Accelerated increase in mercury contamination in north Atlantic mesopelagic food chains as indicated by time series of seabird feathers

Monitor species and monitoring units

The monitor species used here (Table 1) were selected on the basis of a priori information on the ecology of the local seabird assemblage. Although they all feed predominantly on fish, ecological segregation leads to dietary specialization on epipelagic or mesopelagic resources [23, 24] ensuring vertical integration of mercury contamination. Species and sites sampled were: epipelagic monitors—Cory's shearwater Calonectris diomedea borealis (Azores, Madeira, Salvages), little shearwater Puffinus assimilis baroli (Azores, Salvages), common tern Sterna hirundo (Azores); mesopelagic monitors—Bulwer's petrel Bulweria bulwerii (Madeira, Salvages), Madeiran storm petrel Oceanodroma castro (Madeira, Salvages). The main taxa in the diet of each seabird species are [24]: for Cory's shearwater, trumpetfish Macroramphosus scolopax, Atlantic saury Scomberesox saurus, boarfish Capros aper; for common tern, trumpetfish; for Bulwer's petrel and Madeiran storm petrel, lanternfish, especially Myctophum punctatum and Electrona rissoi.

Adult ventral feathers were used as monitoring units. Contemporary samples (post-1990) were obtained from live birds at breeding colonies. Historical samples (pre-1970) were obtained from dated preserved study skins of birds collected at the same colonies and held in museum collections as indicated in the Acknowledgement. Contour feathers are the most representative for estimating whole-bird mercury content [25], and up to 12 ventral feathers were collected from each individual by plucking from live birds or cutting at the basis with scissors from study skins and placed in polyethylene bags prior to analysis (numbers of birds sampled by species and time period are given in Table 1). An assessment of the effect of two washing regimes on feather mercury concentrations showed identical concentrations in unwashed and washed samples [26] and only four historical feather samples with major dust-surface contamination, likely to alter feather weight, were subjected to a chloroform/acetone washing regime [27].

Determinations of total and organic mercury

Contemporary feather samples were analyzed for total mercury. Subsamples of 0.02 to 0.1 g were subjected to a wet mineralization digestion in sulfuric acid and potassium permanganate as described in detail elsewhere [28], prior to total mercury determination by cold vapor atomic absorption spectrophotometry (CVAAS) with a Bacharach mercury analyzer system Coleman 50B. The limit of detection of the method, taken as twice the standard deviation of triplicate analysis at blank concentrations [29], was 10 ng, equivalent to 0.01 μg/g for a 1-g sample. Within- and between-laboratory quality control procedures were employed throughout the study period. Accuracy (expressed as relative error) was within 10% and monitored with standards of inorganic mercury and reference materials (NRCC dogfish muscle DORM 1: measured = 0.829 ± 0.013 SE [n = 9]; certified = 0.798 [±95% confidence limits = 0.074]) and by participation in the intercomparison program for mercury in human hair undertaken by Health and Welfare Canada. Precision (or reproducibility; expressed as coefficient of variation) within and between batch was within the usual 10% for total mercury determinations in biological samples [29], and it was independent of the mercury concentration in the feather sample over a wide range of levels (0.1–30 μg/g; product-moment correlation, r = 0.14, p = 0.14, n = 110). Interference on sensitivity due to matrix and pretreatment were assessed by the method of standard additions before the wet mineralization digestion, and the mean recovery of added inorganic mercury was 99.8% (SE = 5.6, n = 8).

Historical samples were analyzed for organic mercury to overcome potential postmortem contamination of study skins with inorganic mercury used as preservative [27, 30]. Mercury in seabird feathers is entirely organic [27] and this allows a direct comparison with total mercury concentrations determined in uncontaminated (with preservative) contemporary feather samples. That was confirmed for the monitor species by replicate determinations of organic and total mercury concentrations (range: 0.9–22.8 μg/g) in contemporary samples, which yielded a mean organic fraction of 101.0% (SE = 4.1, n = 10). Subsamples of 0.04 to 2 g were subjected to an extraction of methylmercury into toluene, as methylmercury bromide, followed by stripping of methylmercury from toluene into thiosulfate as described in detail elsewhere [26, 27]. Methylmercury in the extracted samples was subjected to the wet mineralization digestion used for contemporary samples (except for a reduction to 1 h digestion in sulfuric acid) prior to determination as total mercury by CVAAS. The efficiency of extraction was 0.910, and concentrations were corrected accordingly. Accuracy and precision of organic mercury determinations were within 10% and monitored throughout the study with standard reference materials and standards of methylmercury chloride. The mean organic mercury in NRCC dogfish muscle DORM 1 was 0.770 ± 0.040 μg/g (SE, n = 3) compared with the certified value of 0.731 (±95% confidence interval = 0.060); the mean organic mercury in IAEA tuna fish RM 350 was 3.90 ±0.14 μg/g (SE, n = 3, CV = 6.2%) compared with a value of 3.96 ± 0.12 μg/g (SE, n = 5) given by Horvat [31]. The mean coefficient of variation for replicate standard solution samples with 0.5 to 1 μg of mercury as methylmercury chloride was 4.1 ± 1.0% (SE, n = 9).

Because consistent weights of dried oven feather samples were difficult to obtain due to rapid reabsorption of moisture, feather weights were taken at ambient laboratory temperature, and all feather mercury concentrations are expressed in micrograms per gram on a fresh weight basis (μg/g fw).

Data analysis

Data were tested for goodness of fit to a normal distribution using the Kolmogorov-Smirnov one-sample test and the requirements of homogeneity of variances using the Levene test, prior to statistical analysis. Parametric and nonparametric statistics were employed as appropriate, and analysis followed standard statistical procedures [32].

Validity and assumptions

The rationale for using seabird feathers as a monitor for mercury is well established [33, 34]. Briefly, seabirds excrete dietary methylmercury in a linear dose-dependent fashion into the plumage by endogenous incorporation during feather growth [26, 35], and mercury is stably bound to the feather keratin [36, 37] and thus can be measured in feathers from museum study skins [8, 30, 37].

Though the trends observed here are clear and well supported, their representativeness relies on the validity of the assumption that the diet of monitor species has remained constant over the study period. This is reasonable because the species selected have narrow and relatively inflexible diets [26] and ecological processes underlying segregation and partitioning of epipelagic and mesopelagic resources should remain unchanged [23, 26]. Furthermore, increase rates are consistent between species feeding on different prey within the same vertical compartment (Table 1). Field and experimental evidence indicate that mercury intake during the breeding period is largely responsible for concentrations in contour feathers of these monitor species [26], and at the very least, trends may be taken as reflecting contamination over the wide north-south Atlantic midlatitude sector where the migratory monitors spend the annual cycle [24].

Trends

The long-term increase of mercury levels near the apex of the epipelagic food web observed here (1.1–1.9%/year) is consistent with a predicted threefold increase of global mercury concentrations in atmosphere and surface oceans since pre-industrial times (i.e., 1.3%/year over 150 years) due to anthropogenic inputs [2] and increase rates of 1.2 to 1.5%/year between 1977 and 1990 in the global background of atmospheric mercury measured over the Atlantic Ocean [18]. The equality of increases in anthropogenic atmospheric inputs of mercury and levels in epipelagic organisms is presumably due to the rapid equilibrium of mercury between the atmosphere and the surface ocean [5, 38]. The trends observed here for the epipelagic oceanic environment are consistent with peat bogs [7, 9], lake sediments [1-12], and seabird feathers [8], all recording two- to fivefold increases since the beginning of the 19th century in midlatitudes of the Northern Hemisphere. Slight increases in the subAntarctic region between pre-1950 and post-1950 were also indicated by seabird feathers [14].

The long-term increase of mercury contamination near the apex of the mesopelagic food web observed here (3.5–4.8%/year) represents a threefold magnification of the anthropogenic-derived pulse of mercury comparative to analogous rates for the epipelagic food web, atmosphere, and surface ocean. Explanations for this difference must be sought in the particular biogeochemistry of mercury in low-oxygen subthermocline seawater, which results in net production of highly bioaccumulative methylmercury through microbial-mediated methylation of reactive mercury supplied by scavenging of particulate mercury from the mixed layer [21]. Assuming linearity in the food chain transfer of methylmercury, the mesopelagic trend indicates that methylmercury production and bioaccumulation grew faster at depth than at the surface, where photodegradation [22] and loss due to gas exchange result in no detectable accumulation of methylmercury [21]. Concurrently, given that methylation is substrate limited [21], the faster historical increase of mercury in depth may reflect long-term increase in fluxes of particulate mercury from the mixed-layer to subthermocline waters. Indeed, upwelling intensification due to climate change enhanced primary productivity [39] and may have increased both oceanic evasion of elemental mercury and scavenging of particulate matter to subthermocline waters for subsequent methylation. There is clearly considerable scope for more research in this topic regarding conflicting views on the role of mixing into the ocean interior as a sink of anthropogenic-derived mercury (see Mason et al. [2] versus Hundson et al. [6].

Historical increases do not appear to have been linear in the epipelagic and mesopelagic food chains due to enhanced increases at the turn of the century (cf. Table 2). This pattern is remarkably consistent with variations in rates of atmospheric mercury deposition in remote midcontinental North America in the same period [7]. Such coincidence may be accounted for by the peak use and loss of mercury to the atmosphere from gold and silver mines in 1850 to 1900 [3] and some slowness of mixing causing a slight time lag. Apparent declines in atmospheric mercury deposition post-1960 recorded in midcontinental North American peat bog presumably reflect regional-scale changes in emissions [7] and were not recorded here in seabird feathers.

Implications

The trends presented above must be perceived under the extreme role of atmospheric processes on mercury environmental cycling. The flux of mercury from the atmosphere at any location on the Earth's surface is a sum of contributions from the global cycle, and regional and local sources [1]. If the only significant mode of deposition of atmospheric mercury is associated with the global cycle, then mercury distribution should be relatively uniform on a regional basis [13]. This is true for the contemporary levels of mercury contamination in the epipelagic and mesopelagic environments of the study area [26]. Because the main regional-scale mercury-related process (fluxes from volcanism and occurrence of Mediterranean water below 1,000 m) were steady over the study period, the historical increases in mercury concentrations in seabirds feeding near the apex of food chains reported here should reflect the impact of the perturbed global cycle of mercury. This provides an empirical linkage between increasing accumulation of methylmercury in aquatic organisms and the anthropogenic influence in the global mercury cycle. Large increases, especially in mesopelagic organisms, are of concern because of the current public-health problem resulting from widespread incidence of elevated levels of methylmercury in fish [40] and the increasing importance of deep-sea resources as a source of protein for humans. Besides, potential interactions of environmental change, as global warming and increases in oxidants and particles in the atmosphere, on the global mercury cycle should not be neglected [6]. The uncertainties concerning the outcome of anthropogenic-derived global processes in current levels of mercury contamination beg for an appraisal of long-term trends at remote locations. In this respect, and given that mercury cycling in marine ecosystems includes many biologically mediated processes [19-21], seabird study skins present good prospects for a comprehensive appraisal of the ecological hazards of global pollution by mercury into epipelagic and mesopelagic food webs.