathways at the Kalamazoo River Superfund Site (KRSS). 150 objectives that included collection of blood plasma from nestling GHOs and monitoring of productivity. To minimize nest disturbances and to avoid bias this required that collection of pellets and prey remains were coordinated with these other investigations. Pellet and prey remains were collected from the nest, the base of the nest tree and beneath adult perch trees during the nestling blood sampling event. Additional samples were collected from the base of the nest tree and beneath feeding perches after the nestling GHOS fledged from the nest (2 to 3 wk after blood was collected), and on 10-d intervals thereafter until no more samples could be collected. A final sample was collected from the nest during a “post—fledge” nest climb to clean and maintain the artificial nesting platform. These climbs occurred between 4 and 10 wk after the young had fledged. During each collection event, pellets and prey remains were systematically and completely removed from each location to reduce the chance of overestimating the frequency of occurrence of large prey species because of their tendency to be represented in more than one pellet or prey sample (Marti 1974). Each respective sample of pellets and prey remains was assembled and packaged in a plastic jar as a composite sample representative of each collection location (e. g., nest, nest tree base, feeding/roosting perch). Samples were labeled to identify the nesting pair, dated and transported to the laboratory where they were exposed to naphthalene while drying to eliminate invertebrate scavengers. Prior to processing, pellet samples were sterilized by autoclave. Prey Item Identification/Dietary Composition Relative proportions of prey items in the site-specific diet were determined by examining unconsumed prey remains (bones, fur and feathers of animals too large to consume 151 whole) as well as the skeletal remains in regurgitated pellets (Errington 1932, 1938; Hayward et al. 1993). All identifiable remains were sorted and quantified as to the minimum number of individuals from each taxon necessary to account for the assemblage of remains present in any given composite of samples. For mammalian prey items too large for owls to swallow whole (~ 100 g) and avian prey, the remains of the same prey item were frequently present in multiple samples. When this occurred, the items from within each discrete sampling event were examined together to reconcile the frequency of occurrence of larger prey and birds. Multiple prey item identification keys were utilized for comparative identification of mammalian and avian remains including owl pellet identification keys (Carolina Biological Supply Company, Burlington, NC, USA) and the vertebrate skeletal collection from the Michigan State University (MSU) museum. Avian remains (feathers) were identified with the aid of MSU Kellogg Biological Station bird sanctuary personnel. Prey items were identified to the best practical taxonomic classification and grouped by species/family and order into seven prey categories relating to food web (aquatic vs. terrestrial) and trophic level (primary vs. secondary consumers) position. These categories included: passerine (terrestrial avian); waterfowl (aquatic avian); mice/vole (terrestrial primary consumers, small mammal); shrew (terrestrial secondary consumers, small mammal); muskrat (aquatic primary consumers, medium- size mammal); rabbit/squirrel (terrestrial primary consumers, medium-size mammal); and crayfish (detritivor/aquatic primary consmner, invertebrate). The estimated dietary composition was based on the frequency of occurrence of all identifiable prey items and compiled on the basis of percent composition on a numeric basis (% number) and percent composition on a biomass basis (% biomass) (Wink et al. 152 1987). Percent biomass was calculated by multiplying each identified prey item by the mean adult weight (male +‘ female) for the particular species or family (Dunning, 1984; Baker 1983). The small number of individual prey items that could not be positively identified to family or order was limited to unidentifiable parts of terrestrial birds and medium-size mammals. For biomass calculations, these items were assigned a mass value equal to the average mass computed for the representative species identified for. that category at each respective nest site. Prey Collections for Chemical Analyses Prey species represented in site-specific and literature-based GHO diets were collected from the FC and TB study sites and analyzed for total PCBs. Analyses of pellet and prey remains samples collected from PC and TB identified six general categories of GHO prey. These included passerine birds, waterfowl, mice/voles, shrews, muskrats and rabbit/squirrel. A seventh category, crayfish, is represented in the literature-based diet and included in the diet analysis. Field sampling and processing methods for representative individuals from each of the seven prey categories are described below. Passerine birds collected from the FC and TB sites included the tree swallow (T achycineta bicolor), house wren (Troglodytes aedon), and American robin (Turdus migratorius). A single European starling (Sturnus vulgaris) also was collected at TB. All live birds were collected at the end of the nesting period. Adult wrens and swallows were captured with mist nets or a trap-door mechanism. Additionally, dead individuals found at nest boxes were salvaged for analyses (N eigh et al. 2006a, 2006b). Adult robins were collected using pellet guns [MSU Aquatic Toxicology Laboratory (ATL), 153 Unpublished]. The starling (carcass) was recovered beneath an active GHO nest. Birds were promptly euthanized by cervical dislocation and carcasses were placed in solvent- rinsed sample jars and frozen at -20°C. For chemical sample analysis, feathers, beaks, wings, legs, and stomach contents were removed and the whole body was homogenized in a solvent-rinsed grinder. Waterfowl species sampled included merganser (Mergus spp.), mallard (Anas platyrhynchos), wood duck (A ix sponsa), and blue-winged teal (Anas discors). Waterfowl sampling was not included in the MSU Kalamazoo River food web investigations. Waterfowl samples used in this GHO diet exposure study were collected in August 1985 by the United States Fish and Wildlife Service (U SFWS). The USFWS collected adult and immature ducks from five locations in the KRSS. Sampling locations included Morrow Pond and the Menasha and Trowbridge impoundments in the Upper KRSS, and the Allegan State Game Area and Saugatuck Lake downstream of Allegan Dam in the Lower KRSS. For chemical analysis, feathers, beaks and entrails were removed and the remaining carcass was homogenized in a solvent-rinsed grinder (MDNR 1987) Small mammals collected included mice (Peromyscus spp., Zapus hudsonius), voles (Microtus pennsylvanicus), shrews (Sorex cinereus, Blarina brevicauda), red squirrels (T amiasciurus hudsonicus) and chipmunks (Tamias striatus). All small mammals were trapped using pit-fall or Sherman live traps placed alternately within a 30 x 30 m2 sampling grid sited in the floodplain. Two sampling grids were located at FC and four grids were set up at TB (See Figure 1 of Blankenship et al. 2005). Captured species were sacrificed by cervical dislocation and carcasses were placed in solvent- 154 rinsed sample jars and frozen at —20°C. Prior to chemical analysis, stomach contents were removed and the remaining whole body (including pelage) was homogenized in a solvent-rinsed grinder (Blankenship et al. 2005). Muskrats (0ndatra zibethicus) were collected along the riverbank throughout FC and TB using body-gripping “conibear” traps. Samples were frozen at —20°C until processing for chemical analysis. Processing of whole-body samples included removal of the pelage, a coarse grind, and further homogenization in a commercial blender (Millsap et al. 2004). Crayfish (Cambarus and Orconectes spp.) were collected along the riverbank at F C and TB by use of wire minnow traps set adjacent to the small mammal sampling grids. For chemical analyses, the whole body was homogenized in a solvent-rinsed grinder (Millsap et al. 2004). Chemical Analysis — Extraction/Clean-up Concentrations of PCB congeners were determined by use of US. Environmental Protection Agency method 3540 (SW846). The details of the soxhlet extraction and sample preparation and clean-up have been described previously (Neigh et al. 2006a). Prey items were homogenized with anhydrous sodium sulfate (EM Science, Gibbstown, NJ, USA) using a mortar and pestle. All samples, blanks, and matrix spikes included PCB 30 and PCB 204 as surrogate standards (AccuStandard, New Haven, CT, USA). Extraction blanks were included with each set of samples. Quality Assurance/Quality Control sets composed of similar tissues were included with each group of 20 samples. Concentrations of PCBs, including di- and mono-ortho-substituted congeners were 155 determined by gas chromatography (Perkin Elmer AutoSystem and Hewlett Packard 5890 series II) equipped with a 63Ni electron capture detector (GC-ECD). Concentrations of non-ortho-substituted PCB congeners (coplanar) were determined by gas chromatograph mass selective detector (GC-MS) (Hewlett Packard 5890 series II gas chromatograph interfaced to a HP 5972 series detector). PCBs were reported on a mass wet weight (w) basis. A solution containing 100 individual PCB congeners was used as a standard. Individual PCB congeners were identified by comparing sample peak retention times to those of the known standard, and congener concentrations were determined by comparing the peak area to that of the appropriate peak in the standard mixture. Di-and mono-ortho-subsititued PCB congeners were detected by selected ion monitoring of the two most abundant ions of the molecular cluster and the limit of quantification was conservatively estimated (minimum surface to noise ratio of 10.0) to be 1.0 ng PCB/g, ww, using an extraction mass of 20 g, a 25 pg/ul standard congener mix and 1 ul injection volume. For coplanar PCB congeners, method detection limits varied among samples but were maintained at 5 0.1 ng/g, w for all samples using the sarnple- specific extraction mass and a minimum surface to noise ratio of 3.0. TurboChrom (Perkin Elmer, Wilmington, DE, USA) was used to identify and integrate the peaks. Total concentrations of PCBs were calculated as the sum of all resolved PCB congeners. Total PCB concentrations in waterfowl samples collected by the USFWS were quantified as Aroclor 1260 (MDNR 1987). 156 TEQ Computation Concentrations of TEanomm in prey item tissues were calculated by summing the products of concentrations of individual non-ortho and mono-ortho PCB congeners (77, 81, 105, 118, 126, 156, 157, 167, 169) and their respective World Health Organization (WHO) 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) toxic equivalency factors for avian receptors (Van den Berg et 'al. 1998). Polychlorinated-dibenzo-p-dioxins and polychlorinated-dibenzo-furans were not measured and were not included in TEQ computation. Whenever a PCB congener was not detected, a proxy value equal to one- half the limit of quantification was multiplied by the toxic equivalency factor to calculate the congener-specific TEQs. Co-eluting congeners were evaluated separately. PCB congener 105 frequently co-eluted with congener 132, congener 156 frequently co-eluted with 171 and 202, congener 157 co-eluted with congener 200, and congener 167 co- eluted with congener 128. In order to report the maximum TEQquMm the entire concentration of the co-elution groups was assigned to the mono-ortho congener. Among the six GHO prey categories analyzed at the MSU-ATL (excludes waterfowl samples), the maximum combined contributions to total TEanomm of congeners 105, 156, 157 and 167 was 4.2% (passerine), 46% (mice/vole), 5% (shrew), 1.2% (muskrat), <1% (rabbit/squirrel), and <1 % (crayfish), respectively. Individual non-ortho and mono-ortho PCB congener concentrations in waterfowl samples were estimated from the quantified Aroclor 1260 total PCB concentrations by multiplying the geometric mean total PCB concentration by a congener-specific fractional composition value (Schwartz et a1. 1993). The greatest observed congener- specific fractional value (percent composition basis) determined among four technical 157 Aroclor mixtures was selected to account for inherent differences in Aroclor batch production processes. Bioaccumulation factors of 10 and 3 were also applied to PCB congeners126 and 169 to account for observed selective enrichment (weathering, metabolism) of these two congeners, as observed by Schwartz et al. (1993). Using this conservative format, the combined contribution of congeners 105, 156, 157 and 167 to total TEQWHO-Avian in waterfowl was < 1%. Toxicity Reference Values In this study, TRVs were used to evaluate the potential for adverse effects due to PCBS including TEanOmm Ideally, TRVS are derived from chronic toxicity studies in which a total PCB or TEQeromm dose-response relationship has been observed for ecologically relevant endpoints in the species of concern, or alternately in a Wildlife species rather than a tradition laboratory species. Chronic studies should also include sensitive life stages to evaluate potential developmental and reproductive effects, and there must be minimal impact from co-contaminants on the measured effects. Toxicity reference values used in this assessment were based on values reported in the literature for no observable adverse effect levels (NOAELS) and least observable adverse effect levels (LOAELs) for total PCBS and TEQWHO-Avian. The dietary PCB NOAEL for GHO was based on the controlled, laboratory study on the reproductive effects of PCBS on the screech owl (Otus asio) (McLane and Hughes 1980). In that study, screech owls were fed a diet that contained 3 mg PCB/kg, ww. Conversion of concentrations in the diet to a daily dose (4.1 x 102 ng PCB/ g, body weight (bw)/d) was accomplished by use of the relationships presented by Sample et a1. (1996). At this dose, "158 no effects were observed on eggshell thickness, number of eggs laid, young hatched and young fledged. The LOAEL value of 1.23 x 103 ng PCBS/g, bw/d was estimated by multiplying the NOAEL by an uncertainty factor of 3. No additional uncertainty factors were applied to account for potential inter-taxon variability, since the NOAEL is in the range determined for the chicken, the most sensitive bird species tested (Platonow and Reinhart 1973; Lillie et al. 1974). No studies of the effects of TEanGmm were available for deriving TRVS, and no studies were found in which there was a closely related test species to GHO. A sub- chronic laboratory study (10 wk exposure period) by Nosek et al. (1992) found that intra- peritoneal injections of 2,3,7,8 - TCDD at concentrations of 1.0 x 103 pg TCDD/g/wk (1.4 x 102 pg TCDD/g, bw/d) caused a 64% decrease in fertility and a 100% increase in embryo mortality in ring-necked pheasants (Phasianus colchicus). This exposure concentration was used as the dietary TEQ-based LOAEL for GHO. A NOAEL was not directly available from this study and had to be derived from the limited dose data. Because effects due to the exposure were pronounced in the test subjects, a safety factor of 10 was applied to derive a NOAEL of 1.4 x101 pg TEQ/g, bw/d. (Table 4.1). Limitations of the study include the use of injections of 2,3,7,8-TCDD instead of feeding 2,3,7,8-TCDD contaminated food to the test species and the evaluation of TCDD exposure and not PCB-TEQ exposure. Because co-eluting congener contributions are included in some mono-ortho PCB congener concentrations used in this risk assessment the PCB-based TEQs may overestimate exposure relative to 2,3,7,8-TCDD. In this instance, the use of a TRV based on 2,3,7,8-TCDD exposure is likely to yield conservative estimates of risk when applied 159 Table 4.1. Toxicity reference values (TRVS) used to calculate hazard quotients for total Polychlorinated Biphenyls (PCBS) and 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQWH0-Avian) in great horned owl (B. virginianus) diet and eggs. dietary-based TRVS tissue-based TRVS (ng PCBS/g bw/d) (ng PCBS/g) (Pg TEQWHO-Avian/ g bW/d) (Pg TEQWHO-Avian/ g) a b c d NOAEL LOAEL NOAEC LOAEC e e e e “m“ PCBS 410 1,230 7,000 21,000 . f f h,i h,i TEQWHO'AW 14 140 1353 4003 a. No observable adverse effect level. b. Lowest observable adverse effect level. c. No observable adverse effect concentration. (1. Lowest observable adverse effect concentration. e. McLane and Hughes (1980). f. Nosek JA et al. (1992). g. Elliott JE et al. (1996). h. Elliott JE et al. (2000). i.Woodford JE et al.(l998). to PCB exposure. Also, tolerance to TEQ-based exposure by birds is species-specific (Woodford et al. 1998), and the TRVS derived from Nosek et al. (1992) are likely conservative because the Galliformes used in the study are among the more sensitive species to the effects of 2,3,7,8-TCDD (Hoffman et al. 1998). The available information indicates that raptors, such as the American kestrel (Falco sparverius), osprey (Pandion haliaetus), and bald eagle (Haliaeetus leucocephalus), are more tolerant than gallinaceous species to the effects of PCBs and TEQ (Elliott et al. 1997; Hoffman et al. 1998; Woodford et al. 1998; Elliott et al. 1996). Thus, for GHO a more closely related raptorial species such as American kestrels would be the ideal basis for TRVS. However, 160 in the few studies in which kestrels were exposed to PCBs, there was either inadequate dose-response information or incomplete assessment of ecologically relevant endpoints. Tissue-specific TRVS (egg-basis) for total PCBs and TEQwuomm used in previous assessments of GHO exposure at the site (Strause et al. 2007) are included (Table 1). The egg-based TRVS are included to aid interpretation of the “bottom-up” and “top-down” methodology comparisons completed in this study. Average Potential Daily Dose (APDD)/Risk Assessment The amount of PCBS ingested by GHOS was calculated using the wildlife dose equation for dietary exposures (U SEPA 1993). The APDDs for total PCBs and TEanGMm were calculated for GHOS using the site-specific diets for GHO determined in this study, and for comparison purposes a literature-based diet for a separate population of Michigan GHOS (Craighead and Craighead 1956). All APDDs were based on diets with prey composition compiled on a biomass basis (eq 1). Average potential daily dose calculations also included the incidental ingestion of floodplain soils that could potentially be associated with GHO foraging activity. APDD = Z (Ck X FRk X NIRk) (I) Ck = Geometric mean concentration of total PCBS or TEQWHO-A,,,an, w in the kth prey item category of GHO diet. FRk = Fraction of GHO diet (based on mass) represented by the kth prey item category. NIRk = Normalized GHO ingestion rate of the kth prey item (g prey/ g, bw/day, w). 161 Concentrations of PCBs and TEQWH0-Avian in representative prey items collected from the KRSS were determined using the methods described previously and are presented in the following section. FR.c (mass-basis) was determined for the GHO subpopulation cohorts at both PC and TB, and from a previous study (literature-based diet) of GHO populations in southeast Michigan. (Craighead and Craighead 1956). A conservative assumption for the value of FRk is that GHO at the KRSS will obtain 100% of their diet requirements from the 100-yr floodplain (site use factor = 1). NIRk (0.056 g/g bw/d) was derived from daily ingestion rates and mean body weights reported for GHO (Craighead and Craighead 1956). Additional PCB and TEanOmiafl dietary exposure from incidental soil ingestion was calculated for TB GHOS using both the site- specific and the literature-based dietary composition and geometric mean concentrations of PCBs measured for TB soils. Incidental soil ingestion contributions to dietary exposure were not calculated for the FC GHOs because of the very low concentrations of PCBS present in PC soils. Geometric mean and upper 95% confidence level (CL) geometric mean concentrations of total PCBs in TB floodplain soils (non-detects were removed from the data set prior to computing mean and upper 95%CL values) were obtained from previous investigations at the site (BBL 1994). Concentrations of total PCBS in soils were considered to be 85% bioavailable and contain 65% moisture (estimated from Studier and Sevick 1992) to make the dry weight (dw) soil concentrations comparable to wet weight concentrations in prey. The dietary fraction of incidental soil ingestion (2%) for GHOS was based on reports in the literature (U SEPA 1993). Dioxin equivalent concentrations in soils were not measured at the site and were 162 estimated from the Aroclor-based soil data using the methods described previously for waterfowl (Schwartz et al. 1993). Comparisons of potential hazard estimated for dietary exposure to PCBS, were based on HQS. Hazard quotients were calculated as the APDD (ng PCB/g, bw/d or pg TEanoAvian/g, bw/d) divided by the corresponding TRV (eq 2). APDD (ng PCBs/g bw/d or pg TEQs/g bw/d) d‘ (2) letary TRV HQ: Other lines of evidence from previously published studies on KRSS GHO populations were examined to minimize uncertainties in the analysis and calculation of risk from dietary exposure. These include the “top-down” risk assessment approach that quantified concentrations of PCBs present in GHO eggs and nestling plasma, and examined the effects of chlorinated hydrocarbons on egg viability through measurements of egg shell thickness and Ratcliffe index (Ratcliffe 1968; Hickey and Anderson 1968). A third and ancillary line of evidence investigated potential population-level effects of PCB exposures at the KRSS by monitoring productivity (fledgling success) and relative abundance between the contaminated floodplain habitat (TB) and the reference location (FC). By evaluating multiple lines of evidence together it was possible to provide the best available information for remedial decision-making at the site, especially when two or more lines of evidence converged on a common finding. 163 Statistical Analysis Both parametric and nonparametric statistics were applied depending on which assumptions were met. Concentrations of total PCBS and TEQWH0-Avian in prey populations from the site were analyzed for normality by use of the Kolrnogorov- Smimov, one-sample test with Lilliefors transformation. Concentrations of the COCS were generally log-normally distributed and therefore all data sets were log-transformed to more closely approximate the normal distribution. Data sets that were normally distributed were compared using a t-test. If the data did not exhibit a normal distribution, than a nonparametric version of the t-test (Mann-Whitney U test) was used. Associations between parameters were made with Pearson Product Correlations. Tests for normality and treatment effects (spatial trends) were completed using the Statistica (Version 6.1) statistical package (Statsoft, Tulsa, OK, USA). The criterion for significance used in all tests was p< 0.05. Statistical methods for comparing COC concentrations in GHO tissues (eggs, plasma), egg shell parameters (shell thickness, Ratcliffe Index) and call/response survey measurements of GHO abundance employed in the multiple-lines-of-evidence evaluation of site-specific risk to KRSS GHO populations have been described previously [Strause et al. 2007]. RESULTS Composition of the Diet of GHOS A total of 285 discrete prey items were identified in 59 pellet and prey remains samples collected from a combined total of seven active nests in the FC and TB study sites from 164 2000 to 2002. Excepting four post-fledge prey remains samples collected between June 4 and June 28, all samples were collected prior to June 1 in each year of the study, and, as such, the data provide a characterization of the spring or nesting- season diet for KRSS GHOS. Only prey items represented by the classes Aves and Mammalia were observed. Prey from classes Reptilia, Amphibia or Crustacae were not observed in the diets of GHO at the KRSS. Dietary compositions at PC and TB varied slightly when compiled on a class basis. Fort Custer GHOS consumed a slightly lesser proportion of birds and slightly greater proportion of mammals (birds; 15.5 % -numeric, l3.8%-mass: mammals; 84.5% -numeric, 86.2% -mass) compared to TB GHOS (birds; 27.5%- numeric, 24.8% -mass: mammals; 72.5% -numeric, 75.2% -mass) (Table 4.2). Similar results are produced whether one compiles the class-level data on either a numeric- or mass-basis with only a slight increase in the proportion of mammalian prey when diet is compiled on a mass- basis. Within-class differences were observed between the FC and TB diets of GHO. Large differences were observed in the proportions of passerine/terrestrial birds represented in diets of GHOS at PC and TB (11% versus 25.8% on a numeric-basis and 5% versus 22% on a mass-basis, respectively) and in the proportion of rabbits represented in GHO diets at PC and TB (46% versus 16% on a numeric-basis and 75% versus 50% on a mass-basis, respectively). Within-class differences also are seen between diet compilations based on % number vs. % biomass. On a numeric-basis, small mammals (mice/voles) and shrews account for up to 33% and 51% of the GHO diet at F C and TB, respectively. These proportions decrease to 2.2% (FC) and 6.2% (TB) of the diet on a 165 no 3 ed on . o.o 3 5:50 we 3. _ 3. cm H; 3. eeesmao was 0.8 gm can 3 H 0.2 93 efieé 3 SH ed 3 3 ea “Ea: 0.0 do «.0 H3 H: on 32% 3 3 ea oHe ode o. H m 06285 . mHm ‘ SH. New 3% mi 3% 22832 was 3 H3 3 3 S 3 ezaeteea mam 98 ea 9% 3m 0.: eeecemmee one a..." ”.2 3.. ms" m.mH 3% $30 «H2 .00 380333 owummmofi and H830 .E «3 do 338%“? owcmmofi Hemwwvfi E0: >95 88 u 75 . fix u H6 88 u 5 Gem u E 85 vammnébdfiotq H05 omnooqmexm 35 339-288qu 35 oEoogTSE 93:93:06 go 8853 905530 Ae\ev 885$ flmmm $32 mwmmm 2.88: Z .oHBSBHq 05 88w 98 AomHoommBHHmX mvac oHHm ugomsm Hoax OONmHHHwHM 05 Ha HHoHHHmomEoo $6 madam H30 BEE 3on ..N v 2an 166 .8888 Aafififimcu HHBoSHHHHHL BEES: 36-8338 vomuHHoEHHHH HR Home mHobEvm 89:03 bomofio annex d . .mEE osmium :m 89:05 bowofio 38.533 .0 .8888 p.95 HHBochHHxv EB vomuaoUEHH dd 98 meB 35893: EN 86205 >8ng ontommmm .9 .63 Hv Seemed es... esteemed e 8.88 329$ 167 mass-basis. Likewise, the combined proportion of rabbit and muskrat prey on a numeric- basis increases from 51% (FC) and 21.5% (TB) to 84% (FC) and 69% (TB) on a mass- basis. A diet compilation based on mass provides the most accurate characterization of the relative importance of prey to avian predators (Marti, 1987). Mass-based characterizations of KRSS GHO diet at PC and TB are compared to diet composition values for a nearby GHO subpopulation residing in southeast Michigan (Craighead and Craighead 1956) (Figure 4.3). Class-level differences in prey composition between KRSS GHOS and the literature-based (LB) values are greatest for classes Aves and Mammalia at FC (13.8% (FC) versus 68% (LB), 86.2% (FC) versus 31.5% (LB), respectively). Crayfish (class Crustacae) were also present in the LB diet of GHO. Within-class prey proportions for KRSS and literature-based GHO diets were similar with passerine/terrestrial birds and rabbits/muskrats comprising the great majority of bird and mammalian prey. Total PCB and T EQWHaAw-an Concentrations in Prey Concentrations of total PCBs and lipid content of 130 discrete whole-body prey item samples were determined. Prey items collected previously from the KRSS included 17 waterfowl samples that also were used to estimate GHO exposure to PCBs via the diet. Budget limitations prevented the collection and PCB/TEQ analyses of rabbit and grey/fox squirrel samples from FC and TB. Available data for chipmunk and red squirrel were used to fill this gap in the data base. Geometric mean concentrations of total PCBs in PC prey ranged from 2 ng PCBS/g, w in rabbits to 9.6 x101 ng PCBS/g, w in passerines 168 Figure 4.3. Site-specific great horned owl (GHO; B. virginianus) diet composition based on a biomass contribution basis for the Ft. Custer and Trowbridge sampling locations, and a literature-based (Craighead and Craighead 1956) diet composition for GHO populations in southeast Michigan. 169 PasseII'neITerrestIiaI/Unknown Méizgat ,, , /(5%l Waterfowl/Aquatic - » 2%: 9°/ GHO Dlet , < o) . $33: MiceNoIe Mass-baSIs eggs; <2%) Oi; Shrew (<1%) F: Eng? Rabbit/Squirrel/Unknown " ' (75%) Muskrat Passerineflerrestrial/Unknown (22%) Waterfowl/Aquatic Rabbit/SquirreI/Unknown (2%) (50%) MiceNole (6%) Shrew Trowbridge (<1%) (n = 240) Crayfish Muskrat (<1%) ..m- Passerine/Terrestrial/Unknown $333“ (65.5%) Rabbit/SquirreI/Unknown (23%) MiceNole (2.5%) Literature-based Waterfowl/Aquatic (n = 260) (25%) 170 (Table 4.3). Geometric mean concentrations of total PCBs in TB prey ranged from 5.6 x101 ng PCBs/g, w in muskrats to 1.3 x103 ng PCBS/g, w in passerines. Total PCB concentrations in waterfowl were 8.9 x102 ng PCBS/g, ww, and this value was used for both the FC and TB sites because of the uncertainty associated with residence and exposure of these mobile and migratory species. PCB concentrations in prey items from TB were significantly greater (small mammals, muskrats, crayfish; t-test p < 0.01; passerines, shrews; Mann-Whitney U test p < 0.01) than those from the upstream reference area at F C. Waterfowl and rabbit samples (TB sample size = 1) were not tested for significant differences. Geometric mean concentrations of TEQquMm in PC prey ranged from 0.52 pg TEQ/g, w in rabbits to 7.5 pg TEQ/g, w in crayfish (Table 4.4). Geometric mean concentrations of TEQWHQAvian in TB prey ranged from 1.3 x101 pg TEQ/g, w in muskrats to 7.1. x101 pg TEQ/g, w in rabbits. Using congener-specific fractional composition values (Schwartz et al. 1993) conservatively estimated TEQwuomm concentrations in 17 waterfowl samples were 2.4 x102 pg TEQ/g, ww. Prey items from TB contained significantly greater concentrations of TEQ than those from FC (t-test p < 0.02) with the exception of muskrats and small mammals, which were not statistically different (Mann-Whitney U test p = 0.26 and p = 0.22, respectively). Waterfowl and rabbit samples (TB sample size = 1) were not tested for significant differences. Contributions to total TEQWHOMam from the four non-ortho and eight mono-ortho PCB congeners showed that over 90 % of total TEQ was contributed by non-ortho congeners 77, 81 and 126 for all prey item categories at each of the two sampling 171 .AHod w no .8000 um 00 803880008 85 808% >385?me 80 808000080 mOHH H38 038345lo... 8888080 0308 .H0.H 028 000888 8 H000: AS 88930 08803on HHH 0000808008 mom 30H .0 00808080 0.50.8 .H0.H 038 888.50 8 H0000 AHV H8880 H08 H08 Amv 88930 .8800 am HHH 808000080 mom H38 808 0308000 .0 ..H0>H.H 06 8 0380 008208 3028800818 08800 80382 wHHHHnHHHHmm 808800 H08 8088: 802500 002wa H00 80B 08 H0>H~H 0083M 08. 00 88.82 m 808 0080:00 803 00380 150.8803 .0 08888008 080003 380088 .8 030380898 8 H0000 Q 038 808002 H08 A3 3238 080. A8 083 0000: AS mHHHHHSm 0wHVH0~H>00HH H08 A$ 30.8 808094 H08 ANV 30:88 08..— A3 083 008: H0080 .E H 8038000080 mom H800 808 0888.0 .0 SH SH .8. 2.3 I es 2 Se 8 a. a - R H. 880 3.4 men men men . H H2 4 N e - H e 8088 SN. a .8 N: - E A 8.“ mm 2 em - w 4 H0292 03 . $3 *8 one I R S 09m 2 w H: I N 2 seem A? 8H ...S 8 - 8 em 03 a 2 2: - m H eHeZeeHE owe HRH 8 08.8 I 02 2 3m Hm: 8 80.8 I 02 AH agree? 8.4 SE .88 80.8 I S cm SH. 8. ea 08; I e H H 03888 08 as: A8 502 8H: H0 .330 8.0880 88 2 2e: 8 $3: 80880 00:8 z 82530.5. 80:0 .5 ..Ho>Hm 00883 08 8 008 .025 808 880:8 88H 88H HHH A33 memom mav Ammmvmv 03885 0000803033 H88. .8 80808008 Ago $38 H0>0H 0082.88 X3 H08: H08 808 08008000 .34 030B 172 .Amod W 8 88:0 uh: 8 88888800 8% 8808 3:88:88 08 80888800 OE 868638 :1. 80888080 88:8 80.: 88> 08880 8 600: AS 88:30 08888:. H8: 80888080 82-08015. .6 .8888080 80:8 8.: 05:8 08888 8 608 AS :088 68 68 Am: 88:30 8850 an: 8: 80888080 §H><-0m30m:.:. :38 808 8.88000 .0 .83: 08. 8 888 008688 :80: 88008 :0 0880: 80800: 8:880 8088306 68 8088 8038: 60636 88 803 68 83: 008863: 0:: 8 80800: m 80¢ 6080:00 083 87:80 :38883 H: 88888800 08808: 3.8088 :0 038.8888 8 608 AD 8:8 80:88 68 Am: 30:830 008 .on H83 008: AC 8:88 08838:. 68 Av: . 838 8088 68 AN: 30:830. 008 Am: 883 088: 8830 an: 8: 808.8088 5H><-0:30m:,:I :38 808 0:.88000 .8 8: 8H .8 8 - H.m 2 8.0 mm 2 mm - 2 HH H880 n3 . E H5 :. H H8 H3 N2 3 I 2.0 o 8.0882 :8 2. 2 a. - as h 88 H.H H.H 2 - So H. “332 . 88 E E 8 I HV 2 85 ”H8 8H 3 I as 2 58 EH. 3.0 3.0 8 I :3 ON and H8 H8 8 I 8 NH 2028:: 08 8 3m 8.» I on S 08 8 . $8 8.5 I 8 t 08833 . . . . . 2H: 0? EH .3 ETHH 2 OH H. 2 mm 842 HH a .38 8 HHS: 8 82 H85 8 $3: 08880 . 83H 2 8H: 8 $3: 08880 08m 2 82.5389 88:0 4.: 83: 008863: 86. 8 00:0 03980.8 6080:00 88: >08 8: A33. w\SH><.080m.:I mm: 82.230er 38:88:00 88:68. 68086803888 mafia :0 80888800 AAU $33 :08: 0086800 $3 88:: 68 808 8.88000 .86 28:. 173 locations excepting TB mice/voles (45% contribution) (Figure 4.4). The same three coplanar congeners were among the three greatest contributors to concentrations of total TEQ for all prey item categories at each sampling location excepting TB mice/voles and TB rabbits (chipmunk). The three greatest PCB congener contributors and their combined contribution to total TEQ for each prey item category included: PCB 77>126>81[FC passerine(95.4%), TB passerine(97%), FC and TB waterfowl(98.l%- estimated), FC and TB crayfish(99%)]; PCB 77>81>126 [TB muskrat(98.6%)]; PCB 126>77>81 [FC muskrat(99.7%), FC rabbit (chipmunk and red squirrel)(99.1%), TB shrew(93.3%)]; PCB 126>81>77 [FC shrew(99.4%), FC mice/vole(97.1%)]; PCB 126>77>118 [TB rabbit (chipmunk)(97.9%)]; PCB 105>81>126 [TB mice/vole(71.l%)]. Average Potential Daily Dose (APDD) Average potential daily doses for GHOS were calculated based on geometric mean concentrations of both total PCBs and TEQWHQAVian of each prey item category for the numeric- and mass-based range of dietary composition at the FC and TB study sites. Calculations of both total PCB and TEanGMm exposures at TB included contributions from incidental soil ingestion. Based on site-specific diet and prey item COC concentrations, GHO ingestion of total PCBS were from 7 to lO-fold greater at TB than at PC, and TEQSWHQ-AVian were 3-fold greater at TB than FC (Table 4.5). Average potential daily doses calculated using the upper 95%CL (geometric mean) of total PCBS and TEQWHGAvian displayed a range of differences that were similar (6 to 7-fold difference and 2-fold difference, total PCBS, TEQWHQ-AVian, respectively) to values of APDD based on the geometric mean. 174 100% ‘ Percent of total TEQWHMM N ‘3. o\ 0%‘ 100% ‘ 80% ‘ 60% ‘ 40% ‘ Percent of total TEQWHMM N 2 e\ 80% ‘ 60% “ 40% ‘ PCB Congener E Other PCB -156 PCB -157 PCB -167 PCB -169 PCB-118 "r 5=' - WIIIIIIIIIIIIIIIWHH PCB-105 W wllllllllll |||||| PCB-126 ‘llll' § PCB-81 PCB-77 \ \ /// i IIWWIIIIIII Illlllll. PasserineWaterfowlMice/Vole Shrew Muskrat Rabbit Crayfish Ft. Custer GHO Prey Item 0%l Figure 4.4. Passerine WaterfowlMiceNole Shrew Muskrat Rabbit Crayfish Trowbridge GHO Prey Item Percent contribution of polychlorinated biphenyl (PCB) coplanar and mono- ortho-substituted congeners to total 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQWH0_Avian) in great horned owl (B. virginianus) prey at the Kalamazoo River. 175 Table 4.5. Range of Average Potential Daily Doses (APDD)a based on geometric mean and the upper 95% confidence level (U 95% CL) of prey items for total Polychlorinated Biphenyls (PCBS) (ng PCBS/g bw/d) and 2,3,7,8 tetrachlorodibenzo-p-dioxin equivalents (TEQWH0-Avian) (pg TEQ/ g bw/d) when assuming two different dietary compositions for great horned owl (GHO) at the KRSS. b Ft‘ Custer Trowbridge PCBs TEQs PCBS TEQs PCB Based Dietary Models Site-Specific APDDC Geometric Mean 3 - 5 0.676 — 1.25 30 — 37 2.39 — 3.81 U95% CL 7— 10 2.03 —3.76 59—61 5.34—6.98 Literature-Based APDDd Geometric Mean 4 — 5 0.428 — 0.549 38 — 60 2.72 — 3.97 U95% CL 9— 12 1.34— 1.75 80— 127 6.81 — 10.18 a. The range of calculated APDD results from using diet estimations based on both total frequency (numeric-basis) and biomass contribution (mass-basis ) (see Table 2). b.Includes incidental ingestion of floodplain soils at the former Trowbridge impoundment. c.Based on results of field collected GHO pellets and prey remains from active nests at each Kalamazoo River study site. d.A study of GHO diet in Washtenaw County, Michigan (Craighead and Craighead 1956). Comparisons of geometric mean, mass-based ranges of APDD between the site- specific and literature—based GHO dietary compositions yielded APDD values for total PCB exposures at FC that were equivalent (5 1.5-fold difference), and TB total PCB APDD values that differed by a factor of 1.6 (literature-based > site-specific APDD) (Table 4.5). Average potential daily dose values for mean TEanGMm were 1.6 to 2.3- fold greater for FC site-specific based dietary exposures and equivalent for TB based exposures. The literature-based TB APDD calculations for both total PCB and TEQWHO- 176 Avian included contributions from incidental soil ingestion consistent with the calculations for site-specific exposures at TB. Average potential daily doses (site-specific vs. literature-based) calculated using the upper 95%CL for total PCBS displayed a range of values similar to mean-based APDDs (F C exposures were equivalent, literature-based APDDs were 2.1-fold greater for TB exposures). Average potential daily doses calculated using the upper 95%CL concentrations of TEano-Avian in prey produced site-specific APDDs that were 1.5 to 2.1-fold greater than literature-based values at PC, and APDDs that were equivalent at TB. The greatest calculated APDDs for GHOS at the KRSS originated from a mass- based dietary compilation. The IO-fold greater APDD at TB than at PC was consistent with the significant differences in total PCB and TEQwquvian concentrations in prey items collected from the two sites. The APDDs based on total PCBS and TEQSwnoAvian for the site-specific and literature-based diets were less than 3-fold different for both PC and TB. This indicates that the moderate differences in dietary composition observed at a class and within-class level between the two studies did not influence APDD to any great extent. Due to greater concentrations of COCs or greater proportions in the diet, exposures are often dependent on a few types of prey. This phenomenon was observed for GHOS at the KRSS where two to four prey item categories combine to account for over 90% of the APDD at any given site and diet composition. At FC, mass-based APDDs for geometric mean total PCB concentrations in prey show that waterfowl (91% of APDD) and passerines (5%) drive exposures for the site-specific diet, and the same two prey items, albeit in 3n inverted ratio: waterfowl (25%) and passerines (72%), also 177 drive the literature-based exposure (Figure 4.5a). At TB, passerines (45%), rabbits (43%) and soil ingestion (5%) figure predominantly in APDDPCBS for site-specific exposures and also literature-based exposures (passerines (62%), rabbits (26%), soil ingestion (5%)). The principal prey items responsible for APDDs calculated for TEanaAvian include the same prey identified for APDDPCBs but in some cases additional prey contribute to the 90% threshold. At FC, waterfowl (97%) and rabbits (2%) drive APDDTEQ for site- specific exposures and literature-based exposures come from waterfowl (62%) and passerines (35%). At TB, rabbits (52%), passerines (18%), soil ingestion (16%), and waterfowl (10%) drive APDDTEQ for site—specific exposures, and literature-based exposures come from passerines (34%), rabbits (28%), waterfowl (16%) and soil ingestion (14%) (Figure 4.5b). Assessment of Hazard Hazard quotients were calculated for each location based on the site-specific and literature-based APDDs for total PCBs and TEQWH0-Avim To conservatively estimate potential hazard to resident GHOS at the KRSS and to capture the broadest reasonable 178 43.8: SB: 40.4.3: :85. m.— mw m1. mm m.— wm m.— mm 32.532... 33:0 an. _ m_mmm moms. I :35. 3.52.806 Eozoeev means. =me E— __ow a .6232 E 8.; 5:50 ”69 325 N 8833826; U csocxcgotaawannam I gecxcngmmtotmoctowmma % . o._. 0N o.m 0.? o.m 0.0 OK 0d 0d odw Rep I M 61“"""‘°“Moar 6d no.3»: cams. 40.33: :82 m... mm 9 mm m._. Mm. 9. mm ., wattage... .395 .E $QO moms. I cums. 056E060 6.382 Beams. =95 =___ :8 mm” 56.3.2 5 as amigo Have 295 fi 658338363 U gecxcgotsowannmm I Eocxcngmotmtmoctmmmmm Z Rap/M01 5 I 9806 Bu 179 4. .920 mmmmnammav mqumomEoo EBB 36 33905528: 28 Ammv ommoomméxm Ea .momowwtngeh $on owvunBoHH use .8550 .E 3. 3235588 62: 30338 $3 8&5. use 888 3.508on 08. no woman. Anamomfimx? .3 <50 3an 3on 8m 925 .3339 @Omflr mm ”Dam/b omow $3 @333 ommuozw castomaomb $83??? aflowwiéuaonEoHoEombofiwK.QN 3 cosmowfi :8 #85205 98 E05988 >on Ramadan none mo 5359.550 308$ QC .330 $39325 mqumomaoo @806 ad 339-0538: 23 Ammv oEoommézm 98 £8 owEBBoHH $on 03:83»lo wad H830 .E E mdouubaoonoo H32 Bouwman. £63 .53: 98 888 033.8% 06 no woman. mazmcmfiwha .3 H30 380m 3on 8m A33 c8339. mRmOm ma ”Gama/b omow 3.2% finfiabm owmcgm 50% T3233 nowaquoaobom :33 8 doumomfl :8 35203 was “noqomaoo than BASES :98 mo nousngqoo 388m Add .men oSmE '180 range of variability in characterizations of prey item COC concentrations and composition of the GHO diet, HQs are calculated from the range of APDD values encompassing the geometric mean and associated upper 95%CL values for each respective prey item and the dietary proportion contributed by each prey item compiled on both a numeric- and mass-basis. The range of HQs discussed for the NOAEL and LOAEL effect levels (HQNOAEL/HQLOAEL, respectively) will typically represent potential hazard associated with exposures to geometric mean concentrations for a numeric-based diet (low range) up to the upper 95%CL concentrations for a mass-based diet (high range). All HQs (total PCBS and TEanoAvian) for site-specific and literature-based diets determined for both PC and TB geometric mean and upper 95%CL exposures were less than 1.0. The maximum FC HQNOAEL for total PCBS was 0.02 and 0.03 for the site- specific and literature-based diets, respectively (Table 4.6). The maximum TB HQNOAEL for total PCBS was 0.15 and 0.31 for the site-specific and literature-based diets, respectively. The maximum FC HQNOAEL for TEQWHOAvim ranged from 0.27 to 0.13 for the site-specific and literature-based diets, respectively; TB HQNOAEL/I‘EQ ranged from 0.5 to 0.73 for the site-specific and literature-based diets, respectively. 181 Table 4.6. Hazard Quotient (HQ) values based on geometric mean and the upper 95% confidence limit (U 95% CL) of average potential daily doses (APDD) of total Polychlorinated Biphenyls (PCBS) and 2,3,7 ,8—tetrachlorodibenzo-p-dioxin equivalents (TEQWHQ-Avian), when assuming two different dietary compositions for great horned owl (GHO) at the Kalamazoo River Superfund Site (KRSS).a Ft. Custer Trowbridge HQ NOAEL HQ LOAEL HQ NOAEL HQ LOAEL PCB Based Dietary Models Site-Specific ° c _ Geometric Mean 0.01 <0.01 0.0., _ 0.09 0.02 0.03 U95% CL 0.02 0.01 d 0.14 - 0.15 0.05 Literature-Based Geometric Mean 0.01 <0.01 0.09 — 0.15 0.03 — 0.05 U95% CL 0.02 — 0.03 0.01 0.20 — 0.31 0.07 — 0.10 TEQWHGM“ Based Dietary Models Site-Specific Geometric Mean 0.05 — 0.09 0.01 0.17 - 0.27 0.02 -— 0.03 U95% CL 0.15 —- 0.27 0.01 -— 0.03 0.38 -— 0.50 0.04 - 0.05 Literature-Based Geometric Mean 0.03 — 0.04 <0.01 0.19 -— 0.28 0.02 - 0.03 U95% CL 0.01 - 0.13 0.01 0.49 — 0.73 0.05 — 0.07 a. Toxicity reference values used to calculate HQs are provided in Table l. b. Based on results of field collected GHO pellets and prey remains from active nests at each Kalamazoo River study site. 0. The range of calculated APDD results from using diet estimations based on both total frequency (numeric-basis) and biomass contribution (mass-basis ) (see Table 2). d. A study of GHO diet in Washtenaw County, Michigan (Craighead and Craighead 1956) 182 DISCUSSION Dietary Composition The three-year sampling program at active GHO nests for pellets and prey remains was an effective approach for characterizing site-specific exposures of nestling GHOS to the COCS at the KRSS, and to concurrently allow for non-intrusive monitoring of GHO productivity at each nest. Additionally, some potential biases commonly associated with . pellet and prey remains sampling were also addressed in this study. Our approach capitalized on GHO preferences to use nests built by other bird species or in this instance artificial nesting platforms that were located at appropriate floodplain locations in the KRSS. In doing so, we successfully induced resident GHOS to occupy areas of the site having maximum exposure potential. Only one of the active nests sampled for pellets and prey remains was a natural nest, all remaining nesting activity included in the diet characterization study occurred at artificial nesting platforms. Additional advantages of GHO behavior incorporated into this study included their propensity to forage within relatively small areas because of their sedentary lifestyle and highly versatile prey capture ability (Marti 1974). In this study, sampling of pellets and prey remains from beneath feeding perches and nest trees (ground collections) was augmented with collections from active nests during the blood sampling event and again afier fledging was complete. Nest sites were the best locations for collecting avian prey remains, particularly feathers that could be positively classified as evidence of owl predation. Nest collections eliminated a significant source of uncertainty associated with feather remains collected on the ground 183 beneath or in the general vicinity of active nests and feeding perches for which solid evidence of owl predation was mostly lacking. Feeding studies with captive great horned owls (Errington 1930; Glading et al. 1942) showed that pellets quite consistently reflected the food habits of adult and juvenile owls. Potential biases associated with pellet studies (under-representation of very small prey, prey with more easily digestible components, boneless pellets from newly hatched owlets) or prey remains collections (over-representation of larger species and avian prey) can be adequately managed through collections of both types of samples, segregation of samples among active nests, and reconciliation of all concurrently collected samples. Very small prey items (e.g., very young animals) are unlikely to carry high COC concentrations because of a very limited period of potential exposure, and even in the exception will contribute negligibly to APDD because of their low mass. Prey with easily digestible components may still be identified in prey remains collected from the nest, and the small numbers of prey that are completely absorbed by recently hatched owlets will most likely be adequately represented in subsequent pellets. Studies of the dietary habits of North American GHO populations show the GHO is an opportunist hunter, plastic in foraging behavior, a generalist in prey selection with the broadest diet of any North American owl (Marti and Kochert 1996). Dietary preferences depend upon habitat, season, and prey vulnerability, and GHOS are capable of expressing the most diverse prey profile of all North American raptors (V oous 1998). Contributing factors to the species’ broad diet include a large body size and crepuscular/nocturnal activity range. Maj or determinants upon prey selection by any individual include habitat, prey abundance and prey vulnerability (a measure that 184 combines ephemeral and interrelated determinants of prey density/prey behavioral patterns, habitat condition, and seasonality) (Houston et a1. 1998). In general, temperate North American GHO populations feed predominantly on terrestrial mammals followed by terrestrial and aquatic birds, and a minor mix of reptiles, amphibians and arthropods (Marti and Kochert 1995; Murphy 1997). Great horned owl diets vary between physiographic regions (Wink et al. 1987) and even among individual nesting territories when land use and habitat type distributions diverge within distinct physiographic units (Marti 1974). Great horned owl diets can also show significant temporal variation when pronounced changes in prey availability occur due to natural small mammal population cycles or anthropogenic modifications to habitat or prey populations (Adamcik et al. 1978; Fitch 1947). In addition, GHO diets also may vary with seasonal changes in prey vulnerability, although this source of variation tends to be minor compared to diet alterations stemming from differences in habitat and temporal prey availability (Wink et al. 1987; Fitch 1947; Errington et al. 1940). If present, significant seasonal variations in GHO diets may originate from changes in prey vulnerability caused by a combination of factors including vegetation changes, altered activity patterns of prey and GHOS, day length, GHO reproductive cycles, prey hibernation patterns, prey migration patterns, and prey reproductive lifecycle events (e. g., mating, dispersal of young). Great horned owl foraging preferences are difficult to predict from surveys of prey populations in most North American temperate habitats and attempts to correlate GHO predation preferences with prey abundances have yielded mixed results (Murphy 1997). This is due in part to the fact that prey density apparently has little effect on prey 185 vulnerability in some GHO territories (Peterson 1979; Adamcik et al. 1978). GHOS tend to display a density-independent dietary relationship to prey species. Changes in prey vulnerability do not necessarily correspond with changes in numerical status. Some species remain more vulnerable to GHO predation regardless of annual fluxes in abundance/density, even in periods of low population densities (Peterson 1979; Errington 1932) Attempts to correlate GHO predation preferences to habitat types show greater success where GHO’s food habits appear to depend largely upon where the bird is situated because many birds studied seem to limit their activities to a few acres of certain favorite habitat (Enington 1932; Fitch 1947). While there is evidence to support a strong interaction between GHO foraging preferences and habitat (Rusch et al. 1972), there were still instances where some GHOS did appear to respond opportunistically to the availability of certain prey types. (Marti and Kochert 1996). However, wetlands habitats appear to be a notable exception to the habitat-prey use relationship identified for GHOS where studies of GHO use of wetland-dependent prey and proximity and extent of wetlands within nesting territories have shown almost no relationship between habitat and prey type. In these studies, GHOS sought wetland prey regardless of proximity or abundance of wetland habitats. This may stem from the fact that prey species may be more available and vulnerable due to high prey density, high prey diversity and abundance, and more favorable locations and numbers of elevated hunting perches in wetlands and wetland edge habitats (Murphy 1977; Houston 1998). Our studies of GHO predation at the KRSS included efforts to control for spatial and temporal variability in owl diets. Great horned owl nesting platforms were 186 specifically located in riparian floodplain habitats that were buffered from most human disturbances and situated within 100 m of the river to provide uniformity in foraging habitat, available prey populations and habitat-dependent influences on prey vulnerability. Nest trees were selected afier completing a qualitative survey of nesting habitat quality so as to provide an optimal mix of cover and foraging habitat for breeding owls. Pellets and prey remains from multiple years were collected only during active nesting and brooding periods of the annual reproductive lifecycle to provide uniformity in environmental cues on GHO behavior and seasonal influences on prey availability/vulnerability. Site-specific and Literature-based Diets Because the dominant factors influencing GHO diet originate from spatial differences in habitat type and temporal alterations in prey populations, a literature-based diet selected for use in the absence a site-specific value must match the physiographic region and dominant habitat types at the site. If multiple studies of equivalent quality are available, temporal considerations can also be addressed. For instance, in the 1940s, nesting habitat of GHO populations in southeastern Michigan (Superior Township, Washtenaw County) included plant and animal community assemblages that were very similar to those present at KRSS (Craighead and Craighead 1956). When dietary composition for GHOS from KRSS was compared to the Washtenaw study, differences were observed that were principally related to differing proportions of rabbits and passerine/terrestrial birds (Table 4.2). Kalamazoo River owls consumed greater proportions of rabbits, and Washtenaw County GHOS consumed greater proportions of passerines/terrestrial birds. The greater 187 proportion of birds in Washtenaw County GHO diets is directly attributable to a greater number of ring-necked pheasants in the diet that were at their historical peak of abundance in Michigan during the 1940’s. However, their number significantly decreased by the 1980’s (Luukkonen 1998) and as a result, this loss of an important dietary item was compensated for by increased GHO consumption of cottontail rabbits in the KRSS study (Springer and Kirkley 1978; Peterson 1979). The impact of wetlands on GHO dietary composition also needs to be taken into account in that wetland habitats are well represented and in close proximity to GHO nesting territories at KRSS and the pr0portion of species with a direct-link (habitat-based; e.g., muskrat, waterfowl, crayfish) and indirect-link (foraging-based; e.g., insectivorous passerine birds, bats, weasels) to the aquatic food-web may have an important contribution to the overall diet. The combined diet composition for KRSS GHOS (F C + TB) shows that 8.8% (numeric-basis) and 21% (mass-basis) of GHO prey originated wholly or in-part from the aquatic food-web. In comparison, aquatic prey comprised 9% of Washtenaw Co. GHO diet on both a numeric- and mass-basis. A review of GHO diet studies available in the literature showed that aquatic prey are common in diets of GHOS residing in close proximity to wetland habitat types, and in select Western and upper Midwestern habitats, the proportion of aquatic prey (numeric-basis) in resident GHO diets can exceed 20% and 50%, respectively (Bogiatto et al. 2003; Murphy 1997). Average Potential Daily Dose (APDD) A conservative approach was used to calculate APDD values for GHOS such that when site-specific PCB or TEQWH0-Avian concentration data were not available for a specific 188 component of GHO diet (e. g., rabbits, grey/fox squirrels), the shortcoming was addressed by using site-specific data for red squirrels and chipmunks to represent potential COC exposures for the group. This approach incorporated a conservative estimate of potential exposure since the omnivorous diets of squirrels and chipmunks place them in a higher trophic level compared to rabbits. The conservative nature of this substitution is evident in that the total PCB and TEQWHOAvian concentrations used to calculate rabbit contributions to Trowbridge APDD were one to two orders of magnitude higher than concentrations expressed by both terrestrial and aquatic herbivorous counterparts to rabbits at the site (mice/vole, muskrat) (Table 4.3). A similar approach was used to address the absence of site-specific data for pheasants and other galliform prey where these species were grouped with passerine prey. Passerines at the site included insectivorous representatives (e. g., tree swallows) with forage-based links to the aquatic food-web at the site. Passerines had the highest concentration of total PCBs and second highest concentration of TEanaAvian among prey groups from the TB site (Table 4.3). The USFWS waterfowl database also provided an unbiased characterization of potential exposure through this aquatic pathway by including representative concentrations from both piscivorous (merganser) and omnivorous (mallard) feeding groups. Great horned owls prey indiscriminately upon waterfowl and a variety of wading birds (Rusch et al. 1972). Piscivorous waterfowl and shorebirds have been shown to accumulate total PCB and TBanoAvian concentrations that are ten- to fifteen-fold greater than their closely related avian counterparts who are more herbivorous (Jones et al. 1993). Processing of small mammals and avian prey that included the removal of stomach contents (both prey types) and feathers, beaks, wings, legs (avian prey) is a common practice in exposure and 189 effects studies and is typically used to conservatively estimate soil to organism bioaccumulation factors. The method contributed to conservative measurements of the COCS in these samples because the substantial mass excluded from analyses (keratin, herbaceous forage) contained much lower contaminant concentrations compared to the remaining tissues (predominantly muscle and lipids) analyzed for the target organism. Finally, incidental soil ingestion was also included in the site-specific APDD calculations to account for soil that may be associated with the pelage of small mammalian prey that tunnel through vegetation or use burrows for shelter, nesting, or food storage, and also present in avian prey that consume grit and associated soil particles as a normal course of their foraging activities (Mayoh and Zach 1986). A conservative approach also was used to estimate TEanGMan from the PCB Aroclor data for waterfowl and soils in the calculation of APDD—”5Q for GHOS. Dioxin equivalent concentrations (TEQWHGAvm) were estimated by selecting the greatest proportional contribution of each individual non-ortho and mono-ortho PCB congener across four technical Aroclor mixtures as the fractional composition value for each medium. The greatest potential concentrations for each congener were supplemented with additional “enrichment factor” increases to two bioaccumulative and toxic non-ortho congeners (PCB 126 and 169). As a result, the estimated TEQwquvian concentrations for waterfowl and soils have much greater toxicity than would be predicted from the original Aroclor mixtures. Waterfowl TEQWHOAWan concentrations were the highest among all prey type contributions to APDDTEQ by a factor of three andseven for the geometric mean and upper 95% CL concentrations, respectively (Table 4.3). 190 The only notable difference between estimates of APDD from a site-specific diet (APDDme'asmd) and literature-based diet (APDDpredicted) were higher APDDprcdictcd for total PCB (geometric mean and upper 95% CL values) at TB (mass-based diet), and higher APDDmeasmd for TEQWHomian (geometric mean and upper 95% CL) values at PC (nurneric- and mass-based diets). At TB, the greater APDDpcg values for the literature- based diet was due to the greater proportions of pheasants and to the elevated total PCB concentrations in the passerine/terrestrial avian prey group compared to total PCB concentrations in rabbits, the predominant prey group for TB owls. The difference in predicted versus measured APDDPCB was not observed at FC because the large differences in the proportion of passerine/terrestrial prey (e.g., pheasant) between the site- specific and literature-based diets was mitigated by the low total PCB concentrations in the dietary items collected at F C, a larger proportion of waterfowl in the FC APDDmeasmd vs. literature-based APDmedicted, and the greater waterfowl total PCB concentrations used for the APDDPCB calculations (Table 4.2, Table 4.3, Figure 4.5). At PC, the greater APDDTEQ for the site-specific diet originated from the overriding influence of waterfowl prey. This included a larger proportion of waterfowl in the FC APDDmeasured vs. literature-based APDDpredicted, and the much greater waterfowl TEQWHQAvian concentrations used for the APDDTEQ calculations (Table 4.2, Table 4.4, Figure 4.5). This difference in APDDTEQ was not present between APDDprcdgcted /APDDmeasu,cd at TB because the much larger proportion of passerine/terrestrial prey (e.g., pheasant) in the literature-based diet was mitigated by higher mean TEQ\;VH().A.,i,.,,n in rabbits vs. passerines/terrestrial avian prey coupled with additional TEQ contributed from muskrat 191 prey (with greater variable TEQ concentrations) to APDDmcasmd that was calculated from the upper 95%CL for TEQWH0-Avian in prey populations. Overall, calculations of APDmedictcd /APDDm¢asmd in this study were primarily influenced by gaps in the site-specific data for principal prey items in both the site- specific and literature-based diets. Although APDmedicwd and APDDmcasmd values were very similar across the range of prey concentration values at PC and TB, the notable differences observed in APDD can be traced to the lack of site-specific data for pheasants and rabbits, and the lack of recent, congener-specific total PCB data for waterfowl. Because all surrogate data used to address these data gaps was chosen to insure that any potential biases contributed by these data erred in a conservative “worst-case” manner, it is reasonable to assume that if site-specific data. were available for these prey, the calculated APDDPCB/APDDTEQ for both diets would have decreased and the relationships between APDDmeasmd and APDmedictcd for both total PCBs and TEanomian would have changed. This exercise also illustrates that differences between APDDmeasmd and APDmedictcd may be exacerbated at sites where the contaminant distribution between proximal aquatic and terrestrial habitats is dissimilar, and prey with links to the aquatic food-web figure predominantly in site-specific GHO diets. In these instances, the unique composition of a site-specific diet that includes aquatic prey may contribute significantly to the overall assessment of exposure, therefore posing significant risk that may be overlooked if the hazard assessment relies upon a literature-based dietary composition that fails to identify important prey items with links to aquatic exposures. 192 Risk Estimates Based on Total PCBS and TEQs Hazard quotients based on TEQWHO-Avian were greater than those based on total PCBS. Hazard quotients calculated from NOAEL TRVS for geometric mean and upper 95%CL concentrations of TEQWHQAvian were four- to 13-fold greater than for total PCBS at FC, and two- to three-fold greater at TB (Table 4.6). Hazard quotients based on total PCB concentrations are considered to be a more accurate estimate of potential risk because the concentration in the diet can be compared directly to values reported in the studies from which TRVS were derived. Congener-specific analyses provided for coplanar PCB congeners to be used in a calculation of TEQ. This approach eliminated the difficulties and uncertainties involved with assessing the toxicity of environmentally weathered PCB mixtures that are quantified as Aroclors, and is generally believed to correlate better with toxicity than measures of total PCBs (Blankenship and Giesy 2002). However, the scientific basis for TEQ derivation and use may contribute to bias that overestimates risk when they are applied to complex mixtures of PCBs. Concentrations of TEQs are calculated by multiplying each PCB congener by a class-specific (mammal, bird, fish) relative potency expressed as a toxic equivalency factor. Toxic equivalency factors are consensus values that were rounded up to be conservative estimates of risk (Van den Berg et al. 1998). This practice, coupled with the use of proxy values for congeners that were present at concentrations less than the method detection limit (e. g., use of one-half the method detection limit for non-detects), the summation of co-eluting congeners into a single value for some mono-ortho PCB congeners, and conservative estimates of congener fractional composition values for historical Aroclor PCB data are likely reasons that HQS based on TEanomm are greater than HQS estimated for total PCBs. 193 Additionally, a recent review of tree swallow exposure studies indicated that TEQSWHQ- Avian calculated from field-based TCDD and PCB exposures did not elicit similar endpoints of effect and may not be toxicologically equivalent (N eigh et al. 2006c). The Multiple-Lines—of-Evidence Approach This study has determined that dietary exposures of resident GHO populations to total PCBs and TEQWHGAvim present in contaminated floodplain soil of the KRSS are well below the threshold for effects on reproductive success. Even when the most conservative estimates of HQ are considered in the “bottom-up” assessment of potential hazards at the site, all HQ values for calculated APDDs using a site-specific dietary composition are less than 0.5. Similarly, all HQ values for calculated APDDs using a literature-based dietary composition at the site are less than 0.75 The greater HQ value for the literature-based diet originated fi'om overestimates of avian prey (fiom diet) and aquatic exposures (from COC concentrations used to calculate APDD fi'om passerine/terrestrial avian prey) in the hazard assessment. The “bottom up” assessment was one component of a multiple-lines-of—evidence approach that also included a tissue-based “top down” investigation of GHO exposure by investigating PCB concentrations in eggs and nestling plasma (Strause et a1. 2007). Results of the tissue-based studies were consistent with the dietary findings. The observed total PCB/TEQWHQAvim-l concentrations in eggs resulted in HQS less than 1.0 for all exposures indicating that tissue-based exposures did not pose a significant risk to GHO populations at the Upper KRSS. Hazard quotients calculated for tissue-based and site-specific dietary exposures show strong agreement at both the Reference and Upper 194 KRSS study sites with less than a three-fold difference between the ranges of HQNOAEc/NOAEL and HQLOAEc/LOAEL (mean and upper 95%CL concentrations) for both COCS, excepting TEQWHO-Avian concentrations at PC where a seven-fold range in HQs was present. (Figure 4.6). The multiple-lines-of-evidence approach included ancillary investigations to the “top-down” assessment that focused on evaluating the relative abundance, site use and productivity of resident GHOS at the Upper KRSS relative to the upstream Reference location. (Figure 4.7). The relative abundance of territory-holding nesting pairs of GHOS in the Upper KRSS was near the carrying capacity for the available habitat area included in the study. Nest acceptance rates and nest fidelity of actively breeding Upper KRSS GHOS across all nesting seasons included in the study were consistent with previous studies of artificial nest acceptance and habitat usage by Strigiforms in Midwestern forests (Holt 1996). Mean productivity rates (fledglings/active nest) were similar among locations where exposures to PCBs were much different, and were consistent with productivity measures for healthy Midwestern GHO populations (Strause et al. 2007; Holt 196). These results agree with findings for both the “top down” and “bottom up” approaches to evaluate chemical exposures at the site, and serve to reduce the uncertainties associated with assessment endpoints and strengthen the conclusion that potential risk to GHOS from exposures to total PCB/ TEstnoAvian in the Upper KRSS are unlikely to be sufficient to cause adverse population-level effects. Results from this study suggest that it would be appropriate to estimate risk based on either tissue-based or dietary-based methodologies. However, if a dietary-based 195 69.22 m» 5.8.5.05 anon. :23 a man Dam 333-2323: D 2:88.35 E 223:. 3393830 8583-95 umwmnézmmfi W 32.9.33: On. 5.3.5.03 80¢ :33 .H. E. | .H 5:01. .. O. Ffl4m<04 ~ 0m0~ 00803800 X3 00%: 080 8008 0808000. 05 00000980080 85 nomm .Aééémaomb 08010860 8xo€-&.o~80n%80200800.wK“m.m 08.0 Ammofic £80nt 000080030b3 00m 0.53 w0§p-Am2 000.0% 080060 0303030 00.02 000090020 080 690510 8038800800 000%«0 080300 0503030 0002 Aww0v 0003-033 mo 80088800, ma 0802000800 @806 000090008 080 $080500b mam: 0085800 8038800800 D92 00208 0&0 00m 505 80308800800 ~0>0~ 00803800 $3 00%: 080.8008 050800w 0% 0000380080 083 nomm .A§_><-o:30mfl.v 0803650 8xo€-&.0N=0£woSEomb00-wK,m.m 08.0 Ammomv 280893 00008803033 00.“ $55 00man-qm0_ 000mm0 080300 05.03030 08 0003-006 080 Aom0.5). Geometric mean predicted “egg-basis” concentrations at the KRAOC and Riverine Control were each significantly greater than concentrations at the Lacustrine Control (Kruskel-Wallace, p<.0.001 KRAOC; p<0.02 Riverine). T EQWHO—Avian concentrations Dioxin equivalent concentrations (TEanOMan) were calculated for both egg and plasma samples. Because congener specific conversion factors are not available to calculate an predicted “egg-basis” concentration from plasma, TEQWQAVjan concentrations in eagle tissues are presented separately for egg and plasma samples. Concentrations of TEQWHO- Avian in bald eagle nestling plasma analyzed across all three sampling locations were significantly correlated with concentrations of total PCBS in plasma (r =0.868. p<0.05). 226 Table 5.2. Geometric mean, ww (range), total PCBS in bald eagle (H. leucocephalus) egg and plasma samples from the Kalamazoo River Area of Concern (KRAOC), Riverine and Lacustrine sites. Total PCBs (egg) -1 ng PCBs g egg Total PCBs (plasma) -1 ‘ ng PCBS m1 plasma a Combined Total PCBS (egg-basis) -1 ng PCBs g egg Stud Site N N y (range) (range) b 44302 1 524 23055 4 KR‘ ‘OC (339 — 773) (12498 — 44302) Reference Riverinec 23943 1 212 9348 8 (152 — 291) (6047 — 23943) . 5516 e 38 1735 f Lacusmne 1 (8.8 — 114) (459 -— 4661) 9 a. PCB concentrations in nestling plasma samples converted to an “egg basis” using the Great Lakes bald eagle conversion factor: (loglo 11g PCBegg g'l, w) = (0.905 [loglo ng PCBplasma ml'l]-1.193). PCB concentrations include both addled eggs and converted nestling plasma samples. b. KRAOC sample site includes samples from the Ottawa Marsh and New Richmond bald eagle breeding areas. c. Riverine sample site includes samples from the Manistee River SGA, Manistee River Airport and Pere’ Marquette River bald eagle breeding areas. (1. Lacustrine sample site includes samples fiom the Lake St. Helen East and Backus Lake Flooding bald eagle breeding areas. e. Lacustrine egg sample from a Lake St. Helen East addled egg collected in 1994. f. Egg sample from 1994 is not included in combined egg-basis geometric mean concentration for the Lacustrine reference location. 227 t : i ...“° 5 2°: Z ""“°° :::: g ., :::; A ::: 2 A :::; Figure 5.2. Geometric mean (ww) total polychlorinated biphenyls (PCBs) and ZDDT concentrations in bald eagle (H. leucocephalus) eggs (combined addled egg and predicted egg-basis from nestling plasma) and TEanGmm in nestling plasma from the Lacustrine and Riverine Reference sites and the KRAOC, error bars show the 95%UCL. 228 Concentrations of TEQWHO—Avian in plasma displayed spatial trends that were identical to trends for total PCBS in plasma. KRAOC birds had the greatest concentrations of TEQWHOAvim and the least concentrations were in Lacustrine Control birds (Table 5.3). Statistical testing of spatial trends for TEQWHO—Avian concentrations in plasma matched results for total PCBS in predicted “egg-basis” samples described above. Geometric mean concentrations at the KRAOC and Riverine Control site differed by a factor of 1.8, but the margin was not statistically significant (AN OVA w/Tukey’s, p=0.75), and mean concentrations at the KRAOC and Riverine Control were each significantly greater than concentrations at the Lacustrine Control (AN OVA w/Tukey’s, p<0.02 KRAOC and Riverine). The distribution of TEQWHO-Avian concentrations in the three addled egg samples available for analyses differed from plasma TEQWHGAvian concentrations in that the Riverine Control egg (Pere’ Marquette) had a greater TEQWHO—Avian concentration than the KRAOC egg (Ottawa Marsh) (Table 5.4). Egg TEQWHoAvian concentrations at both sites were an order of magnitude greater than the concentration of the Lacustrine Control egg (Lake St. Helen). All four non-ortho-substituted PCBs or coplanar PCBS (IUPAC numbers 77,81,126,169) and five of the eight mono—ortho-substituted PCBS (IUPAC numbers 105,] 18,156,] 57,167) were regularly detected in egg and plasma samples from the KRAOC and Riverine study sites and the egg sample from the Lacustrine site. Lacustrine plasma samples displayed very low concentrations of both coplanar and mono-ortho-substituted PCB congeners. All c0planar PCB congeners were below the method detection limit in two of seven Lacustrine plasma samples. Coplanar PCB 77 was detected in five of seven plasma samples and in three of these samples it was the only non-ortho-substituted congener quantified above the method detection limit. 229 Table 5.3. Geometric mean, ww (range), total PCBS, TEQWHGAvian and relative potency concentrations in bald eagle (H. leucocephalus) plasma samples from the Kalamazoo River Area of Concern (KRAOC), Riverine, and Lacustrine sites. Total PCBS TEQWHO—Avian . a (plasma) (plasma) Relative Potency (plasma-basis) -l -1 -1 pg PCBS ml plasma pg TEQ ml plasma N pg g Study Site N (range) (range) (range) b 0.43 2 33 2 75.5 2 K1“ ‘OC (0.34 — 0.55) (28 — 38) (69.1 — 82.6) Reference Riverinec 9.21 5 18 5 87.4 5 (0.15 — 0.25) (13 — 32) (62.2 — 127) Lacustrine 0.04 7 3 7 70.7 7 . (0.01— 0.11) (0.3 — 10) (34.1 — 189) a. Relative potency = (pg TEanomm ml'l plasma) / (ug total PCBS ml'1 plasma). b. KRAOC sample site includes samples from the Pottawatorrri Marsh bald eagle breeding area. c. Riverine sample site includes samples from the Manistee River SGA, Manistee River Airport and Pere’ Marquette River bald eagle breeding areas. (1. Lacustrine sample site includes samples from the Lake St. Helen East and Backus Lake Flooding bald eagle breeding areas. 230 Table 5.4. Geometric mean, ww (range), total PCBS, TEQwuomm and relative potency concentrations in bald eagle (H. leucocephalus) egg samples from the Kalamazoo River Area of Concern (KRAOC), Riverine, and Lacustrine sites. . a Total PCBS TEQWHO'AW' Relative Potency (eggS) (egg) (egg_basis) PCB '1 I TE '1 N '1 Study Site Pg sg egg N pg Qg egg ngg N b KRAOC 44.3 1 1139 1 25.7 1 Reference Ri . c 23.9 1 1474 1 61.6 1 verrne Lacustrine 5.52 1 135 1 24.5 1 Relative potency = (pg TEQWQ-Avian g’l egg) / (1.1g total PCBs g'l egg). KRAOC sample site includes one addled egg sample from the Ottawa Marsh bald eagle breeding area. 0. Riverine sample site includes one addled egg sample from the Manistee River Airport breeding area. d. Lacustrine sample site includes one addled egg sample collected in 1994 from the Lake St. Helen East bald eagle breeding area. .e‘e 231 Two remaining plasma samples included PCB 77 and 126 present at levels above the method detection limit. Just four of the eight mono-ortho-substituted PCB congeners (PCBS 105,118,156,167) were regularly detected in Lacustrine plasma samples, PCB 157 was not present. Mono-ortho-substituted congeners 114, 123, and 189 also were not detected in any of the plasma samples from the three study sites. The relative proportion of TEanGMian contributed by the three coplanar PCBS having the greatest TEFWOAVtan values relative to other congeners was seen to vary between egg and plasma samples. For egg samples, PCB congener 126 contributed the greatest proportion toward total TEQWHO-Avian, ranging from 45.7% at the Lacustrine site to 69.9% at the Riverine site (Figure 5 .3). In plasma samples, PCB congener 77 contributed the greatest proportiOn toward total TEQWHO-Avian, ranging from 47% at the Riverine site to 80.9% at the KRAOC. Together, the relative proportion of TEQWHQAvm contributed by the four coplanar PCBS did not vary greatly between the three sampling sites or between egg and plasma samples, plasma concentrations had slightly greater proportional values, ranging from 90% (Riverine) to 95.2% (KRAOC). Coplanar PCBs accounted for 85.4% (Lacustrine) to 90% (Riverine) of total TEQWH0-Avian in egg samples (Figure 5.3). Relative potency The relative contributions of non-ortho- and mono-ortho-substituted congeners can be evaluated by standardizing the TEanOMm to the total PCB 232 A. 100% - \ one Q P0881 80% PC8126 KN (R 60% W ’////////// 40% ’///////////// 20% 0% Egg Plasma Egg Plasma Egg Plasma Lacustrine Riverine KRAOC 3- 100% ) 80% l E Mono-ortho 60% l I Coplma'r 40% 20% J 0% Egg Plasma Egg Plasma Egg Plasma Lacustrine Riverine KRAOC Figure 5.3. Percent contribution of polychlorinated biphenyl (PCB) coplanar and mono- ortho-substituted congeners to total TEQWHO—Avian in bald eagle egg and plasma samples at the Lacustrine, Riverine and Kalamazoo River Area of Concern (KRAOC) sites. Contribution of individual coplanar congeners to total TEQ and total coplanar verses total mono-ortho-substituted coplanar contribution to TEQ. 233 concentration to obtain a relative potency value (Froese et al., 1998). Relative potency values can be used to assess the degree of weathering and in evaluations of exposure and bioaccumulation between trophic levels of an impacted food web and resulting changes in toxic potency of the weathered mixture. Geometric mean relative potency values for TEanomm and total PCBS in bald eagle plasma are similar among the KRAOC and reference sites. The greateSt geometric mean concentration, (8.74 x101 11g g'l, ww) was observed for plasma samples collected at the Riverine site (Table 5.3). Relative potency values for the three addled egg samples available for the study sites were lower than plasma at each site, with the maximum concentration in eggs (6.16 x101 ug g'l, ww) also observed at the Riverine site (Table 5.4). EDDT concentrations/Eggshell measurements Total DDT was detected in all egg and plasma samples analyzed. p,p’-dichloro-diphenyl- dichloro-ethylene occurred at the greatest concentration of the measured DDT analytes and contributed from 92 to 99% and 85 to 100% of the ZDDT in all egg and plasma samples, respectively. Total DDT concentrations in eggs were greater from the Riverine site compared to samples from the KRAOC and Lacustrine site (Figure 5.2). The formula describing the egg-to-plasma relationship (conversion factor equation) for p,p’- DDE in Great Lakes bald eagles is expressed on a log-basis as “log p, p ’-DDEegg (ug g'l) = 0.676[log p,p’-DDEp.alsma (ng ml")] -0.578” (R2=0.324, p<0.001) (Strause et al, 2007a). Geometric mean predicted “egg-basis” total ZDDT concentrations at the Riverine site and the KRAOC differed by a factor of two, but the margin was not statistically significantly (Kruskel-Wallace, p>0.5). Geometric mean predicted “egg-basis” concentrations at the 234 Riverine site were significantly greater than concentrations at the Lacustrine site (Kruskel-Wallace, p<0.05). Predicted egg-basis EDDT concentrations in KRAOC birds were not significantly greater than the Lacustrine site (Kruskel-Wallace, p>0.5) (Figure 5.2). Egg shell thickness measurements from addled eggs and shell fragments recovered from active nests displayed a significant negative correlation with EDDT concentrations in the combined egg-basis data set (r= -0.65, p<0.05), and spatial trends for geometric mean egg shell thickness measurements (mm) were inverse to trends observed for geometric mean predicted “egg-basis” EDDT concentrations. Shell thickness was significantly less at the Riverine site compared to the Lacustrine site (AN OVA w/I‘ukeys, p<0.01), and shell thickness at the KRAOC was not significantly less then thickness at the Riverine Site, or significantly greater than thickness at the Lacustrine site (ANOVA w/Tukeys, p=0.17, Riverine; p=0.37, Lacustrine) (Figure 5.4). Productivity From 2000 to 2004, a large difference in productivity (fledglings/active nest) was observed between the KRAOC and the two reference sites. During the five reproductive seasons, there were 10 nesting attempts (active nests) in the KRAOC, and 11 and 10 nesting attempts at the Riverine and Lacustrine sites, respectively. The three KRAOC territories included in the study had greater frequencies of inactive breeding territories (5 inactive territories/ 15 possible) and failed nesting attempts (7 failures/ 10 possible) compared to the Riverine (4 inactive territories/ 15 possible; 4 failures/10 possible + one unknown outcome) and Lacustrine sites (0 inactive territories/10 possible; 3 failures/ 10 235 E ZDDT - Egg (measured) I ZDDT - Egg-basis (from plasma) El Eggshell Thickness g 0.68 g Pro-1947 ;: mean 0.66 8 eggshell N ..5 0.64 3 0 3° .62 o\ A 3.. E 0.60 :3 ii :.~ g g 0.58 g a 3 3 °\¢ 8 g 0.50 35 o c .2 A E 5 3 s e 9 0.54 E a .. 5 gig g 0.52 5 2’ 3 0.50 ‘9 Na 1 7 7 1 5 5 1 2 4 (p<0.05) (p<0.01) Lacustrine Riverine KRAOC Study Site Figure 5.4. Geometric mean (ww) EDDT concentrations in bald eagle (H. leucocephalus) eggs and the predicted egg-basis (from plasma) sample compared to egg shell thickness at the Lacustrine, Riverine and Kalamazoo River Area of Concern (KRAOC) sites, error bars show the 95%UCL. 236 possible). Four fledglings were produced at the KRAOC for an annual productivity rate of 0.4 fledglings per active nest. At the Riverine and Lacustrine sites, successful fledglings totaled 9 birds and 10 birds for annual productivity rates of 0.9 and 1.0, respectively (Table 5.5). Risk assessment Hazard quotient (HQ) values based on the geometric mean PCB concentration for the KRAOC combined egg-basis data set ranged from 1.2 for the LOAEC TRV to 7.7 for the NOAEC TRV. On an individual sample basis, two of four (50%) KRAOC egg-basis samples exceeded the LOAEC TRV and four of four (100%) samples exceeded the NOAEC TRV. Hazard quotients for the 95% confidence interval (95%CI) on the geometric mean concentration ranged from 0.7 — 1.9 (LOAEC) and 4.6 — 13.0 (NOAEC) (Figure 5.5). At the Riverine site, HQ values for the geometric mean PCB concentration in eggs ranged from 0.5 (HQLOAEC) to 3.1 (HQNOAEC), with one of eight (12.5%) and eight of eight (100%) samples exceeding the LOAEC and NOAEC TRVS, respectively. Hazard quotients for the 95%CI on the geometric mean concentration were < 1.0 for the LOAEC and ranged from 2.3 — 4.2 for the NOAEC. At the Lacustrine site, HQ values for the geometric mean PCB concentration in eggs ranged from 0.1 (HQLOAEC) to 0.6 (HQNOAEC), with zero and three of nine (33%) samples exceeding the LOAEC and NOAEC TRVS, respectively. Hazard quotients for the 95%CI on the geometric mean concentration were 51.0 for both the LOAEC and NOAEC. 23’? Table 5.5. Kalamazoo river study - bald eagle (H leucocephalus) productivity within the KRAOC, Lacustrine and Riverine reference locations. . 3 Annual Nest Site/Fledgling Success Study Location Total b 2000 2001 2002 2003 2004 Productivity KRAOC 0'4 ' 0 Swan Creek Highbanks (AN—02) d/F b/F e/F MI: W Ottawa Marsh (AN -03) c/ l c/F c/F I I Pottawatomi Marsh (AN-04) I I I b/l b/2 Riverine Reference 0.9 Manistee River SGA (MN-05) d/2 d/2 I f/F f/U Manistee River Airport (MN-11) I I a/2 a/F a/l Pere’ Marquette River (MS-04) a/l all a/F I d/F Lacustrine Reference 1.0 Backus Lake (RO-O4) b/ 1 b/F b/2 b/F b/2 Lake Saint Helen E. (RO-Ol) j/2 l/F m/2 j/F j/l a. United States Fish and Wildlife Service (U SF WS) bald eagle breeding area nest site descriptors/number of successful fledglings per nest: F=nest failure, no successful fledglings; I=nest site inactive (annual activity not included in calculations of total productivity); U=fledgling success unknown (annual activity not included in calculations of total productivity). b. Total productivity=total number of fledglings (all years, all occupied nests)/total number of occupied nests, all years)(Postupalsky, 1974). c. USFWS bald eagle breeding area designator. 238 Figure 5.5. Hazard quotients (HQ) for the effects of total polychlorinated biphenyls (PCBS) and ZDDT (combined addled egg and predicted egg-basis sample) and TEQWHO. Avian (egg samples only) for bald eagles (H leucocephalus) based on the no observable adverse effect concentration (N OAEC) and the lowest observable adverse effect concentrations (LOAEC). Each box encompasses the 95% CI about the geometric mean concentration. 239 Hazard Quotient (95% LCLIUCL) Hazard Quotient (95% LCLIUCL) Hazard Quotient (95% LCLIUCL) 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 - Lacustrine Reference HQs : E Total PCBs - - TEQ WHO- Avian - E] Total DDT L_ {1 HQflfl g L] _ — l:| NOAEC LOAEC Riverine Reference HQs E D - HQ=1.0 NOAEC LOAEC KRAOC HQs J] "Enruururjflgiisr. NOAEC LOAEC 240 Hazard quotient values for TEQWH0-Avian concentrations in eagle tissues were calculated for measured egg exposures only (sample size = 1 at each study site), because coplanar and mono-ortho-substituted congener—specific conversion factors are not available to calculate an egg-basis concentration from plasma measurements, and as noted earlier, there are no TRVS available for plasma TEQWHQAVian exposures/effects. At the KRAOC, HQ values for the Ottawa Marsh addled egg ranged from 2.9 (HQLOAEC ) to 8.4 (HQNOAEC). The Riverine site egg from Pere’ Marquette had HQ values ranging from 3.7 (HQLOAEC ) to 10.9 (HQNOAEC), and the Lacustrine site egg (Lake St. Helen—collected in 1994) had HQ values ranging from 0.3 (HQLOAEC ) to 1.0 (HQNOAEC). Hazard quotient values based on the geometric mean EDDT concentrations for the KRAOC combined egg-basis data set ranged from 0.5 for the LOAEC TRV to 1.8 for the NOAEC TRV. On an individual sample basis, zero samples exceeded the LOAEC TRV and three of three (100%) samples exceeded the NOAEC TRV. Hazard quotients for the 95%CI on the geometric mean concentration were <1.0 for the LOAEC and ranged from 1.1 — 2.9 for the NOAEC (Figure 5.5). At the Riverine site, HQ values for the geometric mean ZDDT concentration in eggs ranged from 1.0 (HQLOAEC) to 3.2 (HQNOAEC), with three of six (50%) and six of six (100%) samples exceeding the LOAEC and NOAEC TRVS, respectively. Hazard quotients for the 95%CI on the geometric mean concentration ranged from 0.8 — 1.1 (LOAEC) and 2.8 — 3.7 (N OAEC). At the Lacustrine site, HQ values for the geometric mean ZDDT concentration in eggs ranged from 0.3 (HQLOAEC) to 0.9 (HQNOAEC), with one of seven (14%) and three of seven (43%) samples exceeding the LOAEC and NOAEC TRVS, respectively. Hazard quotients for 241 the 95%CI on the geometric mean concentration were <1.0 for the LOAEC and ranged from 0.5 — 1.8 for the NOAEC. Geometric mean egg shell thickness at the KRAOC site was 4% less than the pre- 1947 benchmark for bald eagles and was not significantly different than mean thickness measurements recorded for the Lacustrine (ANOVA w/Tukeys, p=0.37) and Riverine (ANOVA w/Tukeys, p=0.17) sites. Mean egg shell thickness at the Riverine site was 11.8% less than the pre-l947 benchmark and was significantly less (AN OVA w/Tukeys, p<0.01) than mean thickness measurements at the Lacustrine site (+1.3% pro-1947 benchmark) (Table 5 .6). 242 3.2+ 1 NE fies 1 some _ 88.2 1S5 2+ s fed s 6% z ewe: assesses A3- .. 5.3 5.2 1 None 322 1 33 mi- m was e an: a $5 anagram G? .. 33 ago .. See as; 159 3.- a and m 28 z a: duo; Axe Z 88 Z wwo Tm HQQN we Z wwo Tm BOON m: 8% .355 emmofioa Emmanwwm wwm $2-65 seam eaeflfie flannm semen semen magnum “Bohm .motm sfibmsomq 63.35% was. 605 Hoodoo mo mode. 83m oowaEmRM Bu 8on wwwo «afieemseoosfi .mb 2mg 23 com 83? mmofioflu sonmmwo $3-8m 89a 83.83% Afiv 685m 28 $2225 scamwwo £0383 mmmgmmo £wa 8 gouebqoodoo HQQN mo Aswafiv e5 53 §on 0568006 .oh 2an 243 .803 3 000003300 $0 @038 gm 5303 am 008.3 8 303 33830080 HOD 800.3. .808 m30003 0380 33 w3000E 0084 938m 38 0.8m 530$ am 883 05 00303 33...” 03988 03.53084 . 880.8 “30003 03080 23 023m 08.03082 .000m 38 0.892 003m 00382 .<0m 003% 008382 03 00303 .0030 0338.6. 0303M 880.8 330003 0380 280. 988303 302 38 3005 83m 4882 83.300 05 00303 003m 033.8 003 85 .6063 eds aeaaeeév 8808.0 B arena: was earns 3266 a 05...: 801388830383 :28 $0 38 $0 00238 .8.“ 830303.803 80533 2030 0033300 83383000380 @383 308 30 00003003 38 wwo 00.5803 80303 03:80 389mg 0033300 .399 38 man 0.0309803 38 PDQ 00303 3380300000 HQQN 013.23 8350000.: 6.0800 3 038. 244 DISCUSSION Organochlorine exposures in Great Lakes and KRAOC bald eagles Great Lakes exposures Studies of bald eagle productivity and eagle exposures to environmental contaminants in the Great Lakes ecosystem during the 19703 linked observations of reduced productivity with elevated concentrations of organochlorine contaminants in eagle eggs (W iemeyer et - al., 1984). On the basis of productivity benchmarks that identified the average number of successful fledglings produced for each active nest as a measure of population health, investigators frequently observed sub-populations of resident Great Lakes eagles with fledgling production rates below 0.7 fledglings per active nest, a rate insufficient for population maintenance. Fledgling productivity rates of 0.7 to 1.0 fledgling/nest, and >1.0 fledgling/nest are associated with stable and healthy eagle populations, respectively (Sprunt et al., 1973; Wiemeyer et al., 1984). Concurrent environmental investigations of contaminant exposure in bald eagles recorded elevated concentrations of p, p-DDE (up to 3.0 x104 ng g'l, ww) and total PCB Aroclors (as great as 9.8 x 104 ng g'l, w) in eggs from active eagle nests in Michigan, Wisconsin and Ohio, with decreases in egg shell thickness of 16% compared to the pre-l947 benchmark (Wiemeyer et al., 1984). These observations provided for a weight-of—evidence consensus that environmental exposure from DDT use (e.g., manifested as p, p ’-DDE-induced reproductive failure) and perhaps PCBs were the primary causes of impaired reproduction in Great Lakes eagle populations. Restrictions on the manufacture and use of DDT mitigated the most severe effects of p, p ’-DDE exposures on eagle reproductive success, and as p, p ’-DDE 245 concentrations in eggs declined during the 19803 and early 19903, most regional eagle populations in the Great Lakes basin recovered to produce fledglings at rates that provided for population grthh (e.g., >l.0 fledgling/active nest) (Bowerman et al., ‘ 1998). Eagle monitoring activities over the last two decades verified that the Great Lakes population is growing, but productivity monitoring and environmental sampling indicated that eagles occupying territories along some Great Lakes shorelines (including Lake Michigan) were still exhibiting depressed productivity caused by exposures to p, p ’-DDE and PCBs. (Bowerman et al., 2003). Fbr Lake Michigan in particular, current and uncontrolled sources of PCBs are the suggested cause of localized impacts to resident environmental communities (Stratus, 1999; 2005a), and studies of the aquatic forage base indicate that aquatic biota continue to accumulate and translocate sufficient mass of both PCBs and ZDDT to create hazards to top consumers in the aquatic food chain including bald eagles that occupy territories along the Lake Michigan shoreline and further inland along tributaries accessible to Great Lakes anadromous fishes (U SEPA, 2004; Giesy et al., 1994b). KRAOC exposures Concentrations of total PCBs and ZDDT/p,p’-DDE in eagle eggs and nestling plasma samples at the KRAOC and the Lake Michigan Riverine site fall within the range of concentrations observed in eggs and plasma collected from these same locations by previous researchers during the 1990s (Bowerman, 1993; Best, 2002). KRAOC study contaminant concentrations in eggs and plasma also are comparable to observed PCB and ZDDT tissue burdens in eagle eggs and plasma collected from locations on the Pacific 246 Coast of North America with known sources of PCBs and DDT that originated from paper making and product manufacturing activities (Elliott and Harris, 2002). An assessment of contaminant trends in the Great Lakes basin showed no trends in changes in the concentrations of p,p ’-DDE or total PCBs in un-hatched eagle eggs through the mid-19905 (Bowerman et al., 1998), but more recent nestling plasma data indicated that between '1993 and 1999-2002, total PCB concentrations in plasma samples collected from inland nests have decreased by an order of magnitude and at Lake Michigan nests PCB concentrations have decreased by over 50% (Roe et al, 2004b). Total PCB and ZDDT trend monitoring data for the three collection sites included in the KRAOC study are available for addled eggs (Best, 2002), nestling blood plasma, and productivity (Michigan Wildlife Trend Monitoring annual monitoring reports - Roe et al., 2004a, 2004b, 2003; Summer et al., 2002). Additional productivity data for the Kalamazoo River eagle territories is available in Stratus (2005a). For comparisons to the KRAOC bald eagle data base, the Great Lakes egg and plasma trend monitoring data were grouped to provide for comparisons of like-exposures. To minimize spatial and temporal sources of variability from the historical data bases the data were grouped to generate descriptive statistics for samples from the Kalamazoo River, Lake Michigan lakeshore samples from the Manistee and Pere’ Marquette Rivers only (PM/MR Riverine sites), and inland Lower Peninsula samples from Roscommon County lacustrine habitats. Only plasma samples from 1999 - 2002 and eggs collected from 1996 - 2000 were included in the comparison data set. The KRAOC results were examined to determine if the contaminant concentrations at thethree study sites were consistent with the historical trend monitoring data for site-specific exposures and spatial contaminant distributions. 247 The historical trend data includesfive samples (two eggs, three plasma) from the Kalamazoo river, eight samples (one egg, seven plasma) from the PM/MR Riverine sites, and 18 samples (one. egg, 17 plasma) from Roscommon County Lacustrine sites. For comparison purposes, all samples were treated in the same manner as described previously for KRAOC study samples. Multiple egg samples fiom the same clutch were combined for an average concentration and individual plasma samples were converted to an egg-basis using a plasma-to-egg conversion factor. An additional adjustment of the Michigan Wildlife Trend Monitoring PCB plasma concentration data was required to convert thesum value for the 20 quantified PCB congeners to a total PCB equivalent (all resolvable PCB congeners). This conversion was completed using the relationship: M PCBs = 4.57( sum 20 PCB congenersmgg‘l w) + 0.98 developed for the Wildlife Trend Monitoring data set by Stratus (2005). Spatial distributions of total PCBs in the KRAOC study data are identical to the trends displayed by the combined egg-basis historical trend monitoring data, but differences are found in results of the statistical testing. For the trend monitoring samples, geometric mean PCB concentrations in egg-basis samples were the greatest in Kalamazoo river birds (2.5 x104 ng PCB g'l, ww), and the least PCB concentrations were measured in data for the Roscommon County sites (9.8 x102 ng PCB g'l, ww). Geometric mean total PCB concentrations at the Kalamazoo river and PM/MR Riverine sites (1.1 x104 ng PCB g'l, ww) differed by a factor of 2.3 (compared to 2.5 for KRAOC study samples), and the difference was statistically significant (Kruskel-Wallace, p<0.5). Geometric mean concentrations at the Kalamazoo river and PM/MR Riverine sites were each significantly greater than concentrations at the Roscommon County sites (Kruskel- 248 Wallace, p<0.001). Direct comparisons between the two data sets show that the trend monitoring and KRAOC study PCB data are not significantly different at any of the three locations (Kruskel-Wallace, p=1.0 KRAOC/Kalamazoo river comparison; p=0.34 Riverine sites; p=0.17 Lacustrine sites, respectively). Historical ZDDT concentrations in the KRAOC sample displayed spatial distributions that were inconsistent with the trend monitoring data. The trend monitoring data set showed slightly greater geometric mean egg-basis concentrations of ZDDT in samples from the Kalamazoo river (3.7 x103 ng EDDT g'l, ww) compared to the PM/MR Riverine sites (2.7 x103 ng ZDDT g4, 'ww) and much lesser concentrations were measured at the Roscommon County sites (7.7 x102 ng ZDDT g", ww). Statistical testing of spatial trends show that geometric mean “egg-basis” total EDDT concentrations at the Kalamazoo river and PM/MR Riverine sites were not statistically significant (Kruskel-Wallace, p=0.88), but both the Kalamazoo river and PM/MR Riverine site concentrations were significantly greater than concentrations at the Roscommon County sites (Kruskel-Wallace, p<0.005). Direct comparisons between the two data sets show that the historical trend monitoring geometric mean EDDT concentrations are 42% less at the Kalamazoo River site (not statistically significant, Kruskel-Wallace, p=0.3) and 76% less at both the Riverine and Roscommon County sties (statistically significant, Kruskel- Wallace, p<0.005). Productivity data for the years 1999 thru 2002 included monitoring results for “Anadromous” and “Inland Lower Peninsula” eagle populations. Anadromous eagle populations included all eagle territories on Michigan tributaries to the Great Lakes with access for anadromous fish up to the first dam on the River. Tributaries to lakes 249 Michigan, Superior, Huron, and Erie are included in this group. Productivity data for the Kalamazoo River eagle territories includes nesting success for the years 1990 through 2003. Productivity rates from the Michigan Wildlife Trend Monitoring database show four-year average Inland and Anadromous productivity rates of. 1.0 and ' 0.8 fledglings/active nest, respectively. These values are consistent with the productivity rates for the KRAOC study Lacustrine (1.0 fledge/nest) and Riverine (0.9 fledge/neSt) reference sites. The comparable four-year (1999-2002) productivity rate for Kalamazoo Bald eagles computed using only the 1999 data from Stratus (2005a) and observations made during this study (2000-2002) is 0.4 fledgling/active nest, a value identicalto the five-year (2000-2004) productivity rates identified in this study. Because productivity rates are typically examined on a five-year, basis (W iemeyer et al., 1984) a rate could be calculated for the five-years immediately preceding the KRAOC study (1995-1999). These data show four fledglings were produced after 10 nesting attempts, or 0.4 fledglings/active nest, and indicate no net change in productivity for this population over the last ten years. PCB congener profiles/T E QWHaAvm The relative proportions of coplanar PCB congeners present between egg and plasma samples collected for the KRAOC study are generally consistent with proportions observed in these tissues of resident eagles throughout North America. Similarities include the predominance of PCB126 as the maximum detected coplanar congener in egg samples, and the predominance of PCB77 as the maximum detected coplanar congener in plasma samples (Elliott and Harris, 2002). This is consistent with observations that 250 PCB77 and. 81 are more susceptible to metabolism than PCB126 and 169 (Smith et al., 1990) and thus less likely to be. available in stored lipids that might be translocated to yolk lipids during oogenesis. ‘Relative contributions between PCB77 and 'PCB1‘26 to total TEQWHQAvian in . egg' samples were consistent for KRAOC and Pacific Coast ' (Canada) data sets, but KRAOC samples differed from the data set compiled by Elliott and Harris (2002) regarding the relative contributions from PCB77 and PCB126 to total TEQWHO-Avian in . plasma PCB77 predominates as the greatest relative contributor to total TEQ for plasma samples from each of the three KRAOC study sites in contrast to trends for the 10 plasma sample groupsfrom the Pacific Coast of Canada which show PCB126 as the greatest contributor to total TEQ for the PCB group. Comparable TEQWHGAvm concentrations for eagle egg or plasma samples fi'om the Great Lakes basin are available for only a single Lake Huron addled egg collected in 1986 (Schwartz, et al. 1993). This egg contained a deformed embryo and had a total PCB concentration of 1.0 x105 ug PCB g'l, ww and a TEQWHQAyim-l concentration of 9.8 x103 pg TEanaAvian g'l, ww , nearly 7-fold greater than the maximum KRAOC concentration of TEQWHOAvian in eggs (1.5 x103 pg TEQWHO-Avian g", ww, Riverine site - Pere’ Marquette River). The available data for PCB-based TEanGMan concentrations in eagle tissues show that Great Lakes eagles carry the greatest PCB TEQWHO-Avian burdens of all North American eagle populations studied to date. Eagle egg samples from the KRAOC contained PCB contributions to TEQWHGMan that were 2- to 5-fold higher than comparable PCB contributions to TEQWHOMan in eagle eggs from New Jersey tributaries to the Delaware Bay (Clark et al., 1998), the Columbia River (Buck et al., 2005) and the Pacific Coast of Canada (Elliott et a1. 1996). Similarly, nestling plasma 251 samples from the' KRAOC contained PCB contributions to TEQWHOAvian that were 10- fold. higher than comparable PCB contributions toTEngoAvim in nestling plasma from the Pacific coast of Canada (Gill and Elliott, 2002; Elliott and Norstrom, 1998). - Calculated TEQwaAvim concentrations in eggs and plasma were of limited value to this examination of exposures .and potential, risks to bald eagles and were used ' sparingly to‘ reinforce conclusions drawn from the total PCB, EDDT, and productivity data. Major contributing factors to this condition include the difficulty of collecting an adequate number of egg samples from a protected avian species, the lack of congener- specific conversion factors for converting plasma sample TEQ concentrations to an egg- basis, and the absence of a suitable plasma-based TEanGMian TRV. Hazard assessment Results of the focused KRAOC eagle investigations are consistent with previous Great Lakes basin-wide monitoring efforts for the bald eagle undertaken by the USFWS and the focused wildlife contaminant trend monitoring by the Michigan Department of Environmental Quality (MDEQ). All three of these monitoring efforts have produced contaminant data and productivity measurements that can be interpreted as strong evidence that Kalamazoo River eagle populations are experiencing impaired reproductive success and On a broader scale, indicate an ecosystem with decreased fitness. A close examination of the KRAOC eagle data shows productivity rates (0.4 fledglings/active nest) below the minimum level (0.7 fledglings/active nest) associated with a stable population'(Sprunt et al., 1973). The KRAOC data show total PCB HQLOAEC ranges for the 95%CI‘ on the geometric mean egg-basis concentration were from 0.7 to 1.9, and the 252 95%CI HQNOAEC ranged from 4.6 to 13. The geometric mean HQLOAEC and HQNOAEC values 'were 1.2' and 7.7, respectively. The HQNOAEC value for TEQWHQAvian concentrations in the single egg. sample from the KRAOC was 8.4. The hazard assessment for PCBs indicated that. the degree of exposure to PCBs was at or above the threshold for effects on reproduction in bald eagles. This assessment is supported by the EDDT data for the KRAOC and Riverine sites. The greatest ZDDT concentrations observed for this study were observed at the Riverine site. Together, the Riverine site combined egg basis sample set displayed individual ZDDT concentrations that exceeded the egg-based TRV for the NOAEC in 100% of the samples. The geometric mean HQLOAEC and HQNOAEC values were 1.0 and 3.2, respectively and the 95%CI HQNOAEC ranged from 2.8 to 3.7. Geometric mean egg shell thickness from two addled egg shells and 3 egg shell fragments showed a decrease of over 11% compared to the pre-1947 benchmark. Two of five egg shell samples (40%) showed shell thinning in excess of the -15% value generally expressed as a conservative benchmark of reproductive failure in raptors (Newton, 1979), and both nests failed the year these samples were collected. Despite these elevated ZDDT exposures, productivity rates at the Riverine site (0.9 fledglings/active nest) were above maintenance levels and close to values indicative of a healthy population. Exposures to ZDDT for KRAOC birds were roughly half the levels observed at the Riverine site, and compared to the much greater PCB exposures, effects related to XDDT can be expected to be negligible. 253 Risk Assessment Uncertainties Riverine exposures to ZDDT provided a clear illustration of the fact that the true effect leVel for individuals’ lies somewhere between the NOAEC and LOAEC, and that population effects are seldom observed when HQ values exceed the 1.0 threshold by small margins. Even conservatively, population effects have not been expected at HQS of 10.0 (Neigh et al., 2006). The uncertainties associated with HQ estimates between 1.0 and 10.0 supported using a multiple lines of assessment and weight of evidence approach to assess population health. Taken together, contaminants data in tissues and direct. observation of productivity provided a compelling argument for population level effects at the KRAOC study site. As part of the weight-of-evidence approach, additional factors that have been observed to influence productivity outcomes at individual eagle nest sites should be considered in the ecological risk assessment. Correlations between contaminant exposure and lower nest productivity can be confounded by other correlated factors, anthropogenic or natural, that an investigator fails to recognize when a study is completed (Karasov and Meyer, 2000). One of these factors is human disturbance to the nest site during critical time periods of eagle courtship and egg incubation (USFWS, 1983), or later in the reproductive cycle, when disturbances to foraging adults can adversely impact foraging success and ability to provide sufficient food to nestling birds. Human activities that have been shown to greatly disturb eagles include residential development and associated daily activities, hiking, and boating (fishing). Studies of disturbances to nesting and foraging eagles indicated that individual eagles vary in their tolerance to human encroachment. Some birds become habituated to human activities and are extremely 254 tolerant of incursions into their territories, and others are quickly disturbed to the extent . . they leave the nest or abandon all activity in the zone of disturbance during all or part of the day (e.g., foraging activity). Studies of foot traffic in proximity to actively nesting eagles have indicated that birds can be flushed from the nest from mean distances of over 400 m (Fraser et al., 1985). Studies of boating activity under similar conditions of eagle activity showed that the distance (from the nest) and duration of a boating activity are factors that cause eagles to be flushed from the nest, and the authors recommended buffer zones that included critical minimum distances of 100 m to pass and 400 m to stop (Grubb et al., 2002). The lower reach of the Kalamazoo River where eagles are currently nesting is a popular destination for sportspersons and outdoor recreationists. Overland foot access is available to critical buffer areas (e.g., g 400 m) surrounding most of the actively maintained nests in each of the three eagle territories in the KRAOC, via residential property, public hiking trails, or through illegal trespass in restricted areas. There are presently few institutional controls available to restrict overland access to nest sites in these areas. A recent survey of recreational fishing activity in the lower river downstream of Calkins Dam estimated that there were a total of 19,416 to 20,193 fishing days by Kalamazoo River anglers in 2001 (Stratus, 2005b). These figures did not include winter fishing, and most of the fishing activity focused on salmon, trout, walleye and bass. Trout fishing is concentrated in the early spring (March/April) When substantial runs of steelhead trout make their way into the river fi'om Lake Michigan and run to the Calkins dam. Walleye also make spawning runs up the river and the popular opening day/week of the walleye season is in late April. Both of these periods of high fishing 255 activity coincide with the critical period for eagle courtship and egg incubation in the KRAOC which typically extends from mid-February through early May. ' Trout and walleye 'can be fished with efficiency from the shore and from a boat. There are public .. ‘ launch facilities in the immediate vicinity of each of the three established eagle territories in the KRAOC, at calkins Dam (2 'km‘upstream of Swan Creek), at the M-89 bridge (adjacent to Swan Creek and 7 km upstream of Ottawa Marsh), and at the New Richmond and 63rd Street ramps (5 km upstream and 2 km downstream, respectively of Pottawatomi Marsh). Most of the actiVely maintained eagle nests are fairly isolated from the potential effects of boat traffic on the river, however the cumulative effects of heavy fishing pressure during the early spring critical period have not been investigated with regard to potential adverse effects on reproductive success at the KRAOC. Prey delivery rates and nestling energy intake are a principal determinant for successful fledgling production. Previous studies of eagle productivity and contaminant exposure in the Great Lakes basin also have identified the overriding influence of prey availability and the necessity to also consider energy intake by nestlings to avoid erroneous assignment of causation for observations of depressed productivity (Dykstra et al., 1998). Delivery rates are influenced by prey availability, human disturbances to foraging activities, and perhaps contaminant exposures. Nest observations to record prey delivery rates and adult attentiveness can provide an indication of how energy provision to nestlings affects productivity. Nest attentiveness data are available for nests included in the KRAOC eagle study, and these data are discussed in a companion assessment of potential risks through dietary exposure for KRAOC eagles (Strause et al., 2007b). 256 Predation also can adversely affect nest productivity. Nestling eagles are subject to predation by great horned owls and nestlings that fall from the nest and land unharmed at the base of the nest tree can succumbtto a variety of mammalian predators. Aside from constant surveillance of the nest site (e.g., continuous video surveillance) accurate estimates of predation impacts to eagle productivity are unavailable. Predation may certainly have played a role .in depressing eagle productivity at the KRAOC. Great horned owls nest in close proximity to active eagle nests and at the KRAOC site have been observed to nest in existing eagle nests. The potential for owl predation on eagle nestlings was present throughout the duration of the study. At the 2001 Swan Creek nest, the predated remains of a single' 7 to 9 wk old nestling were found at the base of the nest tree during mobilization to access the nest and band said nestling. Although all indicators suggest that rough weather caused the nestling to fall from the nest, there is no way to know if the young bird was severely injured or expired from the fall, or was preyed upon while healthy and active at the base of the nest tree. Nestling falls from the nest are a fairly common event, and as long as the nestling is visible, adults will continue to tend to grounded birds until they reach the fledgling stage. CONCLUSIONS The “top—down assessment of potential hazards to resident bald eagles at the KRAOC, and the Riverine and Lacustrine reference sites has employed an unbiased and conservative investigation of “worst-case” exposures to environmental contaminants. Sampling of nestling plasma is recognized as the most reliable medium for sensitive 257 measurements of localized contamination at the point of collection, much more so than are concentrations of residues in eggs or adult‘plasma samples (Donaldson et al., 1999; Olsson 'et al., 2000; Bowerman et al., 2000). Investigative methods included state-of-the- art analytical techniques with multiple levels of quality assurance/quality control safeguards for identification and quantification of PCB congeners and ZDDT analytes, multiple lines to estimate exposure to PCBs, and assessment of potentially confounding chemical stressors. Use of a Great Lakes derived plasma to egg conversion factor with conservative consensus bencMarks of toxicity furthers a technique for evaluating hazard using plasma data, and has broad recognition in the literature (Stratus, 2005 ; Elliott and Harris, 2002). Additionally, all field sampling activities were completed with the assistance of USFWS oversight personnel and split samples of all tissue samples were provided to USFWS scientists to encourage reproducibility and confidence in the analytical data. The KRAOC eagle study demonstrated the utility that bald eagles provide as a sentinel species for site-specific baseline ecological risk assessments employing a multiple-lines-of—evidence approach that includes using “top-down” methodology to combine measured residues of COCS in tissues and counts of productivity. Additional opportunities for focused monitoring using bald eagles are possible. In certain instances eagles will select artificial nesting platforms that are properly constructed and sited (Grubb, 1995). With eagle populations expanding in most North American habitats the competition for optimal nesting sites will increase. Lesser availability of optimal nesting sites and increased tolerance of proximal human activity are two contributory trends that 258 may further facilitate the use of eagles as biological monitors of long term ecosystem health at areas of concern in the Great Lakes and throughout North America. ' Acknowledgement William Bowerman, Clemson Institute of Environmental Toxicology (CIET) provided training and logistical support. CIET graduate students Amy Roe, Carrie Tansy, Faith Wiley, and Kate Parmentier aided in sample collection and processing. Support fi'om the Michigan Department of Natural Resources included nesting activity survey reports (Jerry Weinrich) and lodging for field crews (Dan Moran; Porter Ranch). Jack Holt and Sergej Postupalski provided valuable advice. Climbing assistance was provided by Jack Holt, Brad Richardson, William Bowerman, Dave Best, Charles Mehne, and Craig Thompson. Tameka Dandridge, Pamela Moseley and Lisa Williams assisted with blood draws fiom nestling birds. Michelle Moffat recorded photographic and video documentation of sampling activities. We thank the generous landowners who provided access to nests and warm hospitality including Ken and Janet Petoskey, Mark and Ann Mavis, Robert Adamczak, Bill and Terry House and the Manistee Airport, and Thom Becker and the accommodating members of the St. Helen Hunt Club. Funding for this study was provided by the Kalamazoo River Study Group. 259 REFERENCES Anderson, B. W. ,Hickey, J. J., 1972. Eggshell changes in certain North American birds. Proceedings of the International Ornithological Congress, The Hague, The Neatherlands. August 30- 5 September, 1970. 15: 514- 540. Bartolotti, GR, 1984. Criteria for determining age and sex of nestling bald eagles. J Field Ornithology 55:467-481- . ' BBL, 1993. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site description of the current situation. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. BBL, 1994a. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site Remedial Investigation/Feasibility Study. Technical memorandum 2: Results of phase I TBSA soil sampling. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. BBL, 1994b. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site Remedial Investigation/Feasibility Study. Technical memorandum 3: Results of the floodplain soils investigation, volume 1 or 2. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. BBL, 1994c. Allied Paper, Inc./Portage Creek/Kalamazoo River Superfund Site Remedial Investigation/Feasibility Study. Draft technical memorandum 12: Former impoundment sediment geochronological plot study, volume 14. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. BBL, 2000. Allied Paper, Inc/Portage Creek/Kalamazoo River Superfund Site Remedial Investigation Report - Phase I. Blasland, Bouck & Lee, Incorporated, Syracuse, NY, USA. Best, D.A., 2002. Monitoring of bald eagle exposure to organochlorine compounds in the Great Lakes via addled eggs, 1986-2000. Environmental Contaminants Program, United States Fish and Wildlife Service, East Lansing Field Office, MI, USA. Blankenship, A.L., Zwiemik, M.J., Coady, K.K., Kay, D.P., Newsted, J .L., Strause, K., Park, C., Bradley, P.W., Neigh, A.M., Millsap, S.D., Jones, P.D., Giesy, J .P., 2005. Differential accumulation of PCB congeners in the terrestrial food web of the Kalamazoo River superfund site, Michigan, USA. Envrion Sci Technol 39:5954- 5963. Bowerman, W.W., 1993. Regulation of bald eagle (Haliaeetus leucocephalus) productivity in the Great Lakes Basin: An ecological and toxicological approach. Ph.D. Dissertation, Michigan State University, East Lansing, MI, USA. 260 Bowerman, W.W., Best, D.A., Grubb, T.G., Zimmerman, G.M., Giesy, J .P., 1998. Trends of contaminants and effects in bald eagles of the Great Lakes basin. Environ Monit Asses. 53:197-212. ' Bowerman, W.W., Best, D.A., Grubb, T.G., Sikarskie, J.G., Giesy, J .P., 2000. Assessment of environmental endocrine disruptors in bald eagles of the Great Lakes. Chemsophere, 41 : 1 569-1 574. Bowerman, W.W., Best, D.A., Giesy, J .P., Shieldcastle, M.C., Meyer M.W., Postupalsky, S.J., Sikarskie, J .G., 2003. Associations between regional differences in polychlorinated biphenyls and dichlorodiphenyldichloroethylene in blood of nestling bald eagles and reproductive productivity. Environ Toxicol Chem 22:371-376. Buck, J .A., Anthony, R.G., Schuler, C.A., Isaacs, F.B., Tillitt, D.E., 2005. Changes in productivity and contaminants in bald eagles nesting along the lower Columbia river, USA. Environ Toxicol Chem 24: 1779-1792. Camp, Dresser & McKee, 2003. Baseline Ecological Risk Assessment, Allied Paper, Inc/Portage Creek/Kalamazoo River Superfirnd Site. Final (Revised). Prepared for the Michigan Department of Environmental Quality Remediation and Redevelopment Division. Clark, K.E., Niles, L.J., Stansley, W., 1998. Environmental contaminants associated with reproductive failure in bald eagle (Haliaeetus leucocephalus) eggs in New Jersey. Bull Environ Contam Toxicol 61 :247-254. Council on Environmental Quality, 1972. Development of a continuing program to provide indicators and indices of wildlife and the natural environment. Final Report. Ecology Program, Smithsonian Institution, Washington, DC. Donaldson, G.M., Shutt, J .L., Hunter, P., 1999. Organochlorine contamination in bald eagle eggs and nestlings from the Canadian Great Lakes. Arch Environ Contam Toxicol 36:70-80. Dykstra, C.R., Meyer, M.W., Wameke, D.K., Karasov, W.H., Andersen, D.E., Bowerman, W.W., Giesy, J .P., 1998. Low reproductive rates of Lake Superior bald eagles: Low food delivery rates or environmental contaminants? J G Lakes Res. 24:32-44. Elliott, J .E., Norstrom, R.J., Lorenzen, A., Hart, L.E., Philibert, H., Kennedy, S.W., Stegeman, J .J ., Bullward, G.D., Cheng, K.M., 1996a. Biological effects of polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyls in bald eagle (Haliaeetus leucocephalus) chicks. Environ Toxicol Chem 15:782-793. Elliott, J .E., Norstrom, RJ., Smith, G.E.J., 1996b. Patterns, trends, and toxicological significance of chlorinated hydrocarbon and mercury contaminants in bald eagle eggs 261 from the Pacific Coast of Canada, 1990-1994. Arch Environ Contam Toxicol 31 :354- 367. Elliott, J.E., Norstrom, RJ., 1998. Chlorinated hydrocarbon contaminants and productivity of bald eagle populations on the Pacific Coast of Canada. Environ Toxicol Chem 17:1142-1153. Elliott, J .E., Wilson, L.K., Henny, C.J., Trudeau, S., Leighton, F.A., Kennedy, S.W., Cheng, KM. 2000. Biological effects of chlorinated hydrocarbons in osprey chicks. Environmental Toxicology and Chemistry 20:866-879. Elliott, J .E., Harris, M.L., 2002. An ecotoxicological assessment of chlorinated hydrocarbon effects on bald eagle populations. Reviews in Toxicology 4: 1-60. Fairbrothe, R.A., 2003. Lines of evidence in wildlife risk assessments. Hum Ecol Risk Assess 9:1475-91. Fraser, J .D., Frenzel, L.D., Matlrisen, J .E., 1985. The impact of human activities on breeding bald eagles in north-central Minnesota. J Wildl Manage 49:585-592. Froese, K.L., Verbrugge, D.A., Ankley, G.T., Nieme, G.J., Larsen, C.P., Giesy, J .P. 1998. Bioaccumulation of polychlorinated biphenyls from sediments to aquatic insects and tree swallow eggs and nestlings in Saginaw Bay, Michigan, USA. Environmental Toxicology and Chemistry 17:484-492. Giesy, J .P., Ludwig, J .P., Tillitt, D.E., 1994a. Deformities in birds of the Great Lakes region: Assigning causality. Environmental Science and Technology 28:128A-135A. Giesy, J .P., Verbrugge, D.A., Othoudt, R.A., Bowerman, W.W., Mora, M.A., Jones, P.D., Newsted, J .L., Vandervoort, C., Heaton, S.N., Aulerich, RJ., Bursian, S.J., Ludwig, J .P., Lugwig, M., Dawson, G.A., Kubiak, T.J., Best, D.A., Tillit, D.E., 1994b. Contaminants in fishes from Great Lakes-influenced sections and above dams of three Michigan rivers. 1: Concentrations of organic chlorine insecticides, polychlorinated biphenyls, dioxin equivalents, and mercury. Arch Environ Contarn Toxicol 27:202-212. Gill, C.E., Elliott, J .E., 2003. Influence of food supply and chlorinated hydrocarbon contaminants on breeding success of bald eagles. Ecotoxicology 12:95-11 1. Grubb, T.G., 1995. Constructing bald eagle nests with natural materials. Rsearch Note RM-RN-535. US. Department of Agriculture, Forest Service. Flagstaff, AZ, USA. Grubb, T.G., Robinson, W.L., Bowerman, W.W., 2002. Effects of watercraft on bald . eagles nesting in Voyageurs National Park, Minnesota. Wildl Soc Bull 30:156-161. 262 Hollamby, S., Afema-Azikuru, J ., Waigo, S., Cameron K., Gandolf, AR, Norris, A., Skiarskie, J .G., Suggested guidelines for use of avian species as biomonitors. Environ Monit Asses 118:13-20. International Joint Commission, 1991 . Report of the ecological committee’s biological , effects subcommittee to the Great Lakes science advisory board. Proceedings, Expert Consultation Meeting on Bald Eagles, Windsor, Ont., Canada. February 12-13, 1990. p 33. - ' Karasov, W.H., Meyer, M.W., 2000. Testing the role of contaminants in depreSsing avian numbers. Revista Chilena de Historia Natural 73 :461-471. Kay, D.P., Blankenship, A.L., Coady, K.K., Neigh, A.M., Zwiemik, M.J., Millsap, S.D., Strause, K., Park, C., Bradley, P., Newsted, J .L., Jones, .PD., Giesy, J .P., 2005. ' Differential accumulation of polychlorinated biphenyl congeners in the aquatic food web at the Kalamazoo River superfund site, Michigan. Environ Sci Technol 39:5964- 5974. Mauro, L., 1987. Raptor Physiology — Hematology and Blood Chemistry. In Raptor Management Techniques Manual. Edited by B.G., Pendleton, B.A., Millsap, K.W., Kline, and D.A., Bird. Natl. Wildl. Fed. Sci. Tech. Ser. No. 10 National Wildlife Federation, Washington DC. pp.269-275. MDEQ, 1997. A Strategic Environmental Quality Monitoring Program for Michigan’s Surface Waters. Michigan Department of Environmental Quality. Staff Report MI/DEQ/SWQ-96/152. January 1997. Neigh, A.M., Zwiemik, M.J., Bradley, P.W., Kay, D.P., Park, C.S., Jones, P.D., Newsted, J .L., Blankenship, A.L., Giesy, J .P., 2006. Tree Swallow (T achycineta bicolor) exposure to polychlorinated biphenyls at the Kalamazoo River Superfund Site, Michigan, USA. Environmental Toxicology and Chemistry 25:428-437. Newton, 1., 1979. Population Ecology of Raptors. T. & A.D. Poyser Ltd. London, UK. Nisbet, I.J.C., Risebrough, RW., 1994. Relationship of DDE to productivity of bald eagles Haliaeetus leucocephalus in California and Arizona, USA. In Mayberg B-U, chancellor RD, eds, Raptor conservation Today, world working group on birds of prey, The Pica Press, Berlin, Germany, pp.771-77 3. Olsson, A., Ceder, K., Bergman, A., Helander, B., 2000. Nestling blood of white-tailed sea eagle (Haliaeetus albicilla) as an indicator of territorial exposures to organohalogen compounds-an evaluation. Environ Sci Technol 34:2733-2740. Postupalsky, S., 1974. Raptor reproductive success: Some problems with methods, criteria, and terminology. In Hamerstrom. F.N., Harrell, B.E., Ohlendorff, RR. 263 (Eds). Management of Raptors. Proceedings, Raptor conservations Technologies, . Fort Collins, CO, USA, March 22-24, Raptor. Report 2, pp. 21-31. Roe, A., Bowerman, W., Bush, D., Sikarskie, J ., 2003. Michigan Wildlife Trend Monitoring: Year 2000 Annual Report: Nestling Bald Eagles, MI/DEQ/WD-03/088. Michigan Department of Environmental Quality, Lansing, MI, USA. Roe, A., Birrenkoff, A., Bowerman, W., Bush, D., Sikarskie, J ., 2004a. Michigan Wildlife Trend Monitoring. Year 2001 Annual Report: Nestling Bald Eagles, MI/DEQ/WD- 04/012. Michigan Department of Environmental Quality, Lansing, MI, ’ USA. Roe, A., Birrenkoff, A., Bowerman, W., Bush, D., Sikarskie, J ., 2004b. Michigan Wildlife Trend Monitoring. Year 2002 Annual Report: Nestling Bald Eagles, MI/DEQ/WD- 04/024. Michigan Department of Environmental Quality, Lansing, MI, USA. Ratcliffe, D. A. 1968. Changes attributable to pesticides 1n egg breakage frequency and eggshell thickness 1n some British birds. J Applied Ecol 7: 67-105. Schwartz, T.R., Tillitt, D.E., Feltz, K.P., Peterman, RH, 1993. Determination of mono- and non-0,0 ’-chlorine substituted polychlorinated biphenyls in aroclors and environmental samples. Chemosphere 26:1443-1460. Sprunt, A., W., Robertson, W.B., Jr., Postupalsky, S., Hensel, R.J., Knoder, C.E., Ligas, RJ., 1973. Comparative productivity of six bald eagle populations. Trans N Am Wildl Nat Resour Conf 38296-106. Stickel, J ., Wiemeyer, S.N., Blus, LJ., 1973. Pesticide residues in eggs of wild birds: Adjustment for loss of moisture and lipid. Bull Environ Contarn Toxicol 9:193-196. Stratus, 1999. Injuries to avian resources, Lower Fox River/Green Bay natural resource damage assessment. Final Report. Startus Consulting Inc., Boulder, CO, USA. Stratus, 2005a. Stage I Assessment Report, Volume I — Injury Assessment: Kalamazoo River Environment . Final Report. Startus Consulting Inc., Boulder, CO, USA. Stratus, 2005b. Stage I.Assessment Report, Volume II - Economic Assessment: Kalamazoo River Environment . Final Report. Startus Consulting Inc., Boulder, CO, USA. Summer, C.L., Bush, D., Bowerman, W., 2002. Michigan Wildlife Trend Monitoring: Year 1999 Annual Report: Nestling Bald Eagles, MI/DEQ/WD-02/023. Michigan Department of Environmental Quality, Lansing, MI, USA. 264 Strause, K.D., Zwiemik, M.J., Im, S.H., Newsted, J .L., Kay, D.P., Bradley, P.W., Blankenship, A.L., Wiliarns, L., Giesy, J .P. 2007a. Plasma to egg conversion. factor for evaluating PCB and DDT exposures in great horned owls and bald eagles. Environ Toxicol Chem (submitted). Strause, K.D., Zwiemik, M.J., Sikarskie, J .G., 1m, S.H., Hollamby, S., Bradley, P.W., Blankenship, A.L., Park, C.S., Millsap, S.D., Neigh, A., Kay, D.P., Newsted, J .L., Giesy, J .P., 2007b. Comparisong of risk assessment methodologies for exposure of bald eagles (Haliaeetus leucocephalus) to PCBs on the Kalamazoo River. Integr Environ Asses Manag (submitted). US. Environmental Protection Agency, 1976. PCBs in the United States industrial use and environmental distribution: Task 1. EPA 560/6-76/005. Washington, DC. US. Environmental Protection Agency, 2004. Results of the Lake Michigan mass balance study: Polychlorinated biphenyls and trans-nonachlor data report. EPA 905 R-01-011. Chicago, IL, USA. US. Fish and Wildlife Service, 1983. Northern States Bald Eagle Recovery Plan. Northern States Bald Eagle Recovery Team, US. Department of the Interior, Fish and Wildlife Service, Washington, DC. ' . Van den Berg, M., Bimbaum, L., Bosveld, A.T., Brunstrom, B., Cook, P., Freeley, M., Giesy, J .P., Hanberg, A., Hasegawa, R., Kennedy, S.W., Kubiak, T., Larsen, J .C., Rolaf van Leeuwen, F.X., Djien Liem, A.K., Nolt, C., Peterson, R.E., Poellinger, L., Safe, S., Schrenk, D., Tillett, D., Tyskling, M., Younes, M., Waem, F., Zacharewski, T., 1998. Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environ Health Perspect 106:775-792. Wiemeyer, S.N., Lamont, T.G., Bunck, C.M., Sindelar, C.R., Gramlich, F.J., Fraser, J .D., Byrd, M.A., 1984. Oranochlorine pesticide, polychlorobiphenyl, and mercury residues in bald eagle eggs — 1969-79 — and their relationships to shell thinning and reproduction. Arch Environ Contam Toxicol 13:529-549. Wiemeyer, S.N., Bunck, C.M., Stafford, CJ., 1993. Environmental contaminants in bald eagle eggs -l980-84-and firrther interpretations of relationships to productivity and shell thickness. Arch of Environ Contam and Toxicol 24:213-227. Woodford, J .E., Karasov, W.H., Meyer, M.W., Chambers, L. 1998. Impact of 2,3,7,8- TCDD exposure on survival, growth and behavior of ospreys breeding in Wisconsin, USA. Environ Toxicol and Chem 17: 1323-1331. 265