Geographic variation of mercury in breeding tidal marsh sparrows of the northeastern United States

Saltmarsh sparrows (Ammospiza caudacuta) and seaside sparrows (A. maritima) are species of conservation concern primarily due to global sea-level rise and habitat degradation. Environmental mercury (Hg) contamination may present additional threats to their reproductive success and survival. To assess site-specific total mercury (THg) exposure and identify environmental correlates of THg detection, we sampled blood from adult male saltmarsh and seaside sparrows at 27 sites between Maine and Virginia, USA. The mean THg concentration (±1 SD) throughout the entire sampling range was 0.531 ± 0.287 µg/g wet weight (ww) for saltmarsh sparrows and 0.442 ± 0.316 µg/g ww for seaside sparrows. Individual THg concentrations ranged from 0.135–1.420 µg/g ww for saltmarsh sparrows and 0.153–1.530 µg/g ww for seaside sparrows. Model averaging from a suite of linear mixed models showed that saltmarsh sparrows averaged 20.1% higher blood THg concentrations than seaside sparrows, potentially due to differences in diet or foraging behavior. We found no evidence for a relationship between sparrow THg concentrations and land cover surrounding sampled marshes or average precipitation-based Hg deposition. Overall, our results suggest considerable, unexplained variation in tidal marsh sparrow blood THg concentrations over their co-occurring breeding ranges.


Introduction
Mercury (Hg) is a widespread environmental contaminant that can pose a threat to wildlife by affecting neurological and reproductive systems (Evers, 2018). Common anthropogenic sources of inorganic Hg include coal-fired power plants, waste incinerators, chlorine and metal processing plants, landfills, and artisanal and small-scale gold mines that emit Hg into the atmosphere or directly into waterways (UNEP, 2019). Once deposited into aquatic systems, either by atmospheric processes or leachate, inorganic Hg can be converted into the neurotoxic and bioavailable methylmercury (MeHg). Generally, organisms that occupy higher trophic levels are disproportionately affected by MeHg because it migrates through food webs via biomagnification . Increased concentrations of MeHg in tissue samples of piscivorous and insectivorous birds are associated with impaired coordination, foraging apathy, reduced reproductive success, and reduced survivorship (Scheuhammer, 1987;Bouton et al., 1999;Evers et al., 2008).
In anoxic aquatic sediments, sulfate-reducing bacteria accelerate the production of MeHg (Compeau & Bartha, 1985), leading to elevated risks for organisms living in, around, or feeding from contaminated wetlands. Both the saltmarsh sparrow (Ammospiza caudacuta) and seaside sparrow (A. maritima) are tidal marsh specialists and spend their entire life cycles within saltmarsh habitats, where risk of exposure to MeHg can be elevated. Being dependent on salt marshes, which are under increasing threat of degradation and loss, tidal marsh sparrows are species of high conservation concern (Rosenberg et al., 2016). Anthropogenic activities including coastal development, marsh fragmentation, and alteration of hydrology have all reduced habitat available for both saltmarsh and seaside sparrows , and global sea-level rise threatens saltmarsh systems range-wide (Bayard & Elphick, 2011;Field et al., 2017;Greenlaw et al., 2020;. The saltmarsh sparrow is considered "globally endangered" by the International Union for Conservation of Nature in response to rapid population declines (BirdLife International 2020), decreasing at 9% per year with over 70% of the global population lost since the 1990s (Correll et al., 2017;Roberts et al., 2019). Sea-level rise and increased tidal flooding risk are the major drivers of population loss for this species, as approximately 60% of all nest failures occur due to flooding (Shriver et al., 2007;Gjerdrum et al., 2008), and are projected to cause extinction by the middle of the 21st century (Field et al., 2017(Field et al., , 2018Roberts et al., 2019). Because salt marshes and coastal estuaries are often impacted by polluted urban runoff (Schwarzbach et al., 2006), heavy metal contamination may present additional challenges to sparrow reproductive success and survival.
Tidal marsh sparrows can be effective bioindicators for MeHg contamination of saltmarsh systems, and for quantifying the potential threat that MeHg poses to marsh bird populations, at a variety of geographic scales (Shriver et al., 2006;Warner et al., 2010;Lane et al., 2011Lane et al., , 2020Winder, 2012;Ruskin et al., in review). In all prior research, tidal marsh sparrow blood Hg levels have approached or exceeded 0.7 µg/g wet weight (Shriver et al., 2006;Lane et al., 2011Lane et al., , 2020Warner et al., 2010;Winder, 2012;Ruskin et al., in review), a level that has been shown to influence nesting success in the Carolina wren (Thryothorus ludovicianus) (Jackson et al., 2011). Though atmospheric deposition is known to account for the majority of Hg present in the environment (Hudson et al., 1995;Lindberg et al., 2007;Risch et al., 2012Risch et al., , 2017, breeding tidal marsh sparrow Hg concentrations in the northeastern USA vary significantly by site (Warner et al., 2010;Lane et al., 2011Lane et al., , 2020Ruskin et al., in review) without any consistent inter-annual temporal trends (Lane et al., 2020). These findings suggest that annual atmospheric deposition alone does not drive Hg bioavailability in coastal wetlands of the northeastern USA. Rather, variation in tidal marsh sparrow Hg levels among marsh sites is more likely due to sitespecific MeHg production rates, differences in marsh hydrodynamics or biogeochemistry, differences in local food webs and food availability, or nearby point sources of the pollutant and surrounding land use (Lane et al., 2011(Lane et al., , 2020. Little is known about the dynamics of Hg in tidal wetlands or how local land cover influences Hg levels. Wetlands are positively correlated with the presence, production, and export of MeHg in the hydrological system because they are rich in dissolved organic carbon (St. Louis et al., 1994;Krabbenhoft et al., 1995;Driscoll et al., 1995Driscoll et al., , 1998Wiener & Shields, 2000). Forest canopies influence the rate of Hg deposition by absorbing atmospheric Hg through leaf stomata, depositing Hg into terrestrial and aquatic systems through litterfall, or enhancing the Hg concentration of precipitation as it falls through the canopy (Rea et al., 1996). Sources of direct Hg discharges, such as metal-processing plants, landfills, and waste-water treatment plants, tend to aggregate in urbanized environments and are linked to Hg contamination within drainage basins (Niebla et al., 1976;Hildebrand et al., 1980;Gilmour & Bloom, 1995). Impervious surfaces also direct Hg to surface water via storm water drainage systems (Rule et al., 2006). Therefore, we predicted that MeHg levels in marshes and marsh-associated birds would be affected by the relative amount of wetlands, forests, and developed areas in the surrounding landscape.
Range-wide analyses of Hg contamination are lacking for most species, including saltmarsh and seaside sparrows. Within their co-occurring breeding ranges, five major geographic gaps in Hg sampling exist: coastal New Jersey, Delaware, Maryland, Virginia, and the eastern Chesapeake Bay. These subregions collectively support over 65% of both the global saltmarsh sparrow population and the northeastern seaside sparrow (A. m. maritima) population (Wiest et al., 2016). The objectives of our study were to, 1) quantify regional variation in the mercury detected in saltmarsh and seaside sparrows across a large portion of their breeding ranges in the northeastern USA, and 2) identify environmental and spatial characteristics that influence patterns of mercury detection. We sought to conduct an assessment of local mercury contamination, which could potentially inform conservation and remediation efforts.

Sample collection
To quantify regional variation of MeHg exposure in saltmarsh and seaside sparrows, we collected blood samples from 67 adult male saltmarsh sparrows and 53 adult male seaside sparrows at 27 marsh sites along the coast of the northeastern and mid-Atlantic USA from May 29 to July 31, 2018. Marsh sites were located from Maine south to Virginia (Fig. 1, Table 1) and were selected using a randomization process that first used generalized random tessellation stratified sampling to generate bird survey points (Wiest et al., 2016). Points were then randomly subsampled after stratifying by sparrow abundance (see "Rapid Demo SOP," www.tidalmarshbirds.org).
Because blood samples can reflect days to weeks of dietary exposure (Evers, 2018), and can reliably represent site-specific MeHg availability in migratory songbirds within 1-2 weeks (Ma et al., 2018), they are a useful sampling tissue to understand tidal marsh sparrow MeHg exposure in the local environment Shriver et al., 2006;Kopec et al., 2018;Lane et al., 2011Lane et al., , 2020. Tidal marsh sparrows arrive on the breeding grounds from mid-April through late May, and do not begin their prebasic molt until mid-to late August . Therefore, sampling during late-May, June, and July removes the confounding possibility of birds depurating their MeHg load into feathers during molt. We only sampled adult males to avoid the confounding factors associated with females passing on their Hg body burden to eggs during the laying period (Brasso et al., 2010;Ackerman et al., 2013;Ou et al., 2015). To reduce the risk that blood samples reflect Hg levels from nonbreeding or migratory sites, we delayed sampling until after most migration was complete, conducted our earliest sampling towards the southern end of the range where birds initiate breeding sooner than northern subpopulations, and focused on males, which tend to arrive earlier than females (Borowske et al., 2017). Although we do not know the arrival dates of individual birds, which vary greatly even within sites, most birds likely had the opportunity to forage at their breeding site and incorporate dietary-MeHg from the local environment prior to sampling.
Sparrows were captured within a 50-m radius of a given site using mist nets and fitted with an aluminum US Geological Survey leg band. Individuals identified as males via a cloacal protuberance were weighed, and 10-50 µL of blood were collected in 70 µL heparinized capillary tubes using brachial venipuncture. Capillary tubes were immediately sealed at both ends using Leica Microsystems Crito-caps™, placed into a Vacuette™, and stored in a cooler with ice packs. Samples were transferred into a freezer within 8 h and were maintained below −4°C until analysis.

Laboratory analysis
Blood samples were analyzed for total Hg (THg), the combination of methylated and unmethylated Hg in the blood, instead of MeHg because Rimmer et al. (2005) showed that the total Hg in songbird blood is composed of approximately 95% MeHg. THg analysis was performed at the Texas A&M University Trace Element Research Laboratory in College Station, Texas using a combustion, gold trapping, atomic absorption spectroscopy approach. The detector used was a Nippon MA-3000 direct thermal decomposition Hg analyzer equipped with dual cell detectors. Samples were thawed, removed from capillary tubes, weighed to the nearest 0.01 mg, transferred to precombusted cuvettes, and then placed into the Nippon MA-3000's autosampler. Samples were dried and combusted under a constant flow of O 2 , combustion gases were passed through a gold trap to remove Hg gas from the combustion stream, and the gold trap was then heated to release concentrated Hg 0 to be swept into two atomic absorption cells. THg concentrations in samples were quantified by comparing absorption peaks with known calibration standards. Instrument calibration was performed using a blank and three calibration standards in both detector cells and verified every 10 samples using two certified reference materials and a continuing calibration blank. Reference materials, including mussel tissue (NIST 2976) and tuna muscle (ERM-CE464), were used to evaluate the calibration on the more sensitive and less sensitive cells, respectively. This approach has been incorporated by the United States Environmental Protection Agency (EPA) in EPA SW-846 Method 7473 for analyzing solid waste and leachate (USEPA, 1998). Instrument detection limit was 0.05 ng THg and final results for THg concentration are reported in µg/g wet weight (ww).

Statistical approach
Using an information-theoretic approach, we built two linear mixed model candidate sets to determine a statistically relevant spatial extent for land cover extraction, as well as which spatial, environmental, biological, and temporal parameters are associated with blood THg variation in tidal Table 1 Location information, sample sizes (n), and species blood total mercury (THg) concentrations (mean ± SD and range) for all sampling locations, Maine to Virginia, USA, May 29 to July 31, 2018. A dash (−) indicates that there are no data to report marsh sparrows among marsh sites. We selected independent variables based on a set of a priori hypotheses (Burnham & Anderson, 2002) derived from known dynamics of Hg in the environment, and scaled these variables (µ = 0, sd = 1) because raw values occurred at different orders of magnitude. We constructed models in R (v. 3.6.2; R Core Team, 2019) using the package "lmerTest" (Kuznetsova et al., 2017), and fit models with maximum likelihood estimations (Zuur et al., 2009). Model selection was performed using second-order Akaike's information criterion for small-sample sizes (AIC C ; Burnham & Anderson, 2002). We inspected the distribution of the blood THg concentrations for normality using quantile-quantile plots and a Shapiro-Wilk normality test, and rejected the null hypothesis that the data were normally distributed (W = 0.881, p < 0.001). To achieve normality, we used a natural logtransformation (W = 0.979, p = 0.054). Four high outliers were present prior to transformation, but were included in the analysis because we had no reason to believe they were erroneous. Following transformation, there was no evidence that homogeneity of variance, normality of residuals, and normality of random effects assumptions were violated, based on visually inspecting quantile-quantile plots and plotted residuals versus predicted values (Zuur et al., 2009).
To populate the full linear model, we extracted 30 × 30 m resolution land cover data (percent of total area) around each sampling point from the National Land Cover Database 2016 CONUS data (NLCD, Yang et al., 2018) using the R package "raster" (Hijmans 2021) and lumped land cover types into 3 relevant categories: wetland (NLCD categories 90 and 95 combined), forest (NLCD categories 41, 42, 43, and 52 combined), and developed land (NLCD categories 21, 22, 23, and 24 combined). We also included average precipitation-Hg deposition from 2017 (µg/m 2 , the year preceding our sampling season) around each sampling point sourced from the Mercury Deposition Network of the National Atmospheric Deposition Program (NRSP-3 2021). This metric does not account for total atmospheric Hg deposition, and importantly ignores literfall-Hg deposition -which often exceeds precipitation-Hg deposition across the continental USA (Risch et al., 2012(Risch et al., , 2017. However, forested land cover may serve as a rough proxy to account for potential differences in litterfall-Hg deposition rates throughout our sampling range, since canopies track interannual shifts in atmospheric Hg deposition and continental Hg emissions (Risch et al., 2017).
We included species as a fixed effect because the uptake of MeHg occurs mainly through diet; and the foraging and relative size differences between species (Woolfenden, 1968) could influence MeHg uptake measured in the blood. While we made a considerable effort during the sampling design to control for factors that influence avian blood Hg concentrations, including age, sex, and the depuration of Hg into eggs and feathers, tidal marsh sparrow blood Hg concentrations have repeatedly been shown to increase with time spent at their breeding location (Kopec et al., 2018;Lane et al., 2020). Therefore, to help account for an individual's residence time at a given sampling site, we included the Julian day on which sampling occurred as a fixed effect. This method does not account for the temporal and latitudinal stratification of sparrow arrival timing, or withinsite variation in arrival times, which were unknown for individuals.
To account for geographic variation in prey availability, marsh hydrology, and other spatially variable conditions, we initially included latitude and longitude as fixed effects, and included marsh site as a random intercept in all models. When investigating appropriate spatial scales for the analysis (see paragraph below), we discovered that some of our variables were highly correlated at larger spatial scales, as indicated by variation inflation factors (VIF > 3) and correlation coefficients (R 2 > 0.7) using R package "usdm" (Hair et al., 2010;Naimi et al., 2014). After removing latitude, longitude, and percent area of wetlands from the full model, there was no evidence of multicollinearity at any of the spatial scales we evaluated.
To test for correlations between tidal marsh sparrow THg concentrations and landscape characteristics surrounding sampling points, we attempted to define a statistically relevant spatial extent, or "scale of effect," for the independent variables. To accomplish this, we compared linear mixed models of identical structure, but with spatial attributes extracted within different buffer radii (Jackson & Fahrig, 2012). Wiener and Shields (2000) demonstrated that industrially contaminated wetlands produced and exported MeHg 25 km downstream from the Sudbury River, Massachusetts, and Hildebrand et al. (1980) detected noticeable levels of THg in crayfish 130 km downstream from a chemical processing plant. Thus, we selected and compared buffer radii at 1-km increments from 5-10 km and at 5-km increments from 15-50 km to explore a range of potentially relevant scales. The model used to compare scales contained independent variables that would reasonably be influenced by radius selection: developed and forested land cover, in addition to average precipitation-Hg deposition from 2017, within a 5-50 km radius around each sampling point. We defined the scale of effect to be the buffer radius that resulted in the lowest model AIC C value.
To investigate which spatial, environmental, biological, and temporal parameters affect blood THg variation in tidal marsh sparrows, we constructed a set of 32 candidate models, which included a global model with all fixed effects, all possible reduced model combinations, as well as a null model with only site as a random effect. Fixed effects represented a priori hypotheses that tidal marsh sparrow blood THg concentrations are affected by the percentage of forested or developed land surrounding each sampling point, the average precipitation-Hg deposition from 2017 surrounding each point, the species, and the Julian day on which samples were taken. We averaged all models in the candidate set to obtain coefficient estimates for each independent variable using "MuMIn" (Bartoń, 2020). We then estimated 95% confidence intervals and summed Akaike weights (w i ) across all models containing each fixed effect to determine whether each independent variable affected THg levels. If a parameter had a 95% CI that overlapped zero and a summed w i < 0.75, we concluded that our data provided little support for a relationship between the variable and blood THg concentrations of marsh sparrows (Smith et al., 2018).
Models with buffer radii of 25-40 km were statistically indistinguishable (ΔAIC C < 2, Appendix 1), however, a 30 km buffer resulted in the lowest AIC C value, and was thus chosen as the most appropriate scale for describing the landscape. The top seven models in our candidate set, accounting for 57% of total model weight, were statistically indistinguishable (ΔAIC C < 2, Table 2). Species was included as a fixed effect in all fourteen top-performing models, and the model containing species and a random site effect had 5.5 times more weight than the model with the random site effect alone, suggesting interspecific differences in blood THg concentrations (p = 0.0207). Adding other independent variables did not result in better supported models (Table 2); which is consistent with the finding that, after conditional model averaging, all independent variables except for species had estimated 95% confidence intervals that overlapped zero (Table 3).

Discussion
Our conditional model-averaged results indicated that saltmarsh sparrows tended to have higher blood THg concentrations than seaside sparrows both when present at the same marsh sites and across their ranges (Table 1, Table 3, Appendix 2). These results are congruent with previous studies documenting elevated saltmarsh sparrow blood THg concentrations relative to sympatric marsh sparrow species (Shriver et al., 2006;Ruskin et al. in review). However, these results differ from those of Warner et al. (2010), who found no differences between saltmarsh and seaside sparrows in coastal Delaware, though their sample of saltmarsh sparrows was relatively small. We hypothesize that interspecific differences in THg are due to differences in MeHg exposure through diet or foraging preferences. The trophic level at which an organism feeds is one of the principal mechanisms of bioaccumulation and magnification . Some dietary specialization exists between our focal species , potentially leading to differences in the amount of MeHg-rich prey items consumed. For instance, Post and Greenlaw (2006) documented that amphipods (order Amphipoda) and wolf spiders (family Lycosidae) were the first and fourth most important prey item for saltmarsh sparrows, respectively, but ranked much lower for seaside sparrows. Amphipods are scavengers that consume organic matter that originates from aquatic environments (Moore & Francis, 1985;Persson, 1999) and have elevated levels of Hg relative to organisms occupying higher trophic levels (George & Batzer, 2008). In addition, due to their predatory nature and elevated capacity for MeHg biomagnification, spiders pose a potential risk to arachnivorous songbirds (Cristol et al., 2008;Gann et al., 2015). Shriver et al. (2006) found a similar pattern between Nelson's and saltmarsh sparrows breeding in coastal Maine and also attributed differences in Hg to variation in foraging strategies between the taxa. Quantifying MeHg concentrations within food items consumed by tidal marsh sparrows remains an unexplored, yet crucial step in characterizing MeHg exposure for these species, as well as Hg dynamics and biomagnification within saltmarsh estuarine food webs as a whole.
Despite previous statistical differences of Hg exposure within the tidal marsh sparrow clade, collectively saltmarsh, seaside, and Nelson's sparrows, we found these results to be counterintuitive in principle. All of these species are members of the genus Ammospiza: a group of similar-sized, invertivorous, obligate wetland specialists that have overlapping distributions on both the breeding and nonbreeding grounds. Saltmarsh and Nelson's sparrows have previously been considered the same species (Shriver et al., 2020), yet saltmarsh sparrows have repeatedly been shown to have around twice the concentration of blood THg (Shriver et al., 2006;Cristol et al., 2011). These relationships highlight the potential for subtle differences in foraging, habitat use, and perhaps internal anatomy and pharmacokinetics, to affect a bird's exposure or sensitivity to ecotoxicants (Fuchsman et al., 2017;Cristol & Evers, 2020)-even among closely related species. We should use caution when applying the understanding of toxicological risk in one species across taxa, and refrain from making inferences about poorly sampled members of the same order, family, and genus based on presumed similar behavior and ecology.
Defining relevant land cover characteristics at an appropriate spatial scale is challenging. Hg can be transported over large distances via atmospheric and aquatic systems, and coastal wetlands can have differential Hg loading and site-specific biogeochemical processes-all of which can influence MeHg exposure to wildlife. Relevant spatial boundaries likely vary in both size and shape across marshes, depending on marsh characteristics. Despite exploring the effects of geographic scale on model fit, we found no evidence that the proportions of developed or forested land cover in the surrounding landscape were associated with blood THg concentrations of tidal marsh sparrows. Similarly, we found no evidence that average precipitation-Hg deposition was associated. These landscape-level factors were intended to cumulatively account for regional atmospheric Hg deposition and the potential abundance of Hg point sources, such as industrial facilities and landfills, surrounding sampling points. Some of these variables may play an important role in influencing MeHg exposure of tidal marsh sparrows, which might be revealed by more direct measures of their abundance in the landscape.
Several water quality parameters that correlate with increased organismal Hg bioaccumulation and MeHg production in wetlands were also unaccounted for in this study, including low pH and high dissolved organic carbon levels (Wiener et al., 1990;Wren et al., 1991;Scheuhammer & Graham, 1999;Ravichandran, 2004;Hall et al., 2008). High Table 2 Model-selection results examining the effects of spatial, environmental, biological, and temporal parameters on the natural logtransformed blood THg concentrations (µg/g ww) of breeding tidal marsh sparrows in salt marshes from Maine to Virginia, USA, May 29 to July 31, 2018. Site is treated as a random effect in all models salinity sediments may promote Hg demethylation (Compeau & Bartha, 1987) and the frequency of inundations of each marsh could influence methylation rates and MeHg mobilization from anaerobic wetland sediments (Snodgrass et al., 2000;Hall et al., 2008). Freshwater wetlands that are shallow and experience long bouts of periodic drying have higher THg and MeHg concentrations in water, sediment, and fish than those with consistently moist sediments (Caldwell & Canavan, 1998;Snodgrass et al., 2000). While these complex mechanisms have not been studied extensively in saline environments, brackish water marshes that experience longer periods between inundations (ie. a shorter hydroperiod) have more time to dessicate, and the organic sulfur and sulfides in their sediments have more time to oxidize. A larger oxidized sulfur pool stimulates both sulfate-reducing bacteria and Hg methylation, which then elevates MeHg export and bioavailability after the system is reinundated (Devito & Hill, 1999;Eimers et al., 2003;Hall et al., 2008). Future studies seeking to characterize MeHg bioaccumulation in coastal estuaries should obtain a comprehensive understanding of the unique biogeochemical and hydrological dynamics of each marsh site whenever possible. Previous research suggests that up to 62% of saltmarsh sparrows breeding from Maine to New York, as well as >75% of saltmarsh sparrows and >60% of seaside sparrows wintering in North Carolina, are potentially at risk of deleterious MeHg effects on reproduction (Winder, 2012;Lane et al., 2020). A severe limitation in producing precise at-risk population estimates is the lack of toxicity reference values for our focal species. A common practice in ecotoxicological wildlife monitoring is to extrapolate toxicity reference values across species because many species have not been studied (Warner et al., 2010;Lane et al., 2011Lane et al., , 2020Winder, 2012). Researchers are often cautioned against this practice because the potential exists for extrapolation to be severely biased (USEPA, 2007). Therefore, we provide comparative context with caution as we lack any direct evidence of how THg may influence sparrow reproduction. Based on a predictive linear model using blood THg concentrations from Carolina wrens, a non-migratory passerine breeding along two contaminated rivers in Virginia, USA, Jackson et al. (2011) proposed that females with blood THg concentrations of 0.7-1.2 µg/g ww and 1.2-1.7 µg/g ww were likely at risk of a 10-20% and 20-30% reduction in nesting success, respectively (but see Fuchsman et al., 2017). Applying the Jackson et al. (2011) potential risk categories to birds sampled from Maine to Virginia, approximately 22% of saltmarsh and 11% of seaside sparrows may be at risk of a 10-20% reduction in nesting success, while 3% of saltmarsh and 4% of seaside sparrows may be at risk of a 20-30% reduction in nesting success.
Although approximately 21% of the birds sampled from Maine to Virginia had blood THg levels above the proposed 10% effect concentration (EC10), 58% of saltmarsh sparrows and 33% of seaside sparrows sampled in coastal New Jersey were above the EC10. Coastal New Jersey is estimated to support 37% of the global saltmarsh sparrow breeding population, and 22% of the seaside sparrows that breed in the northeastern USA (Wiest et al., 2016). This may warrant further investigation considering that elevated MeHg exposure cases have been documented previously in the northeastern USA (Lane et al., 2011(Lane et al., , 2020Kopec et al., 2018). For saltmarsh sparrows, especially, the high risk of extinction from sea-level rise across their range suggests that any additional stressors that might exacerbate declines are a concern.
Sea-level rise is the principal driver of tidal marsh sparrow population declines (Shriver et al., 2007;Gjerdrum et al., 2008), and could cause the extinction of the saltmarsh sparrow by mid-century (Correll et al., 2017;Field et al., 2017Field et al., , 2018Roberts et al., 2019). Concurrently, some tidal marsh sparrows may face challenges coming from industrial practices in terrestrial systems that expose them and other wildlife to potentially dangerous levels of Hg contamination. For tidal marsh sparrows to persist into the next century, significant conservation efforts are required (Hartley & Weldon,  There was no evidence of multicollinearity, as indicated by VIF values < 3 from the full model a Modeling estimate for seaside sparrow in reference to saltmarsh sparrow 2020). Ensuring that secondary factors such as heavy metal contamination do not compromise the persistence of remaining populations is beneficial to ensuring longterm success-as long as mitigating the effects of sealevel rise on tidal flooding remains the primary conservation priority.
Author contributions CJS II and DNB contributed to the study conception and design. Material preparation and data collection were performed by CJS II, MRR, LMF, DR, SEA, ARK, AMC, WGS, CSE, BO, and DNB. CJS II, GLBD, DNB, and CSE conducted and contributed to analyses. The first draft of the manuscript was written by CJS II and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Compliance with ethical standards
Conflict of interest The authors declare no competing interests.
Ethical approval Sampling was conducted under approved animal care protocols from the State University of New York: College of Environmental Science and Forestry (180201), University of Connecticut (A17-011), and University of Maine (A2018-03-03).
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix
Appendix 1 ΔAIC C values for models with buffer radii from 5-50 km. Graph indicates that 30 km is the best scale of effect for which developed and forested land cover, as well as average precipitation-Hg deposition in 2017, surrounding each focal marsh can best predict the natural logtransformed blood THg concentrations (µg/g ww) of breeding tidal marsh sparrows in salt marshes from Maine to Virginia, USA, May 29 to July 31, 2018 Appendix 2 Regional-and site-level tidal marsh sparrow blood THg concentration comparisons of other published works through time