Monitoring and assessment of population, reproductive, and health effects in colonial waterbirds breeding at contaminated Great Lakes sites in Michigan

Abstract

Immunological, reproductive, and population endpoints were assessed in fish-eating birds during 2010–2019 in the Saginaw River and Bay and River Raisin Areas of Concern (AOCs) and Grand Traverse Bay, which are ecosystems historically contaminated with polychlorinated biphenyls, dibenzo-p-dioxins, and dibenzofurans. Reference sites were in the lower St. Marys River (herring gulls and Caspian terns), eastern Lake Superior (terns), and eastern Lake Huron (black-crowned night herons). Relative risk ratios for embryonic nonviability (from both infertility and mortality) in gull embryos were 2–3-fold higher than the reference site in both AOCs and Grand Traverse Bay. Twelve of 13 deformed embryos and chicks (e.g., crossed bills and gastroschisis) were observed at the contaminated sites. Productivity of 4-week-old tern chicks in Saginaw Bay was 35% lower than that at reference sites. In the River Raisin AOC, productivity of 4-week gull chicks was poor in 7 of 10 years. Numbers of breeding herring gulls decreased significantly in the River Raisin AOC, and breeding Caspian terns, a state-threatened species, declined in the Saginaw River and Bay AOC. The mean T cell-dependent phytohemagglutinin skin response was suppressed 50%–56% in gull chicks in both AOCs and Grand Traverse Bay, and 49% in terns and 33% in herons in Saginaw Bay. Antibody responses in gull chicks in the River Raisin AOC and Grand Traverse Bay were 1.6–2-fold lower than reference. Time trend analyses showed no significant improvements in reproductive and immune endpoints in either AOC or Grand Traverse Bay over the study period. Embryonic death increased with time in gulls in the lower Saginaw Bay, and antibody responses decreased in terns in the outer Saginaw Bay.

Introduction

Significant concentrations of persistent organic pollutants, including polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins, and dibenzofurans (PCDDs/Fs or “dioxins”), and organochlorine pesticides have contaminated the Great Lakes since their introduction during the 1940s–1960s. Studies since the late 1960s have shown that colonial fish-eating birds of the Great Lakes are excellent sentinel species for assessing and monitoring effects of contaminants, including reproductive problems, deformities, and immune suppression (Bowerman et al., 2003; Fox et al., 1998, 2007a, 2007b; Grasman et al., 1996, 1998; Kubiak et al., 1989). Studies in the 1980s–1990s found associations between planar PCBs and PCDDs/Fs and deformities described as Great Lakes Embryo Mortality, Edema, and Deformities Syndrome (GLEMEDS; Grasman et al., 1998; Kubiak et al., 1989; Ludwig et al., 1996). This condition in wild birds parallels chick edema disease caused by dioxin-like contaminants in chickens (Gilbertson et al., 1991). Many of the more recent studies of contaminants in fish-eating birds of the Great Lakes have focused on monitoring legacy pollutants such as PCBs and other organochlorines (Brady et al., 2024 [part of the present study]; de Solla et al., 2016; Fuentes et al., 2014; Hammond et al., 2024) and assessing the presence of contaminants of emerging concern such as perfluoroalkyl substances and brominated flame retardants (Brady et al., 2024 [part of the present study]; Gauthier et al., 2019; Letcher et al., 2015; Su et al., 2017 [part of the present study]. The study described here reassessed and monitored biological endpoints in colonial waterbirds breeding in historically contaminated areas of the Great Lakes, including Areas of Concern (AOCs; primarily industrialized locations), within the context of restoration and remediation goals.

Contaminants that act through the aryl hydrocarbon receptor (AhR), such as PCDDs/Fs and planar PCBs, can greatly affect immune function, and the AhR is now recognized to play an important role in immunoregulation (Singh et al., 2020). The developing immune system is particularly sensitive to contaminants, and previous research has shown associations between PCBs and suppression of the T cell-dependent phytohemagglutinin (PHA) skin response in herring gull (Larus argentatus) and Caspian tern (Hydroprogne caspia) chicks at highly contaminated sites in the Great Lakes as well as associations with altered antibody-mediated immunity (Grasman & Fox, 2001; Grasman et al., 1996). Furthermore, the PHA skin response was severely suppressed in herring gull and black-crowned night heron (Nycticorax nycticorax) chicks exposed to dioxins and PCBs in lower New York Harbor, and this PHA response showed strong negative associations (r ≈ –0.9) with concentrations of PCBs and dioxins in the livers of herring gull chicks (Grasman et al., 2013).

The types of immune function assays employed in the studies above and in the present investigation are sensitive indicators of immunological functions important for the survival of wild birds (Fairbrother et al., 2004; Grasman, 2002, 2010; Rollins-Smith et al., 2007). The PHA skin test has been employed in a large number of studies of the interactions between immune function and ecological variables in wild birds, and meta-analyses of these studies have demonstrated the ecological relevance of immune function as assessed by the PHA skin response (Møller & Cassey, 2004; Møller & Saino, 2004). Wild birds with lower PHA responses demonstrate reduced abilities to survive and colonize new areas (i.e., found new local populations). Hence, associations between environmental contaminants and reduced PHA responses in wild birds potentially are associated with population-level consequences (Grasman, 2010).

A second, complementary measure of immune function frequently used in avian immunotoxicology studies is the antibody response following immunization with sheep red blood cells (SRBCs; Grasman, 2010; Grasman & Fox, 2001; Grasman et al., 1996). The avian anti-SRBC hemagglutination test is very similar to standard assays used in laboratory rodents, which have been recognized as extremely sensitive immunotoxicity screening assays and have been incorporated into regulatory toxicology programs in the United States and Europe (Grasman, 2010).

The overall purpose of our study during 2010–2019 was to reassess the reproduction, growth, and immunological health (PHA and SRBC responses) of colonial waterbirds in the Saginaw River and Bay and River Raisin AOCs and Grand Traverse Bay, which have been historically contaminated with legacy organochlorines. Re-evaluation of biological endpoints is important because despite significant improvements in the 1970s–1980s, in many instances, concentrations of legacy organochlorines in Great Lakes colonial waterbird eggs have declined at much lower rates, leveled off, or even increased since the 1990s (de Solla et al., 2016). In the most recent published report available at the beginning of our study (Bowerman et al., 2011), during 2002–2006, herring gull eggs in the Saginaw River and Bay and River Raisin AOCs had some of the highest median concentrations of PCBs (5,952–10,783 ng/g) and dioxin toxic equivalents (TEQs; 686–739 pg/g) in the Great Lakes. Relative to other contaminated areas of the Great Lakes, herring gull eggs from Bellow Island in the Grand Traverse Bay ecosystem had low median PCB concentrations (3,144 ng/g) but similarly high median concentrations of TEQs (739 pg/g) and dichlorodiphenyldichloroethylene. This unique mixture may be a result of the herbicides and insecticides used on orchards in the area surrounding the Grand Traverse Bay. This study allowed comparison of waterbird health in Grand Traverse Bay with that in AOCs with different mixtures of PCBs and PCDDs/Fs contributing to total TEQs. Herring gull eggs from the St. Marys River and Lake Superior (reference areas for our study) contained only 3,094–3,207 ng/g PCBs during 2002–2006 (Bowerman et al., 2011). Median PCB concentrations in plasma of herring gull chicks in the present study mirrored the earlier egg values described above, with low concentrations at the reference site (26.3 ng/g) and in Grand Traverse Bay (27.8 ng/g) and higher concentrations in Saginaw Bay (55.3–113 ng/g) and the River Raisin AOCs (158 ng/g; Brady et al., 2024). Notably, plasma PCB concentrations showed no or minimal declines at all five sites throughout the decade-long study.

Assessment of biological effects in fish-eating birds is relevant to a number of Great Lakes contaminant issues managed by federal and state agencies, including considerations about legacy contaminants and chemicals of emerging concern in AOCs, in historically contaminated areas not designated as AOCs, and in lake-wide regions. In AOCs, such assessments are used to evaluate the Great Lakes Water Quality Agreement Beneficial Use Impairments of bird or animal deformities or reproductive problems and degraded fish and wildlife populations (Bush & Bohr 2012, 2015; Bush et al., 2020). Endpoints employed in this assessment included embryonic mortality, deformities, overall reproductive success, chick growth, and immune function (PHA skin response and SRBC antibody response) in chicks. Specific objectives of this monitoring and assessment program included: (1) investigating population-level effects associated with contaminants in Great Lakes fish-eating birds by assessing breeding numbers and reproductive rates (e.g., embryonic mortality and deformities, fledging success), (2) investigating immunological functions associated with potential population-level effects, (3) investigating these endpoints in certain species (e.g., Caspian terns) whose conservation status is of special concern (e.g., designated as threatened or endangered by state agencies), and (4) assessing whether these health, reproductive, and population effects in birds changed over time (2010–2019), including after remediation efforts (e.g., dredging and capping) in a portion of the lower Raisin River in Monroe during 2014 and 2016.

Methods

Study species and colonies

Reproduction, populations, growth, and immune function were assessed in colonial waterbirds breeding at contaminated and reference sites in Michigan during 2010–2019 (Figure 1, Tables 1 and 2). Sampling during our long-term study varied by year, site, species, and endpoint for various reasons, including intentional design (e.g., later addition of sites as the monitoring program expanded), changing breeding locations (e.g., rising lake levels eliminated breeding areas on islands, so later sampling shifted to other islands), bad weather, logistical difficulties, and occasional colony failures leaving no survivors for measuring later endpoints. In the Saginaw River and Bay AOC of western Lake Huron, hereafter referred to as the Saginaw Bay AOC because waterbird colonies were found only in the bay, field studies were conducted at two herring gull colonies (the Confined Disposal Facility [CDF] in the southern bay and Little Charity Island in the outer bay), two Caspian tern colonies on three islands (the CDF and Charity Reef/Little Charity Island, hereafter referred to as the Charity colony), and one black-crowned night heron colony (CDF). At the Raisin River AOC in western Lake Erie, studies were initiated on the herring gull colony at the Detroit Edison Monroe Power Plant, which is on the western shore of Lake Erie on the south bank of the mouth of the Raisin River. Caspian terns and black-crowned night herons are not known to breed in or around the Raisin River AOC. In Grand Traverse Bay in northeastern Lake Michigan (added as a study site in 2014), herring gulls were studied on Bellow Island. Reference colonies were located in the Lower St. Marys River and Lake Superior: The Pipe Island Twins in Potagannissing Bay for herring gulls and Two Tree Island in Munuscong Lake (2011–2013) and Tahquamenon Island in Whitefish Bay (2017–2019) for Caspian terns. Terns abandoned the Two Tree Island reference colony because of increasing water levels. However, there were no statistical differences in any biological variables between the two tern reference colonies. Reference data for black-crowned night herons came from previous studies using the same methods on Chantry Island in northeastern Lake Huron in 2001–2002 (Grasman et al., 2013). Procedures described below were carried out in accordance with protocols approved by the Calvin University Institutional Animal Care and Use Committee (approval numbers BR2011-01, BR2014-01, and BR2017-01) and under applicable federal and state scientific collecting and banding permits.

Figure 1.

Colonial waterbird colonies studied in Saginaw Bay and River Raisin Areas of Concern (AOCs), Grand Traverse Bay, and reference sites for this assessment of health and reproduction in fish-eating birds during 2010–2019.

Table 1.

Sampling design and sample sizes for reproduction and health effects assessment in herring gulls at contaminated Great Lakes sites in Michigan, 2010–2019.

	Sample size by yeara
Endpoints (# by age)/site	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	Total
Embryonic viability assessment (# eggs)
Pipe Island Twins (reference)	N.S.	N.S.	N.S.	129	66	111	27	15	45	84	477
Saginaw Bay Little Charity Island	102	N.S.	69	105	33	N.S.	21	66	66	30	492
Saginaw Bay Confined Disposal Facility (CDF)	66	N.S.	105	105	57	99	111	99	99	93	834
Monroe Lake Erie	90	N.S.	120	105	51	87	84	81	75	93	786
Bellow Island Grand Traverse Bay	N.S.	N.S.	N.S.	N.S.	18	N.S.	45	69	99	90	321
PHA skin response (# 3 week chicks)
Pipe Island Twins (Reference)	N.S.	19	26	25	28	14	15	17	13	9	166
Saginaw Bay Little Charity Island	28	N.S.	23	22	20	10	22	21	20	26	192
Saginaw Bay Confined Disposal Facility (CDF)	18	21	30	36	30	42	19	19	12	28	255
Monroe Lake Erie	0	13	10	38	30	12	12	16	30	30	191
Bellow Island Grand Traverse Bay	N.S.	N.S.	N.S.	N.S.	33	22	37	35	26	18	171
Growth and antibodiesb (# 4 week chicks)
Pipe Island Twins (Reference)	N.S.	17	23	23	20	11	7	15	13	10	139
Saginaw Bay Little Charity Island	23	N.S.	20	22	19	9	20	16	20	22	171
Saginaw Bay Confined Disposal Facility (CDF)	18	18	30	34	29	31	17	16	9	28	230
Monroe Lake Erie	0	7	4	38	28	12	10	2	28	27	156
Bellow Island Grand Traverse Bay	N.S.	N.S.	N.S.	N.S.	32	18	32	32	26	15	155

Note. PHA = phytohemagglutinin.

^aN.S. = not sampled by design or because of weather or other logistical difficulties; 0 = colony failure with no survivors for sampling.

^bFour-week sample sizes reflect numbers for growth. Sample sizes for antibodies were slightly lower (<7%) because blood samples were not obtained from a few birds, and results for a few samples did not pass quality control criteria for the antibody assay.

Table 2.

Sampling design and sample sizes for reproduction and health effects assessment in Caspian terns at contaminated Great Lakes sites in Michigan, 2010–2019.

	Sample size by yeara
Endpoints (# by age)/site	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	Total
PHA skin response (# 3-week chicks)
Two Tree Island (Reference)	N.S.	33	31	36	N.S.	N.S.	N.S.	N.S.	N.S.	N.S.	100
Tahquamenon Island (Reference)	N.S.	N.S.	N.S.	N.S.	N.S.	N.S.	N.S.	36	18	32	86
Saginaw Bay Charity	17	N.S.	N.S.	N.S.	10	14	40	24	21	N.S.	126
Saginaw Bay Confined Disposal Facility (CDF)	13	21	31	30	10	0	0	26	32	29	192
Growth and Antibodiesb (# 4-week chicks)
Two Tree Island (Reference)	N.S.	33	30	34	N.S.	N.S.	N.S.	N.S.	N.S.	N.S.	97
Tahquamenon Island (Reference)	N.S.	N.S.	N.S.	N.S.	N.S.	N.S.	N.S.	27	18	32	77
Saginaw Bay Charity	N.S.	N.S.	N.S.	N.S.	10	13	36	23	19	N.S.	101
Saginaw Bay Confined Disposal Facility (CDF)	10	20	26	N.S.	10	0	0	17	29	24	136

Note. PHA = phytohemagglutinin.

^aN.S. = not sampled by design or because of weather or other logistical difficulties; 0 = colony failure with no survivors for sampling.

^bFour-week sample sizes reflect numbers for growth. Sample sizes for antibodies were slightly lower (<7%) because blood samples were not obtained from a few birds, and results for a few samples did not pass quality control criteria for the antibody assay.

Embryonic nonviability, reproductive success, growth, and population censuses

Infertility, embryonic mortality, and deformities were assessed quantitatively during mid-late incubation in herring gulls using an embryonic viability detector (Avitronics Digital Egg Monitor, Cornwall, England) and visual observations of non-viable embryos. The egg monitoring device detects vibrations in the egg from heartbeat or motion. One and two egg herring gull clutches were marked during laying and revisited during mid-late incubation (approximately 20–21 days of incubation). Embryonic viability was assessed only in clutches that were completed and retained three eggs through this mid-late incubation stage. At that time, nonviable eggs were opened and examined for infertility, embryonic development, and deformities.

For gulls and terns, enclosures were erected around groups of 5–20 nests during early incubation to facilitate determination of chick survival and to confine chicks for immune assays. Generally, 2–6 enclosures per site were erected for each species, depending on nest density. Enclosures were constructed with metal conduit poles supporting 1.25 × 1.5 cm plastic mesh, approximately 0.7–0.8 m high with a weighted or staked flap along the ground to prevent escape underneath. Chicks were confined until fledging or until the enclosures were removed. Extensive experience has shown that chick survival rates measured within enclosures are generally similar to estimated survival rates outside enclosures. Adult birds adapt to fences and fly in to attend eggs and chicks. All birds were banded with standard USFWS leg bands for individual identification. Chick survival rates at 3 and 4 weeks after median hatch (chick age ranges 2.5–3.5 and 3.5–4.5 weeks, respectively, using age criteria from Grasman et al. (1996)) were determined by dividing the number of chicks by the number of nests for each enclosure. Chick productivity measures for the multiple enclosures at each site were used as replicates for statistical analyses. Even though chicks do not fledge until around 6 weeks of age in these species, survival rates through 3–4 weeks post-hatch are often used as a surrogate measure of fledging success. Daily growth during this phase of linear growth was calculated as the change in body mass between the 3–4 weeks divided by the number of days between measurements of mass (Grasman and Fox, 2001; Grasman et al., 1996, 2013). While not all colonies could be censused because of logistical reasons (e.g., large size; limited time, weather, or personnel), nest censuses (i.e., ground-based counts) were conducted to monitor population sizes of breeding herring gulls at Monroe (in the “Seagull Hill” area of the Detroit Edison Monroe Power Plant) and Caspian terns in Saginaw Bay (Charity Reef, Little Charity Island, and CDF).

T cell-dependent PHA skin response

Immune function tests were initiated at 3 weeks after hatch in gulls and terns and 2 weeks after hatch in herons. The PHA skin response test, a measure of cell-mediated immunity, involves the influx of a variety of mononuclear and polymorphonuclear white blood cells (Martin et al., 2006; Stadecker et al., 1977) and can be modulated by a number of factors, including environmental contaminants, body condition, parasitic infection, hormonal status, resource availability, and life-history traits (Martin et al., 2006). Multiple avian studies have shown the importance of T cells in initiating this response, which is significantly decreased by reducing T cell numbers through thymectomy or irradiation at hatch (Edelman et al., 1986) and through administration of several different immunosuppressive drugs affecting specific T cell functions (Edelman et al., 1986; Grasman and Scanlon, 1995; Schrank et al., 1990). Phytohemagglutinin stimulates T lymphocytes to release cytokines (chemical messengers) that attract a diversity of other white blood cells and increase vascular permeability. Thus, the extent of the swelling caused by the influx of cells and fluid is an integrated measure largely dependent on T lymphocyte function. This in vivo skin test incorporates a number of events in the T cell response, including cell proliferation, differentiation, and cytokine production (Lochmiller et al., 1993; Stadecker et al., 1977).

The PHA skin test was conducted following the procedures of Grasman and Scanlon (1995) using PHA-P (Sigma, St Louis, MO) dissolved in phosphate buffer saline. Feathers were plucked from both wing webs to clear an area for intradermal injections. One wing web was injected with 0.1 ml of 1 mg/ml PHA, while the other received a placebo injection of 0.1 ml of phosphate buffer saline alone. The thickness of each wing web was measured to the nearest 0.05 mm immediately before and approximately 24 hr after the injections using a pressure-sensitive caliper with a low-tension spring that does not crush the skin (Dyer Co., Lancaster, PA). A stimulation index was calculated as the change in the thickness of the PHA-injected wing web minus the change in thickness of the phosphate buffer saline-injected wing web.

SRBC test for antibody-mediated immunity

The SRBC test for antibody production following stimulation with an antigen was initiated on the first day of the PHA skin test in herring gulls and Caspian terns. On this day, chicks were injected via the brachialis vein with 0.1 ml of a 1% SRBC suspension in sterile saline using a 30 ga. needle. Plasma samples were collected from chicks 5–7 days after SRBC injection because antibody titers peak in herring gulls at approximately 6 days post-immunization. Total (IgM + IgG) and 2-mercaptoethanol-resistant (IgG) antibody activities were measured by the microtiter method of Grasman and Scanlon (1995).

Statistical analyses

The primary statistical questions addressed here were whether biological endpoints differed among sites, particularly between historically contaminated and reference sites, and whether endpoints showed directional time trends within sites (e.g., improvement or worsening at contaminated sites). For continuous response variables, inter-site differences were analyzed using one-way analysis of variance (ANOVA; or t-test), followed by Tukey’s pairwise test (for ANOVA). Other factors such as year were not included in the site analysis because sampling considerations limited statistical analyses and confounded interpretations. This long-term, multisite study produced a large data set (Tables 1 and 2), but for a variety of reasons, not all sites could be sampled in every year (see the Study species and colonies section for explanation). As such, missing data confounded inclusion of year as a factor in the statistical analysis. For instance, some years lacked reference data, so the overall mean for that year might be statistically different than other years, but only because it was comprised of data only from contaminated sites. Hence, interpretation of the biological relevance of apparent year effects was complicated by the strong influence of missing data. Nest was not incorporated as a covariate because at the time of chicks sampling, enclosures contained mobile 2.5–4.5-week-old chicks from multiple nests. Hatching could not be monitored throughout extended asynchronous hatch periods to allow marking of nonmobile (i.e., just hatched) chicks and attribution to a particular nest. Inter-site differences in embryonic nonviability rates were analyzed using an epidemiological incidence rate ratio test. Relative risk (RR) ratios were calculated based on incidence rates of infertility and embryonic death at contaminated sites relative to the reference site. RR ratios of 1 indicated no difference in incidence rates between groups. While most site comparisons used colony or island, in limited instances, certain contaminated sites were pooled for secondary analyses to determine whether colony-specific conclusions held at higher levels (i.e., the entire Saginaw Bay AOC with multiple islands for embryonic endpoints in gulls and chick productivity in terns, and all contaminated sites sharing high TEQs for embryonic endpoints in gulls).

Time trends in continuous biological variables were assessed separately for each site using the Jonckheere test for ordered alternatives (Hollander & Wolfe 1973), which is a trend test for detecting monotonic (directional but not necessarily linear) changes in the response variable across ordered groups (e.g., time periods). Time trends in nest numbers were assessed using Pearson’s correlation. Time changes in embryonic nonviability rates were analyzed by comparing RR ratios based on the incidence rates for early vs. later time periods at each site.

JMP 12.1.0 (SAS Institute Inc., Cary, NC) was used for ANOVA, t-test, and correlation analyses, and data for those tests met parametric assumptions of normality and homoscedasticity without transformation. Stata 13.1 (StataCorp, College Station, TX) was used for epidemiological incidence rate ratio test and the Jonckheere trend test across ordered groups.

Results and discussion

Embryonic nonviability and deformities

Embryonic nonviability rates in late incubation herring gull eggs were elevated on both islands in the Saginaw Bay AOC (6.5%–6.7%), in the River Raisin AOC (8.1%), and in Grand Traverse Bay (8.7%) compared to the Pipe Island Twins reference site (3.1%; Figure 2A). Infertility was the primary cause of nonviability at the reference site and was further elevated at all AOC colonies and Grand Traverse Bay. Embryonic death (failed development) also accounted for 30%–50% of total nonviability at all contaminated colonies.

Figure 2.

Reproductive endpoints in herring gulls and Caspian terns at reference sites, Saginaw Bay and River Raisin Areas of Concern (AOCs), and Grand Traverse Bay. Embryonic nonviability in herring gulls (A) in 2010, 2012–2019. Numbers on bars are egg totals. “Undetermined” describes infertile or failed eggs that could not be distinguished due to addling. Reproductive productivity (number of 4-week-old chicks surviving per nest in an enclosure) for herring gulls (B) and Caspian terns (C) in 2010–2019. Numbers on bars indicate number of enclosures. Dotted lines indicate site means. Sites with the same letters were not statistically different by Tukey’s test (p < .05). ANOVA = analysis of variance; SB = Saginaw Bay; CDF = Confined Disposal Facility.

There was strong statistical evidence that RR ratios for overall nonviability were elevated 2–3-fold (p values from <0.0002 to <0.0089) at the AOCs and Grand Traverse Bay (Table 3). RR ratios were elevated 2–3.3-fold for infertility (p values from <0.0011 to <0.02) and 2–3-fold for failed development (p values from <0.0023 to <0.043) at the AOCs and Grand Traverse Bay. These conclusions held for individual islands, grouping both islands within the Saginaw Bay AOC to assess reproductive impacts for the entire ecosystem, and grouping all sites contaminated with PCBs and PCDDs/Fs (i.e., both AOCs and Grand Traverse Bay) to assess collectively all sites sharing high TEQs.

Table 3.

Relative risk ratios for incidence rates of embryonic nonviability, fertility, and failed development in herring gulls in the Saginaw Bay and River Raisin Areas of Concern and Grand Traverse Bay compared to the lower St. Marys River reference site (Pipe Island Twins) during 2010–2019.

	Relative risk ratio (one way exact p-value)
Location	Overall nonviability	Infertilea	Failed developmenta
All contaminated sites combined	2.35 (0.0002)	2.46 (0.0019)	2.32 (0.0095)
Saginaw Bay AOC
Both islands combined	2.11 (0.0021)	2.24 (0.0081)	2.00 (0.038)
SB CDF	2.14 (0.0030)	2.10 (0.020)	2.04 (0.043)
Little Charity Island	2.07 (0.0089)	2.48 (0.0086)	1.94 (0.076)
River Raisin AOC	2.59 (0.0002)	2.49 (0.0044)	2.95 (0.0023)
Grand Traverse Bay (Bellow Island)	2.77 (0.0006)	3.30 (0.0011)	2.12 (0.065)

Notes. AOCs = Areas of Concern; SB CDF = Saginaw Bay Confined Disposal Facility. Relative risk ratios and p-values in bold were statistically significant at p < 0.05.

^aIncludes undetermined eggs that were either infertile or early failed development.

During the course of this study, a number of embryos and chicks were observed with deformities, such as crossed bills and gastroschisis, but only at contaminated colonies except for one gull embryo at the Pipe Island Twin reference site. Deformed birds found at contaminated sites included four gull embryos on the Saginaw Bay CDF, one gull embryo on Little Charity Island, one gull embryo on Bellow Island, three gull chicks at Monroe, and one gull chick and two tern chicks on Little Charity Island.

In the 1980s and 1990s, studies employing new methods for assessing complex mixtures of PCBs and PCDDs/Fs found associations between TEQs and embryonic mortality and deformities in Great Lakes colonial waterbirds (Gilbertson et al., 1991; Grasman et al., 1998; Kubiak et al., 1989; Ludwig et al., 1996). These and other reproductive endpoints became important measures for assessing the status of wildlife inhabiting AOCs and other historically contaminated areas (Bush & Bohr 2012, 2015; Bush et al., 2020). While the ability of these chemicals to cause embryonic mortality, often at very low doses, is generally accepted, subsequent avian field and laboratory studies have led some investigators to question whether coplanar PCBs and PCDDs/Fs consistently cause deformities and edema in birds (Harris & Elliot, 2011). Evaluating potential associations between these chemicals and developmental effects is complex, depending on factors such as species/strain, particular chemical or mixture, dose and exposure regime (route, timing, and duration), and especially the need for very large epidemiological sample sizes to reliably measure rates of rare phenomena like deformities. Conversely, other investigators have incorporated studies on diverse vertebrates (including Great Lakes waterbirds) and cellular mechanistic data into an adverse outcome pathway for AhR-activating chemicals causing early life stage mortality, including craniofacial and cardiac malformations and edema (Shankar & Villeneuve, 2023). The present study with epidemiological sample sizes consistently showed elevated rates of embryonic mortality and infertility in herring gull eggs on four islands across three ecosystems with high TEQs comprised of different mixtures (predominately PCBs in the Saginaw Bay and River Raisin AOCs and PCDDs/Fs in Grand Traverse Bay; Figure 2A and Table 3).

Reproductive success

While mean chick productivity, calculated for each enclosure and then averaged, was consistently good (>0.75 chicks/nest) at reference colonies, herring gulls in the River Raisin AOC and Caspian terns in the Saginaw Bay AOC had the lowest mean productivity and experienced complete (or nearly complete) colony-level reproductive failure in multiple years (Figure 2B and C). Mean productivity of herring gull chicks at 3 and 4 weeks differed significantly among sites (ANOVA p-values <0.0001; data shown only for 4 weeks; Figure 2B). All overall site means, when assessed at 4 weeks, were above the level of 0.8 fledged chicks/nest (fledging at 6 weeks) necessary to maintain a stable population (Kadlec & Drury, 1968). However, productivity at Monroe was below 0.8 chicks/nest in 6 of 10 years, including near complete reproductive failure in 3 years. Furthermore, the apparently high productivity for enclosed nests during 2019 did not represent the very low chick productivity in the rest of the colony—very few surviving chicks were observed outside the enclosures at both 3 and 4 weeks. Hence, chick productivity was poor in 7 of 10 years, including 4 of the last 5 years and near complete reproductive failure in 3 of 10 years. In herring gulls, chick productivity was significantly higher than the reference site on the Saginaw Bay CDF but not on Little Charity Island (Figure 2C). Frequent offal and terrestrial food in regurgitated boluses from herring gull chicks on the Pipe Island Twins suggested the low ecological productivity of the surrounding aquatic ecosystem might have limited chick productivity at this site. Aquatic food was abundant in boluses at other sites, suggesting ample food supply. Predation and epizootic diseases did not appear to be major factors affecting productivity in herring gulls.

Mean productivity of Caspian tern chicks differed significantly among sites (ANOVA p < 0.027 at 3 weeks and p < 0.0063 at 4 weeks) and was lowest in the Saginaw Bay AOC (Figure 2C). When the reference sites were pooled into one group and the two Saginaw Bay colonies into an AOC group in order to elucidate reproduction for the entire AOC, productivity of 4-week terns in Saginaw Bay (mean of 0.68 chicks/nest) was 35% lower than at reference sites (mean of 1.04 chicks/nest; one-tailed T-test p < 0.0016). Overall, these data indicate continuing and consistent reproductive effects in Saginaw Bay AOC Caspian terns, a state-threatened species. Overall, in Saginaw Bay terns, there was no evidence that food supply, predation, epizootic diseases, or weather significantly affected overall productivity.

Declines of breeding populations in River Raisin and Saginaw Bay AOCs

While census data are not available for all years, the breeding population of herring gulls in the colony in the River Raisin AOC declined by approximately 90% between the late 1990s and 2015–2019 (r = –0.96, p < 0.0009; Figure 3A). The declining trend also was apparent during recent years. The numbers of declining nests are reinforced by qualitative observations. The breeding area used consistently for this census includes both beach and adjacent upland areas. The number of nests, nesting density, and overall nesting area in the upland area have continued to decline even as the rising waters of Lake Erie over the last 5 years washed out most nests in the beach area. Hence, nesting density and area in the upland area declined even though many pairs from the beach area were facing pressure to relocate to the upland area.

Figure 3.

Breeding populations based on nest censuses for herring gulls in the River Raisin Area of Concern (AOC; “Seagull Hill” area of the Detroit Edison Monroe Power Plant; A) and Caspian terns in the Saginaw Bay AOC (Confined Disposal Facility, Little Charity Island, and Charity Reef; B).

The breeding population of Caspian terns in Saginaw Bay underwent a significant decline over 13 years (Pearson’s r =  –0.64, p < 0.03; Figure 3B), raising concerns about the status of this state-threatened species in this location. Total nest numbers for three islands in the Bay (CDF, Charity Reef, and Little Charity Island) ranged from 198 to 638 during 1994–2003, with a mean of 371 nests. This monitoring project recorded a steep decline in the number of nests since 2007. The mean total number of nests for the Bay during the last 3 years of the study was only 188, 49% lower than the mean for the earlier period reported above. The population decline in Saginaw Bay Caspian terns is consistent with earlier studies documenting low survival and recruitment into the breeding population of young terns raised in this ecosystem and other PCB-contaminated areas in the upper Great Lakes (Ludwig, 1979; Mora et al., 1993) and with the ongoing lower pre-fledgling productivity found in our study.

Growth

Growth (weight gain) in pre-fledgling gulls differed significantly among sites (ANOVA p < 0.0001), with growth rate in Monroe in the Raisin River AOC being significantly higher than in all other colonies except the Saginaw Bay CDF (Figure 4A). Only Monroe had a mean multiyear growth rate in the expected range for herring gull chicks of 14–20 g/day established by a previous study (Grasman et al., 1996). Although Monroe had the highest mean growth rate overall, it was below the expected growth range in 5 out of 9 years of the study. Gull chicks at both AOCs and Bellow Island had poor growth rates in almost half of the years studied. Mean growth in tern chicks was significantly lower on the Saginaw Bay CDF compared to all other colonies (ANOVA p <0.0001; Figure 4B) and was below the expected range of 4.0–18 g/day for Caspian terns (Grasman et al., 1996). Growth in this colony was particularly low in 4 out of 7 years.

Figure 4.

Mean change in body mass between 3 and 4 weeks of age in pre-fledgling herring gulls (A) and Caspian terns (B) in Saginaw Bay and River Raisin Areas of Concern, Grand Traverse Bay, and reference sites in 2010–2019. Numbers on bars indicate sample sizes, and error bars indicate standard error. Dotted lines indicate site means. Sites with the same letters were not statistically different by Tukey’s test (p < .05). ANOVA = analysis of variance; SB = Saginaw Bay; CDF = Confined Disposal Facility.

The extremely poor growth in some years of herring gull and Caspian tern chicks at most sites in the Saginaw Bay and River Raisin AOCs is notable because the Saginaw Bay and western Lake Erie are highly productive ecosystems with an abundance of fish prey. Hence, the poor growth is not likely to be caused by food supply shortages in these locations but instead by other stressors such as environmental contaminants (e.g., wasting syndrome induced by dioxin-like chemicals including PCBs), algal toxins (e.g., microcystins), or epizootic diseases. There was no evidence that weather events coincided with sites and years with low growth.

Immune function tests

Herring gull and Caspian tern chicks had lower PHA skin responses at all AOC sites as well as at Grand Traverse Bay compared to the reference sites (ANOVA p < 0.0001; Figure 5A and B). The mean PHA skin response of gulls was 55%–56% lower at the AOCs and 50% lower in Grand Traverse Bay than at the Pipe Island Twins site. Caspian terns on both islands in the AOC had a 49% lower mean response as compared to the reference sites. Although data were collected in different years, black-crowned night herons on the Saginaw Bay CDF (2010) showed a 33% lower PHA skin response compared to herons at the Chantry Island reference site (2001–2002; t-test p < 0.05; Figure 5C). In the Saginaw Bay AOC, the suppressed PHA skin response was replicated in all three species studied (gulls, terns, and herons) and was replicated on two islands for each of the two species (gulls and terns). The magnitude of the suppression in the PHA response at the AOCs and Grand Traverse Bay was biologically significant, comparable to the maximal reductions in this response caused by immunosuppressive drugs, radiation, or complete removal of the thymus gland (the site of T lymphocyte maturation) in chickens, which causes a reduction in the stimulation index by approximately 50%–60% (Edelman et al., 1986; Grasman & Scanlon, 1995; Schrank et al., 1990).

Figure 5.

Mean phytohemagglutinin (PHA) stimulation index, a T cell-dependent response, in pre-fledging herring gulls (A), Caspian terns (B), and black-crowned night herons (C) at Saginaw Bay and River Raisin Areas of Concern, Grand Traverse Bay, and reference sites in 2010–2019. Numbers on bars indicate sample sizes, and error bars indicate standard errors. Sites with the same letters were not statistically different by Tukey’s test (p < .05). ANOVA = analysis of variance; SB = Saginaw Bay; CDF = Confined Disposal Facility.

Herring gull chicks at the River Raisin AOC and in Grand Traverse Bay had 1.6–2-fold lower total antibody and 2.3–2.5-fold lower IgG responses than chicks at the Pipe Island Twins reference site (ANOVA p < 0.0001 for total, p < 0.0001 for IgG; Figure 6A). Total and IgG responses in gulls in the Saginaw Bay AOC were similar to those at the reference site. Total antibody responses in Caspian terns were significantly higher at the Charity colony than at all other sites (ANOVA p < 0.0002; Figure 6B). IgG responses were significantly lower at Tahquamenon Island than at the Charity colony and Two Tree Island (ANOVA p < 0.0001).

Figure 6.

Mean total and 2-mercaptoethanol-resistant (IgG) antibody titers in pre-fledging herring gulls (A) and Caspian terns (B) in Saginaw Bay and River Raisin Areas of Concern, Grand Traverse Bay, and reference sites in 2010–2019. Sites sharing letters were not significantly different by Tukey’s test (p < .05). Numbers indicate sample sizes. Error bars indicate standard error. ANOVA = analysis of variance; SB = Saginaw Bay; CDF = Confined Disposal Facility.

The immunosuppressive effects of chemicals acting via the AhR have been documented extensively at the mechanistic and physiological levels in cell culture, laboratory animals, and humans, including suppression of delayed-type hypersensitivity responses very similar to the PHA skin response (Singh et al., 2020). Several egg injection studies of PCBs/PCDDs/Fs with no additional post-hatch dosing did not find subsequent effects on the PHA skin response in young birds. These studies examined individual coplanar congeners (PCBs 77 and 126) and a complex organochlorine mixture with high TEQs (extracted from double-crested cormorant [Phalacrocorax auritus] eggs from a contaminated Great Lakes site) in chickens (Lavoie & Grasman, 2007; Lavoie et al., 2007) and an Aroclor mixture in American kestrels (Falco sparverius; Smits et al., 2002). However, several field studies have shown negative associations between PCB and PCDD/F exposure and suppression of the PHA skin response in wild birds. Polychlorinated biphenyls and TEQ concentrations in eggs had strong negative associations with the PHA skin response in both herring gull and Caspian tern chicks in the Great Lakes (Grasman et al., 1996), and a follow-up study found a similar strong negative relationship between this response and PCBs in plasma of individual tern chicks (Grasman & Fox, 2001). The PHA skin response was significantly suppressed in both herring gull and black-crowned night heron chicks exposed to PCBs/PCDDs/Fs in New York Harbor, with strong negative correlations (r ≈ –0.9) between the response and PCBs, TCDD, and TEQs in the livers of individual herring gull chicks (Grasman et al., 2013). Gulls at the contaminated site in that study showed other signs of T cell-mediated immunosuppression, including thymic atrophy (reduced numbers of thymocytes) in embryos and reduced in vitro T cell mitogen-induced proliferation of lymphocytes from chicks. Black guillemot (Cepphus grylle) chicks exposed to PCBs spilled from a demolished military radar installation with no other significant legacy pollutants exhibited both a suppressed PHA skin response and atrophy of the thymus, the organ for T cell maturation (Brown et al., 2013). While the suppressed PHA skin response observed in multiple species and sites in the present study is consistent with these other field studies of birds exposed to similar dioxin-like chemicals, future studies also will investigate the potential for other chemicals, particularly contaminants of emerging concern such as per- and polyfluoroalkyl substances, to contribute to immunological effects in Great Lakes birds.

Although PCBs/PCDDs/Fs often affect B cells and antibody-mediated immunity in laboratory animals and humans, observed effects on the SRBC antibody response in birds have been more variable. In Great Lakes herring gull and Caspian tern chicks, no associations were found between SRBC antibody titers and PCBs or TEQs measured in eggs (Grasman et al., 1996), but a positive association between this response and plasma PCB concentrations was found in Caspian tern chicks (Grasman & Fox, 2001). In chickens, in ovo exposure to an organochlorine mixture with high TEQs extracted from cormorant (P. auritus) eggs resulted in an increased SRBC antibody response in chicks (Lavoie et al., 2007), but similar experiments with individual PCB congeners 77 and 126 decreased the response (Lavoie & Grasman, 2007; Lavoie et al., 2007).

Meta-analyses of numerous avian field studies have demonstrated that wild birds with lower immune responses, including the PHA skin response, have reduced abilities to survive and colonize new areas (Møller & Cassey, 2004, Møller & Saino, 2004). Hence, associations between environmental contaminants and reduced immune responses in wild birds possibly are connected to population-level consequences (Grasman 2010) and may contribute to low chick productivity, low recruitment, and (or) population declines observed in colonial waterbirds in the Saginaw Bay and River Raisin AOCs.

Time trends in biological endpoints at the reference sites

No statistically significant time trends were observed in any biological endpoints at the reference sites in either herring gulls or Caspian terns, except for a moderately significant increasing trend (Jonkheere p < 0.027) of low magnitude (increasing <10% over 9 years) in the PHA skin response of herring gulls at the Pipe Island Twins.

Time trends in biological endpoints in the River Raisin AOC

An important objective of this study was to determine whether health, reproductive, and population-level indicators of biological integrity have improved over time in fish-eating birds in two AOCs (River Raisin and Saginaw Bay) and Grand Traverse Bay, a third area contaminated by dioxin-like chemicals. The wildlife deformities and reproductive Beneficial Use Impairment identified under the Great Lakes Water Quality Agreement (degraded wildlife populations and deformities/reproductive problems) remain designated for the River Raisin AOC. Additionally, both the PHA response for cell-mediated immunity and anti-SRBC antibody titers in River Raisin herring gull chicks were significantly suppressed to a similar degree during the early 1990s (Grasman et al., 1996) and the present study. Overall, for the entire 10-year period of the present study, all measures of reproduction (embryonic and chick survival) and health (growth, PHA skin response, and antibody response) in herring gulls in the River Raisin AOC showed significant deficits compared to the reference site. These effects were explored further using time-trend analyses.

Our study revealed no improvements over time in these biological endpoints despite remediation dredging in 2014 and 2016 to remove PCB-contaminated sediments from the lower River Raisin. Although other contaminants or ecological factors also could have contributed to the ongoing biological effects, elevated PCB concentrations in plasma of gull chicks did not decline at Monroe during the duration of this study (Brady et al., 2024). Analysis of RR based on incidence rates of embryonic nonviability, infertility, and embryonic death did not change significantly irrespectively of whether time groups were partitioned equally over the study period (first 5 years vs. second 5 years; all RR ratios <1.6, p-values >0.19) or pre-/post-remediation (2010–2016 vs. 2017–2019; all RR ratios <1.26, p-values >0.13). Likewise, there were no significant time trends in productivity of 4-week chicks (Jonkheere p < 0.32), growth (Jonkheere p < 0.07), PHA skin response (Jonkheere p < 0.53), or anti-SRBC antibody titers (Jonkheere p < 0.43). The breeding population of herring gulls decreased over 90% since the late 1990s (r = –0.96, p < 0.009), with declines continuing in recent years (Figure 3A).

Time trends in biological endpoints in the Saginaw Bay AOC

Two wildlife-related Beneficial Use Impairments remain in the Saginaw Bay AOC (degraded wildlife populations and deformities/reproductive problems). Both the PHA response for cell-mediated immunity and anti-SRBC antibody titers as well as growth were significantly suppressed in herring gull and Caspian tern chicks in this AOC during the 1990s (Grasman & Fox, 2001; Grasman et al., 1996). Furthermore, chick growth and productivity were greatly reduced in both species in this location during this earlier time period (Grasman et al., 1996).

Overall, total nonviability, infertility, and embryonic death in herring gulls eggs on Little Charity Island and the Saginaw Bay CDF were significantly higher than reference sites (Table 3), and few significant changes were observed with time. On Little Charity Island, no significant time trends were observed when comparing incidences between the first and second halves of the decade (all RR ratios <0.94, p-values >0.28). On the Saginaw Bay CDF, infertility did not change from the first to last 5-year increments (RR ratio =1.16, p <0.69), but there was a marginal increase over time in total nonviability (RR ratio =1.82, p <0.043) that was driven by a large increase in embryonic death in the later years (RR ratio =4.43, p < 0.0063). This increase might have been caused by increased bioavailability of embryotoxic chemicals, either legacy pollutants or contaminants of emerging concern.

Few significant changes over time were observed in biological endpoints measured in gull chicks in this AOC, which paralleled the sustained elevated concentrations of PCBs in plasma PCBs in these gull chicks over the course of the study (Brady et al., 2024). Productivity of 4-week-old gull chicks did not change significantly with time on Little Charity Island (Jonckheere trend p < 0.76) or the Saginaw Bay CDF (p < 0.95). Likewise, the PHA response did not change over time (Jonckheere trend p < 0.84 at Little Charity and p < 0.16 on the Saginaw Bay CDF), but as noted previously, this response was significantly suppressed at both Saginaw Bay colonies relative to the reference site for the entire 10 years of study (Figure 5A). The anti-SRBC antibody response did not change over time at either colony (Jonckheere trend p < 0.17 at Little Charity and p < 0.91 on the Saginaw Bay CDF). Growth between 3 and 4 weeks of age showed an increase over time at both sites (Jonckheere trend p < 0.0001 at Little Charity Island and p < 0.003 on the Saginaw Bay CDF). This trend was driven by two exceptionally poor years of weight loss or minimal growth at both colonies during early years of the study (2012 and 2014; Figure 5A).

Biological endpoints in Caspian terns in Saginaw Bay showed no trends of improvement over the course of this study, and in some cases worsened for terns in the Charity colony. Chick productivity at 4 weeks, growth between 3 and 4 weeks, and the PHA skin response were all lower in young terns on the Saginaw Bay CDF compared to reference sites. None of these endpoints improved over the course of the study (Jonckheere trend test p-values >0.54). Total anti-SRBC antibody titers were higher at the Saginaw Bay CDF compared to those at the reference site but did not change significantly over time (Jonckheere trend p < 0.85). Terns in the Charity colony also showed lower productivity at 4 weeks and PHA skin responses compared to those at reference sites with no improvements over time (Jonckheere trend test p-values >0.23). Growth and total antibody responses in terns at the Charity colony were not different from reference sites for the entire study period, but both responses decreased over time for Charity terns (Jonckheere trend p < 0.001 for growth and p < 0.009 for total antibody). Breeding populations of Caspian terns across Saginaw Bay decreased significantly over the course of the study (r = –0.64, p < 0.025; Figure 3B).

Time trends in biological endpoints in Grand Traverse Bay

Although not an AOC, a substantial amount of dioxin-like contamination, primarily the result of PCDDs/Fs and not PCBs, has been identified in herring gull eggs in Grand Traverse Bay (Bowerman et al., 2011), giving this ecosystem a different mixture of contaminants than the two AOCs that were part of this study. Over the time of the current study, there was marginal statistical evidence of a decreasing trend over time in total embryonic nonviability for herring gulls (RR ratio = 0.45, p < 0.04), but this was not supported by similar trends in infertility (RR ratio = 0.47, p < 0.10) or embryonic mortality (RR ratio = 0.70, p < 0.58), the components that make up total nonviability. Overall growth was not impaired relative to reference, but there was a moderate trend of increasing growth with time (Jonckheere trend p < 0.016), although growth was not impaired relative to reference. The trend in growth was driven by high growth rates in 2017 and 2018, but growth in 2019 was more similar to that in 2014–2016 (Figure 4A). No trends for changes with time were observed in reproductive productivity of 4-week chicks (Jonckheere trend p < 0.55), PHA skin response (Jonckheere trend p < 0.08), or total antibody titers (Jonckheere trend p < 0.51), the latter two of which were lower than the reference site. Overall, no biologically significant improvements over time were observed in herring gulls from Grand Traverse Bay.

Conclusions

This assessment investigated reproduction and immunological health of fish-eating birds at Great Lakes sites with historic and ongoing contamination, including the Saginaw Bay and Raisin River AOCs, as part of the Great Lakes Restoration Initiative with additional support from the U.S. Fish and Wildlife Service. The results of our study clearly indicate ongoing immunological, developmental, and reproductive impacts in birds at these AOCs that are consistent with previous studies on the effects of persistent pollutants, such as PCBs, in Great Lakes wildlife (Table 4). In the AOCs, altered biological endpoints and continued elevated PCB exposure (Brady et al., 2024) have persisted despite significant earlier efforts to control sources of PCBs in and along the lower Saginaw River and lower River Raisin, and ongoing efforts to control sources of PCDDs/Fs upstream in and along the Tittabawassee River. Our study also found similar impacts in herring gulls in Grand Traverse Bay, an area with a different mixture of contaminants than many other Great Lakes sites (i.e., elevated TEQs with a greater contribution of PCDDs/Fs and lower contribution of PCBs). The present study’s conclusions were strengthened by the replication of reproductive and immunological health effects in multiple avian species at multiple sites contaminated with dioxin-like chemicals and by the consistency of the findings with previous field studies and mechanistic laboratory studies.

Table 4.

Key findings for this assessment of health, reproduction, and population status in colonial waterbirds in Michigan, 2010–2019.

Area	Key findings
Saginaw Bay AOC	Herring gulls, Caspian terns, and black-crowned night herons showed reductions in immune responses and (or) reproduction, consistent with past studies
	Embryonic nonviability, including both infertility and failed development, was elevated in gulls Terns had lower overall productivity of chicks in the AOC compared to reference sites Growth of tern chicks was significantly lower on the CDF than the reference site, with particularly low growth rates in 4 out of 7 years Suppressed T cell-mediated immunity was demonstrated in herring gulls, Caspian terns, and black-crowned night herons Time trend analyses showed no significant improvements in reproductive and immune endpoints Embryonic death increased with time in herring gull eggs on the CDF The antibody response decreased over time in Charity terns Numbers of breeding Caspian terns, a state-threatened species, declined significantly (2007–2019)
River Raisin AOC	Herring gulls at the River Raisin AOC showed reductions in immune responses and reproduction, consistent with past studies
	Embryonic nonviability, including both infertility and failed development, was elevated in gulls Chick productivity was poor in 7 of 10 years, including 4 of the last 5 years and complete reproductive failure in 3 of 10 years Low growth rates in gull chicks in 5 out of 10 years Suppressed T cell-mediated immune response Suppressed total antibody and IgG responses Time-trend analyses showed no significant improvements in reproductive and immune endpoints Numbers of breeding herring gulls decreased significantly (1995–2019)
Grand Traverse Bay	Herring gulls at the Grand Traverse Bay colony, a site with high TEQs and DDEs, showed reductions in immune responses and reproduction
	Elevated embryonic nonviability, including both infertility and failed development Low growth rates in 4 out of 6 years Suppressed T cell-mediated immune response Suppressed total antibody and IgG responses Time trend analyses showed no significant improvements in reproductive and immune endpoints

Note. Years of trends including information from before 2010 are noted in the Table.

AOCs = Areas of Concern; CDF = Confined Disposal Facility; DDEs = dichlorodiphenyldichloroethylenes; TEQs = toxic equivalents.

Data availability

The data underlying the results reported in this paper are available from the corresponding author upon request (keith.grasman@calvin.edu).

Author contributions

Keith A. Grasman (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization; Writing—original draft; Writing—review & editing), Mandy Annis (Conceptualization, Funding acquisition, Project administration, Supervision, Writing—review & editing). Carly Eakin (Funding acquisition, Project administration, Writing—review & editing). Jeremy Moore (Funding acquisition, Project administration). Lisa L. Williams (Conceptualization, Funding acquisition, Project administration, Supervision, Writing—review & editing)

Funding

Funding was provided by the Great Lakes Restoration Initiative through the U.S. Fish and Wildlife Service Midwest Region Ecological Services Program (grants F10AP00096, F11AP00577, F14AC01169, F16AP01041, F18AP00638, and F19AC00179) and the Calvin University School of Science, Technology, Engineering, and Mathematics undergraduate research program.

Conflicts of interest

The authors declare no conflicts of interest.

Disclaimer

The findings and conclusions in this manuscript are those of the authors and do not necessarily represent the views of the U.S. Fish and Wildlife Service or the U.S. Environmental Protection Agency.