Introduction

Mercury (Hg) is a neurotoxic pollutant that accumulates in aquatic food webs, making fish unsafe for consumption by humans and wildlife (Chan et al. 2003; Chen et al. 2008). Mercury emitted from coal-fired power plants, mining, and other sources is transported though the atmosphere (Branch 2008) and deposited in remote areas (Selin et al. 2007), where it can be subsequently converted to methyl-Hg. The transfer of methyl-Hg from lake water and sediments to primary producers is highly efficient, and biomagnification results in increases in tissue Hg concentrations at successive trophic levels (Evers et al. 2007). Wildlife at top trophic levels exposed to Hg can experience adverse behavioral, reproductive, and population-level effects (Driscoll et al. 2013; Eagles-Smith et al. 2018). Remote forested regions have been shown to be particularly sensitive to Hg deposition due to enhanced deposition facilitated by the forest canopy, an abundance of wetlands that are hotspots for methylation, and unproductive aquatic ecosystems (Driscoll et al. 2007).

The Clean Air Act of 1990 (https://www.epa.gov/clean-air-act-overview/benefits-and-costs-clean-air-act-1970-1990-retrospective-study) helped regulate Hg emissions to the atmosphere. Additionally, the Mercury and Air Toxics Standards rule (US EPA 2012), which was finalized in 2011, facilitated additional declines in Hg emissions from electric utilities. Anthropogenic Hg emissions declined by 77% from 1990 to 2014 in North America (Zhang et al. 2016) and air Hg concentrations and wet Hg deposition have also decreased across the U.S. and Canada (Weiss-Penzias et al. 2016). In the northeastern U.S., declining regional emissions resulted in declining atmospheric Hg deposition rates from 1992 to 2014 (Gerson and Driscoll 2016; Zhou et al. 2017). Because environmental regulations have been effective in reducing Hg loading, it might be expected that recovery in aquatic ecosystems should follow.

The common loon (Gavia immer), a top-trophic predator in the freshwater food web, has long been recognized as an important bioindicator of surface water Hg pollution in remote areas (Evers et al. 1998). The Hg concentrations in loon blood indicate dietary uptake in the current season in the lakes where loons feed (Evers et al. 1998; Burgess and Meyer 2008), as does Hg concentration in loon eggs (Evers et al. 2003). Examining temporal trends in Hg concentrations in loon blood and eggs should therefore provide insight into the efficacy of regional air quality management. In southern New Hampshire, a 64% decrease in mean Hg concentration in loon blood from 1999 to 2002 was attributed to a 45% decrease in upwind Hg emissions from a coal-fired power plant (Evers et al. 2007). To date, long-term trends in loon Hg concentrations have been reported only from Wisconsin (Fevold et al. 2003), partly due to the logistical challenges associated with capturing individual loons repeatedly over multi-year periods.

In addition to temporal patterns, loon Hg concentrations from multiple lakes can indicate spatial variation in Hg exposure. Loon feather Hg concentrations provide insight into Hg exposure over longer time periods, as feathers are the major excretory pathway for Hg sequestration in loons (Evers et al. 1998). Previous studies have reported spatial variation in loon feather and blood Hg concentrations (Evers et al. 1998), primarily related to spatial patterns in lake chemistry (Schoch et al. 2011). For example, acidic lakes methylate Hg more efficiently (Miskimmin et al. 1992; Barkay et al. 1997; Kelly et al. 2003) and thus often contain elevated Hg concentrations in zooplankton, fish (Chen et al. 2005; Adams et al. 2009; Brown et al. 2010), and loons (Meyer et al. 1995; Yu et al. 2011; Schoch et al. 2011). Acid neutralizing capacity (ANC), dissolved organic carbon (DOC), and other physio-chemical variables have been shown to vary with Hg concentrations in aquatic biota across a spatial gradient (Yu et al. 2011). Examining changes in the spatial pattern in loon Hg concentrations should indicate where recovery is accelerating or where it is delayed.

Biodiversity Research Institute, its successor, the Adirondack Center for Loon Conservation, and their collaborators have been monitoring Hg concentrations in common loons in New York’s Adirondack Park since 1998, an area previously identified as a biological Hg hotspot (Evers et al. 2007). The dominant source of Hg to the Adirondack Park is atmospheric deposition, estimated to be approximately 20 µg m−2 year−1 (Miller et al. 2005; Yu et al. 2013). Declines have been reported in lake Hg concentrations and fluxes (Gerson and Driscoll 2016) and in Hg concentrations in yellow perch (Perca flavescens) (Simonin et al. 2009; Charles Driscoll, unpublished data). However, spatial patterns in loon Hg concentrations have been explored only for the early part of the record (1998 to 2007; Schoch et al. 2014) and temporal trends have not been examined. Thus, our objectives were to identify spatial patterns and temporal trends in loon Hg concentrations in the Adirondacks using a data set spanning 19 years, 116 lakes, and 760 loon samples. Our goal was to determine whether loon Hg concentrations have decreased in response to reductions in regional and national Hg emissions and deposition.

Methods and materials

Study area and lake selection

Study lakes were located within the Adirondack Park of New York State, USA (43°59′N, 74°14′W), an area of 2.4 million ha. The Adirondack Park is the largest public protected area in the contiguous U.S., containing approximately 2,800 lakes, of which 831 are large enough (>10 ha; Parker 1988) to provide breeding habitat for loons. Study lakes were chosen based on accessibility, observations of loons with juveniles, and the availability of water quality data. Samples of loon blood, eggs and feathers were collected from a subset of study lakes every year, with a total of 116 different lakes sampled from 1998 to 2016. Most lakes (90) were sampled at least twice and the maximum number of years sampled was 11 (4 lakes).

Common loon sampling

We captured common loons for a 1- to 2-week period annually from 1998–2016 using night-lighting and playback techniques (Evers 2001). Upon capture, we collected tissue samples following established protocols (Evers et al. 1998, 2003, 2005). Briefly, we collected loon blood samples from the tibiotarsal vein to determine recent Hg exposure, and collected feather samples from adult and juvenile loons with fully emerged feathers to provide an indication of long-term Hg accumulation. Feather samples included two central tail feathers and the second secondary feather from each wing. We also banded adult and juvenile (if they were large enough for an adult-sized band) loons with U.S. Geological Survey aluminum bands and a unique combination of plastic colored bands, enabling identification of individual loons to be made from a distance during field observations to determine reproductive success. In subsequent years, we recorded whether birds had been previously captured, and we collected blood and feather samples from any recaptured loons.

Each breeding season, abandoned loon eggs were opportunistically collected after field staff determined that they were non-viable (i.e., exhibiting a strong odor or abandoned by adult loons). Eggs were processed for Hg following standardized protocols (Evers et al. 2003).

Mercury analysis

Loon blood, feather, and egg samples were submitted to the Animal Health Diagnostics Laboratory, University of Pennsylvania, New Bolton, PA; the Trace Element Research Laboratory, Texas A&M, College Station, Texas; or the Biodiversity Research Institute in Portland, Maine, for analysis of total Hg concentrations. Blood was analyzed as whole blood. Egg samples were freeze dried in a Labconco Lyph Lock 12 freeze dryer and powdered in a Spex Mixer Mill. Blood Hg concentrations are expressed as mg kg−1 wet weight, feather Hg concentrations as mg kg−1 fresh weight, and egg Hg as mg kg−1 wet weight. Egg Hg was not corrected for dehydration; moisture contents averaged 77.1% with a standard deviation of 2.7%. Analyses for methyl-Hg in loon tissues were not conducted because more than 95% of blood Hg in birds occurs in the methyl form (Wolfe et al. 2007).

All tissue samples were analyzed for total Hg concentration either using cold vapor atomic absorption spectroscopy following U.S. Environmental Protection Agency ((USEPA) Method 245.1), or using thermal decomposition and atomic absorption spectroscopy (USEPA Method 7473) with an automated direct Hg analyzer. USEPA approved QA/QC protocols were used for all assays (U.S. EPA 1994; U.S. EPA 2007). Recoveries of certified reference materials (DORM-3, DORM-4, DOLT-4, DOLT-5 and BCR464) were >90% and duplicates were within 10%.

Calculation of a common Hg unit

Our dataset included Hg concentrations from male, female, and juvenile loons, as well as loon eggs, all of which naturally differ in Hg concentration. For some analyses, we used the relationships identified by Evers et al. (2011) and adapted by Schoch et al. (2011) to convert all data to female loon units (FLU):

  • For adult males, \({\mathrm{FLU}} = {\mathrm{e}}^{( - 0.65 + 1.35 \times \ln ({\mathrm{male}}\;{\mathrm{Hg}}\;{\mathrm{concentration}}))}\)

  • For juveniles < 4 weeks old, \({\mathrm{FLU}} = {\mathrm{e}}^{(1.12 + 0.44 \times \ln ({\mathrm{juvenile}}\;{\mathrm{Hg}}\;{\mathrm{concentration}}))}\)

  • For juveniles 4–6 weeks old, \({\mathrm{FLU}} = {\mathrm{e}}^{(1.82 + 0.75 \times \ln ({\mathrm{juvenile}}\;{\mathrm{Hg}}\;{\mathrm{concentration}}))}\)

  • For eggs, \({\mathrm{FLU}} = 0.22 + 1.55 \times {\mathrm{egg}}\;{\mathrm{Hg}}\;{\mathrm{concentration}}\)

These conversions allowed us to combine all Hg concentrations from different tissues in a single common metric, which improved the sample size and thus the statistical power.

Analysis of spatial patterns

Spatial patterns of Hg concentration in blood and feathers of male and female loons from 1998 to 2016 were analyzed using Geographic Information System (ESRI ArcGIS 10.3) software. Digital Elevation Model and National Hydrography Dataset of the Adirondack Park were acquired from the New York State GIS Clearinghouse (https://gis.ny.gov/).

To examine whether spatial variation in loon Hg concentrations could be explained by water chemistry, we used previously collected water chemistry data including total Hg, methyl-Hg, pH, ANC, dissolved organic carbon (DOC), dissolved inorganic carbon (DIC), monomeric aluminum (Alm), and non-labile (organic) monomeric aluminum (Alo). We had water chemistry data from 2003 and 2004 from 44 of the 116 study lakes (Yu et al. 2011). For these 44 lakes, we averaged FLUs from 1998 to 2016, and used a Pearson correlation test on average FLUs and lake chemistry parameters. We also used a Pearson correlation test on average feather Hg concentration from 1998 to 2016 and 20 lake chemistry parameters, for both male and female loons. This approach ignores change over time in both lakes and loons.

To examine whether feather Hg concentrations could be explained by blood Hg concentrations due to a possible depuration during winter remigial molt, we used a Pearson correlation to test for a correlation between average blood and feather Hg concentrations, using lakes as replicates. The test was conducted for both male and female loons with lakes as observations.

Analysis of temporal trends

We examined temporal trends in loon Hg concentrations (as FLUs) across all of our survey lakes using a general linear mixed model. The mixed model used for trend analysis was Yij = x*a + bi + cj + u + eij, where where i and j are index variables for lake and year, Yij is the average log-transformed FLU at lake i in year j, x is a continuous variable for time, a is the slope for temporal trend, b is the coefficient for lake, c is the coefficient for year, u is the intercept, and eij is the residual error. Restricted maximum likelihood was used to quantify the variance component, and a likelihood ratio test was used to determine the p value. Piecewise regression with the NLIN function was used to identify any changes in slope and to identify a breakpoint between two regressions with different slopes. We log-transformed Hg concentration data to meet the assumption of normality of the residuals.

To examine temporal trends by gender, age, and type of sample (blood vs. egg), we applied the mixed model above to Hg concentrations separately for males, females, juveniles, and eggs.

To analyze temporal trends in loon Hg concentration (as FLUs) in individual lakes, we identified 52 lakes that had sampling intensity ≥ 4 years. We used the nonparametric Mann-Kendall (MK) test along with Sen’s method on the annual average FLU. The calculations were performed using the MAKESENS template in Excel (Microsoft Corp., Redmond, WA) which applies the Mann–Kendall test for trends and Sen’s slope estimates (Salmi 2002).

We examined the change in blood Hg concentration (not converted to FLUs) of 29 male and 19 female recaptured loons. A general linear model was used to test the fixed effects of gender, number of years between capture and recapture, and lake pH on change in blood Hg concentration. For loons that were captured three or four times (the maximum number of times captured), we used the average Hg concentration of the two closest years as one endpoint, and the number of years between capture and recapture was the difference between the first and last year of capture. We used a t-test to compare the changes in Hg concentration between loons with initial Hg concentrations greater or less than a threshold to test whether Hg concentrations measured at the time of first capture could predict the subsequent changes in Hg concentrations. We selected a moderate Hg exposure risk level (2 mg kg−1, Evers et al. 2008) as the threshold.

All the statistical tests described above were performed using SAS 9.4 (SAS Institute Inc. 2013). We used an α of 0.10 to reduce the chances of accepting the null hypothesis of no effect, and we report the actual p values so that readers can judge the significance of effects (Amrhein et al. 2019).

Results

Loon Hg concentrations from 1998 to 2016

The concentrations of blood Hg in Adirondack common loons from 1998 to 2016 ranged from 0.48–7.31 mg kg−1 for males (x̅ = 2.30 mg kg−1, n = 153), 0.11–8.98 mg kg−1 for females (x̄ = 2.04 mg kg−1, n = 138), 0.01–0.77 mg kg−1 for juveniles < 4 weeks old (x̄ = 0.24 mg kg−1, n = 82), 0.01–0.86 mg kg−1 for juveniles from 4 to 6 weeks of age (x̄ = 0.24 mg kg−1, n = 64), and 0.01–1.70 mg kg−1 for juveniles > 6 weeks old (x̄ = 0.29 mg kg−1, n = 39). Loon egg Hg concentrations ranged from 0.11–2.65 mg kg−1 (x̄ = 0.78 mg kg−1, n = 326; Table 1).

Table 1 Average concentrations of Hg (mg kg−1) in Adirondack loon blood and eggs from 1998 to 2016, with the number of lakes sampled each year (N) and the standard deviation (SD)
Full size table

Loon feather Hg concentrations ranged from 3.90–147.23 mg kg−1 for males (x̄ = 18.87 mg kg−1, n = 151), and 2.20–35.26 mg kg−1 for females (x̄ = 11.23 mg kg−1, n = 137) from 1998 to 2016 (Table 2). Feather Hg concentrations ranged from 0.12–13.44 mg kg−1 for juveniles < 6 weeks old (x̄ = 6.40 mg kg−1, n = 18) from 2008 to 2016.

Table 2 Average concentrations of Hg (mg kg−1) in Adirondack loon feathers from 1998 to 2016, with the number of lakes sampled each year (N) and the standard deviation (SD)
Full size table

Spatial patterns of loon Hg concentrations

Concentrations of Hg in blood and feathers of male and female loons varied across the Adirondack study area (Fig. 1). Some of the spatial variation in loon blood and egg Hg concentrations was explained by lake chemistry. Higher loon Hg concentrations (using blood and egg data converted to FLUs) were found in lakes with lower pH (r = −0.44, p = 0.002), lower ANC (r = −0.34, p = 0.05), higher Al (r = 0.56, p < 0.001), higher Alo (r = 0.48, p = 0.004), higher water methyl-Hg (r = 0.40, p = 0.03) or higher water total Hg (r = 0.43, p = 0.01).

Fig. 1
figure 1

Spatial patterns of Hg concentrations in blood and feathers of male and female loons in the Adirondack Park from 1998–2016. Colors indicate Hg concentrations (mg/kg)

Full size image

Spatial variation of feather Hg for male and female loons was also partially explained by lake chemistry. Lakes with higher Al (r = 0.31, p = 0.08) had higher Hg concentrations in male loon feathers. Lake with higher Al (r = 0.41, p = 0.03), Alo (r = 0.35, p = 0.06), or methyl-Hg (r = 0.47, p = 0.02) had higher Hg concentrations in female loon feathers.

We also found that lakes containing loons with higher blood Hg concentrations also had higher feather Hg concentrations for both male (r = 0.34, p = 0.01) and female loons (r = 0.52, p < 0.01).

Temporal trends in loon Hg concentrations

Using 760 samples of loon blood and eggs from 116 lakes in the Adirondack Park, we found a significant increase in loon Hg concentrations (as FLU) from 1998 to 2010 (p = 0.04), but no significant change from 2010 to 2016 (p = 0.91) (Fig. 2). From 1998 to 2010, the annual increase in FLU was 5.7% (0.05 mg kg−1 Hg yr−1), while from 2010 to 2016, the FLUs remained stable, with an average of 1.70 mg kg−1. Blood Hg concentration in both male and female loons increased from 1998 to 2010 (males: p = 0.09, n = 153; females: p = 0.02, n = 138), but did not change from 2010 to 2016 (both p ≥ 0.25) (Fig. 3). For juveniles, blood Hg concentration increased from 1998 to 2008 (p = 0.03), and did not change from 2008 to 2016 (p = 0.38). For eggs, Hg concentration increased from 1998 to 2011 (p = 0.01) and did not change from 2011 to 2016 (p = 0.49) (Fig. 3). These concentrations can also be viewed in risk categories from low to extra high (Fig. 4)

Fig. 2
figure 2

Temporal trends in Hg concentrations, as FLUs, in Adirondack loons from 1998 to 2016

Full size image
Fig. 3
figure 3

Temporal trends in Hg concentrations, in the native units, in male, female, and juvenile loons and loon eggs from lakes in the Adirondacks from 1998 to 2016. Arrows indicate the breakpoints in piecewise regression

Full size image
Fig. 4
figure 4

Risk ratios for Hg exposure based on Hg exposure groups: low (0–1 mg kg−1), low-moderate (1–2 mg kg −1), moderate-high (2–3 mg kg−1), high (3–4 mg kg−1) and extra high ( > 4 mg kg−1) for male a and female blood b (Evers et al. 2008, Burgess and Meyer 2008), low (0–0.1 mg kg−1), moderate (0.1–0.3 mg kg−1), high (0.3–0.4 mg kg−1) and extra high (>0.4 mg kg−1) for juveniles < 6 weeks c (Meyer et al. 1998), and low (0–0.5 mg kg−1), moderate (0.5–1.3 mg kg−1), high (1.3–2.0 mg kg−1) and extra high (>2.0 mg kg−1) for eggs d (Evers et al. 2003)

Full size image

In addition to testing for a temporal trend across the entire data set, we characterized change over time within individual lakes; there were 52 lakes with at least 4 years of loon Hg samples. Patterns within these lakes were not consistent (p = 0.51 for a t-test on slopes of FLUs over time). Only eight lakes had statistically significant rates of change, with six lakes showing increases and two lakes showing decreases (Table 3).

Table 3 Temporal trends in loon Hg concentrations (as FLUs) at each lake that was sampled ≥ 4 years, using nonparametric Mann-Kendall with Sen’s slope estimates
Full size table

Changes of Hg concentrations in blood from recaptured loons

Female loons had larger differences in Hg concentrations between their first and last capture than did male loons (p = 0.09 for ANOVA comparing changes in female vs. male loons). Neither the period between capture and recapture (p = 0.45) nor lake pH (p = 0.34) explained changes in Hg concentrations between sampling dates.

Concentration of loon blood Hg measured at the time of first capture was an important factor influencing the subsequent changes in Hg concentration. Loons with a lower initial blood Hg concentration (<2 mg kg−1) tended to have higher Hg concentrations at their subsequent recapture (25 loons had increases and 4 loons had decreases; Fig. 5a). Loons with a higher initial blood Hg concentration (>2 mg kg−1) had lower concentrations upon recapture (Fig. 5b), with only two exceptions. One of these two loons was captured in two different lakes, with the second lake having higher Hg concentrations in zooplankton and fish than the initial lake. The other loon was from a lake with one of the highest total Hg concentrations (3.8 ng L−1) and the lowest pH (5.2) in the water column of the lakes studied.

Fig. 5
figure 5

Temporal changes in Hg concentrations in loon blood from recaptured Adirondack loons with an initial Hg concentration a <2 mg kg−1 and b >2 mg kg−1. Of the 48 recaptured loons, 4 female loons and 6 male loons were captured and recaptured at different lakes (dashed arrows)

Full size image

Discussion

Loon Hg concentrations

In this study of Adirondack common loons, average male blood Hg concentrations (2.30 ± 0.09 mg Hg kg−1) were in the moderate-high Hg risk exposure category for loons (2–3 mg Hg kg−1; Evers et al. 2008) and average female loon blood Hg levels (2.04 ± 0.11 mg Hg kg−1) were at the low end of the moderate-high Hg risk exposure category. Lower blood Hg concentrations of female loons compared to males can be explained by the excretion of Hg into eggs during the breeding season and the ingestion of smaller prey that contain lower quantities of Hg. The Hg concentration in loon eggs averaged 0.78 ± 0.05 mg kg−1 from 1998 to 2016, which is categorized as moderate risk for loon exposure (0.5–1.3 mg Hg kg−1; Evers et al. 2003).

Blood Hg is indicative of recent dietary exposure, whereas feather Hg is reflective of longer-term exposure (Evers et al. 1998; Burgess and Meyer 2008). Feather Hg concentrations averaged 18.2 ± 1.2 mg kg−1 for males and 10.8 ± 0.5 mg kg−1 for females, with only 3% of male and 0% of female loons exceeding the adverse effects threshold of 40 mg Hg kg−1 (Evers et al. 2008). Thus, fewer of the loons studied exceeded the adverse effects category based on feather Hg concentrations, compared to the moderate risk of Hg exposure based on blood and egg Hg concentrations. This might suggest that the summer diet of loons in the Adirondacks contains more Hg than does their diet in their wintering areas. However, feather and blood Hg concentrations were correlated, as has been reported elsewhere in North America using data from 1992 to 1996 (Evers et al. 1998), which suggests that differences in winter diet are not very important to Hg accumulation in feathers.

We note that loon capture success is skewed toward those birds who are raising young. Loons with the highest Hg levels are less likely to hatch chicks successfully (Evers et al. 2008), and thus are less likely to be included in this dataset. This omission suggests that the Hg levels we report, based on captured loons, likely underestimate Hg exposure and its impacts on the health of the overall loon population. However, since this bias is consistent over time, it does not preclude the detection of long-term trends in Hg in the loon population in the Adirondack Park.

Spatial pattern in loon Hg

The relationship between loon Hg concentration and lake chemistry (e.g., pH, ANC) was similar to that reported for the early years of observation (Schoch et al. 2014). In lakes with lower pH, methylation rates are hypothesized to be higher because there is less dissolved organic carbon to complex Hg2+, making Hg2+ more available for methylation and more sulfate-reducing bacteria that enhance the rate of conversion of elemental Hg to MeHg (Miskimmin et al. 1992; Barkay et al. 1997; Kelly et al. 2003). Thus, more acidic lakes could have higher Hg methylation resulting in higher bioaccumulation and biomagnification rates (Steffan et al. 1988; Kelly et al. 2003; Yu et al. 2011). Note that Yu et al. (2011) found no relation between methyl-Hg concentrations with pH in the water column of Adirondack lakes, but observed increases in the methyl-Hg concentrations in zooplankton of those same lakes, and this pattern of elevated Hg with lower pH persisted up the aquatic food chain. This pattern led these investigators to suggest that bioaccumulation of methyl-Hg at the base of the food chain increased with decreases in pH, and this factor controlled the spatial relationship of fish and loon Hg with pH and ANC. Other landscape characteristics including elevation, watershed area (Yu et al. 2011) and land cover (Kramar et al. 2005) might also affect the spatial distribution of loon Hg concentrations.

Temporal trends in loon Hg

We expected to observe declines in Hg concentrations in Adirondack loons over the course of this long-term study, reflecting the reported declines in atmospheric dry (Zhou et al. 2017) and wet (Mao et al. 2017) Hg deposition, lake water Hg concentrations and fluxes (Gerson and Driscoll 2016; Millard et al. 2019), and Hg concentrations in yellow perch (Simonin et al. 2009; Charles Driscoll, unpublished data). However, over the full dataset of 116 lakes, we found that loon Hg concentrations (based on FLUs) increased from 1998 to 2010 and then remained stable from 2011–2016. Juvenile blood Hg concentration stabilized earlier than adults. Adult loons consume older and larger fish that have accumulated more Hg (Driscoll et al. 1994), while juvenile loons consume younger and smaller fish that benefit sooner from lower Hg concentrations in the environment (Charles Driscoll, unpublished data). The delay in recovery of loon Hg compared to fish Hg is likely be associated with the higher trophic position of loons in the aquatic ecosystem, as in the case of bald eagles in New York State, which had high Hg in 2004–2006 (DeSorbo et al. 2019).

Even recaptured adult loons in the Adirondacks did not show a consistent temporal change in Hg concentration (Fig. 5). The fact that recaptured loons showed changes towards a central tendency might be due to random variation, as a random high value would be unlikely to be followed by a higher value. Near Little Rock Lake, Wisconsin, where atmospheric Hg deposition declined 10% annually from 1994 to 2000 (Hrabik and Watras 2002), recaptured adult loons showed detectable improvement (5% yr−1) in blood Hg from 1992 to 2000 (Fevold et al. 2003). The modest reductions of atmospheric Hg deposition in our study area (1.6% annually; Zhou et al. 2017) have yet to produce consistently decreasing loon Hg concentrations either for the entire data set or for recaptured loons.

Implications for loon recovery

It is difficult to predict the recovery of loon Hg concentrations. Our original hypothesis was that loon Hg in the Adirondack Park of New York would decrease in response to decreases in regional Hg emissions and deposition. Although the rate of increase in loon Hg has attenuated in recent years, there is no evidence of recovery from elevated exposure. In addition to decreases in Hg deposition, the Adirondack Park is experiencing decreases in acid deposition (Driscoll et al. 2016) and changing climate (USGCRP 2018), both of which can alter Hg dynamics in complex ways. Decreases in acid deposition have resulted in decreases in sulfate and increases in pH and ANC, which should decrease Hg methylation and bioaccumulation. However, decreases in acid deposition have also coincided with increases in dissolved organic matter (Driscoll et al. 2016), which could drive increases in Hg methylation and have been shown to correspond with increases in fish Hg in Scandinavia (Hongve et al. 2012). In the Adirondacks and the northeastern U.S., air temperature and precipitation have also increased (USGCRP 2018). These changes have altered watershed hydrology through decreases in snowpack accumulation and duration, earlier snowmelt, and changing stratification of lakes (Campbell et al. 2011). Moreover, climate is becoming more variable with more extended dry and wet periods and extreme runoff events. Increases in temperature will tend to increase mineralization and methylation rates. Increases in precipitation and runoff will tend to increase Hg transport from watersheds to aquatic ecosystems. Increases in wet-dry cycles and stratification will enhance methylation (Evers et al. 2007). In total, these climatic effects will likely increase forest watershed sensitivity to atmospheric Hg deposition. Although these hydrologic and biogeochemical effects have been, and undoubtedly will continue to be, profound, the greatest uncertainty in changes in Hg dynamics may be those manifested though changes in biological communities. Changes in acid deposition, climate, species invasion, and other drivers are changing aquatic biological communities in complex ways. Changes in species distribution, biomass, and food chains can have profound and unpredictable effects on Hg concentrations in apex predators.

In conclusion, although Hg concentrations in loon blood, eggs, and feathers have yet to show recovery, the lack of increase in the last decade suggests that loons have benefited from controls on Hg emissions. It will be difficult to predict the ultimate recovery of loon populations because Hg dynamics are influenced by a host of ecosystem biophysical processes, as described above. Given this complexity, it will be important to continue long-term measurements of loon Hg with careful attention to monitoring design (Yang et al. 2019). Interpretation could be facilitated if loon Hg measurements were accompanied by fish and water chemistry observations.