Spatial and Temporal Trends of Gaseous Elemental Mercury over a Highly Impacted Coastal Environment (Northern Adriatic, Italy)

: Mercury (Hg) is a global pollutant, being highly persistent in the atmosphere, in particular gaseous elemental mercury (GEM), which can easily be emitted and then transported over long distances. In the Gulf of Trieste (northern Adriatic Sea, Italy), contamination by Hg is well characterised but little is known regarding the concentrations, sources and fate of GEM in the atmosphere. In this work, discrete measurements of GEM were recorded from several sites at di ﬀ erent times of the year. The database is consistent with temporal night-day variations monitored using a continuous real-time device. The meteorological conditions were collected as ancillary parameters. GEM levels varied from < LOD (2.0 ng m − 3 ) to 48.5 ng m − 3 (mean 2.7 ng m − 3 ), with no signiﬁcant di ﬀ erences found among sites. A clear daily pattern emerged, with maximum values reached just after sunset. Air temperature, relative humidity, wind speed and direction were identiﬁed as the main micrometeorological factors inﬂuencing both the spatial and temporal variation of GEM. Our results show that average atmospheric GEM values are higher than the natural background of the Northern Hemisphere and will be useful in future selection regarding the most suitable sites to monitor atmospheric Hg depositions and ﬂuxes from soil and water.


Introduction
Mercury (Hg) is a well-known global pollutant which is a cause for serious concern regarding ecosystems and human health due to its persistence, toxicity and potential bioaccumulation/biomagnification [1]. The atmosphere represents the main redistribution pathway of Hg mobilised from lithospheric reservoirs and often atmospheric depositions constitute the main source of this element for terrestrial and aquatic ecosystems, even in areas far from points of emission [2]. Once released into the environment, Hg can adversely affect the physiology of many living organisms [3], particularly in its organic form, monomethylmercury (MMHg), which is more toxic and easily biomagnified within the trophic chain due to its lipophilic nature [4]. The consumption of fish contaminated by MMHg is considered the main Hg exposure route for humans [5], whereas inhalation of inorganic Hg vapours occurs mainly via dental amalgams and in certain occupations [6]. Adverse and pristine environments. To better understand temporal patterns and possible dispersion of this contaminant in the atmospheric compartment, continuous night-day measurements were also conducted. Furthermore, GEM values were correlated with the main meteorological parameters in order to elucidate the factors influencing the behaviour of the element in this area.

Study Area
The Gulf of Trieste extends from the Tagliamento River mouth to Savudrija/Punta Salvore (Croatia) and covers an area of approximately 550 km 2 in the north easternmost part of the Adriatic Sea. This coastal ecosystem hosts a Marine Protected Area (Miramare) where, due to a no-entry and a buffer zone, conservation and refuge for overexploited species are guaranteed [33]. On the other hand, this densely populated area is host to several varieties of industry. The coastline is largely exploited for tourism (approximately 60%) with Grado and Lignano Sabbiadoro settlements which significantly increase the resident population during summer (from 8000 to 80,000 and 6000 to 250,000, respectively). The most industrialised areas are represented by the cities of Trieste and Monfalcone (210,000 and 30,000 inhabitants, respectively) with their harbours, and also by the nearby city of Koper (Slovenia); overall these industrial sites cover about 30% of the territory (i.e., iron-steel factory, coal-fired power plant, oil pipeline, ship traffic, considerable vehicular emissions and so on). Finally, mussel farming (Mytilus galloprovincialis) and aquaculture using suspended cages (Dicenthrarcus labrax, Sparus aurata and Mugil cephalus) take place in the easternmost sector of the Gulf.
Several pollutants (i.e., trace elements and POPs) represent a concern for the area as highlighted in numerous previous studies [34][35][36][37][38][39][40][41][42][43][44]. However, mercury contamination due to secular inland mining exploitation from Idrija (NW Slovenia) shows a wide diffusion in sediments, waters and soils [32,[45][46][47].  Figure 1). The Lumex Ra-915M relies on atomic absorption spectrometry (AAS): the instrument has a multipath analytical cell and Zeeman background correction provides both high sensitivity and minimal interference: the accuracy of the method is 20% [48]. The dynamic range covers four orders of magnitude (2-25,000 ng m −3 ), and the detection limit is governed by shot noise and equals 2.0 ng m −3 (average measuring time 5 s) and 0.3 ng m −3 (average measuring time 30 s). A complete calibration is done by a Lumex technician each year, while a calibration check of the instrument is performed prior to taking measurements by means of an internal accessory cell containing a known amount of Hg. During field work, real-time measurements were visualised on a digital display and stored in an internal data logger. Subsequently, the data were recovered by RAPID 1.00.442 software.

Gaseous Elemental Mercury (GEM) Measurements
The data were acquired over a variable time range, from a few hours to a maximum of 14 days, depending on sites and local conditions, and sampling rate (from 1 to 10 s). Moreover, values below 2 ng m −3 were treated with the medium bound approach, thus set to 1 2 LOD (1 ng m −3 ). All GEM time-series were resampled to a fixed sampling frequency of 1 h in order to easily process the dataset and check the relationships with some meteorological parameters provided by the "OMNIA" database from the meteorological observatory of the regional environmental protection agency (OSMER-ARPA FVG, Visco, Italy).
Univariate statistics were computed hourly with Microsoft Excel spreadsheets and then processed in Python. The Pandas library [49] was employed for time-series downsampling and re-sampling and for calculating matrices of Spearman nonparametric correlation coefficient. The wind rose chart

GEM Level and Distribution
The univariate descriptive statistic of the surveyed sites is reported in Table 1. As previously mentioned, the pristine environment was chosen for its suitability for comparison with other impacted sites at a considerable distance from Hg-bearing sediments and soils affected by the mining activity of the Idrija Hg mine and from other potential emissions. These sites (BAS and PIR) were previously investigated and showed GEM contents that were on average lower than those found in the other sites (1.02 ± 0.32 and 1.88 ± 1.07 ng m -3 , respectively) [51,52]. It is notable that MON, which is located in an urban area, showed lower values (1.19 ± 0.40 ng m -3 ) considering the hourly average, but the maximum reached up to 17.28 ng m -3 . Overall, mean hourly values ranged between 1.20 and 3.57 ng m -3 , comparable or slightly higher than the natural background levels estimated for the Northern Hemisphere (1.5-1.7 ng m -3 ) [22] and for the Mediterranean area (1.75-1.80 ng m -3 ) [53]. The highest concentrations were found at the Hg-contaminated FOS site (48.5 ng m -3 ). This area was originally part of the Isonzo River delta and has been affected by human activity since 1800, in particular by land reclamation (several dewatering plants) for intensive agriculture which took place after 1920: the soils are heavily contaminated (up to 40 mg kg -1 ) [54].
These results are comparable to those found in other Hg-contaminated sites. Taking as an example, Muramoto et al [55] recorded GEM ranging from 1.89 to 2.23 ng m -3 (max = 6.11 ng m -3 ) in the Minamata Bay (Japan). In the Mediterranean area Bagnato et al 2013 [56] and Gibicar et al 2009 [57] found from 1.5 ± 0.4 to 2.1 ± 0.98 and from 2.8 to 8.7 ng m -3 in the Augusta Bay (Sicily, Italy) and Rosignano (Tuscany, Italy), respectively, and the GEM values were higher during the summer period.
Due to the large amount of the dataset, the comparison of GEM concentrations was depicted by means of a box and whisker plot representation. Briefly, the median is represented by the horizontal bold line within the box, 25th and 75th percentiles are at the top and bottom. In this case, the presence

GEM Level and Distribution
The univariate descriptive statistic of the surveyed sites is reported in Table 1. As previously mentioned, the pristine environment was chosen for its suitability for comparison with other impacted sites at a considerable distance from Hg-bearing sediments and soils affected by the mining activity of the Idrija Hg mine and from other potential emissions. These sites (BAS and PIR) were previously investigated and showed GEM contents that were on average lower than those found in the other sites (1.02 ± 0.32 and 1.88 ± 1.07 ng m −3 , respectively) [51,52]. It is notable that MON, which is located in an urban area, showed lower values (1.19 ± 0.40 ng m −3 ) considering the hourly average, but the maximum reached up to 17.28 ng m −3 . Overall, mean hourly values ranged between 1.20 and 3.57 ng m −3 , comparable or slightly higher than the natural background levels estimated for the Northern Hemisphere (1.5-1.7 ng m −3 ) [22] and for the Mediterranean area (1.75-1.80 ng m −3 ) [53]. The highest concentrations were found at the Hg-contaminated FOS site (48.5 ng m −3 ). This area was originally part of the Isonzo River delta and has been affected by human activity since 1800, in particular by land reclamation (several dewatering plants) for intensive agriculture which took place after 1920: the soils are heavily contaminated (up to 40 mg kg −1 ) [54]. Table 1. Basic statistics of gaseous elemental mercury (GEM) dataset calculated on raw data (not resampled) and grouped by location. (*) Data published by [52] and ( §) [51]. These results are comparable to those found in other Hg-contaminated sites. Taking as an example, Muramoto et al. [55] recorded GEM ranging from 1.89 to 2.23 ng m −3 (max = 6.11 ng m −3 ) in the Minamata Bay (Japan). In the Mediterranean area Bagnato et al. 2013 [56] and Gibicar et al. 2009 [57] found from 1.5 ± 0.4 to 2.1 ± 0.98 and from 2.8 to 8.7 ng m −3 in the Augusta Bay (Sicily, Italy) and Rosignano (Tuscany, Italy), respectively, and the GEM values were higher during the summer period.
Due to the large amount of the dataset, the comparison of GEM concentrations was depicted by means of a box and whisker plot representation. Briefly, the median is represented by the horizontal bold line within the box, 25th and 75th percentiles are at the top and bottom. In this case, the presence of outliers is shown as circles if values are 1.5 times out of the box and as stars for values which are three times out of the box ( Figure 2). Site FOS showed several outliers such as TS, thus suggesting that there are nearby sources of GEM (i.e., contaminated soils, urban activities) that, in the absence of dilution conditions, can be detected by means of continuous monitoring. On the other hand, pristine areas do not behave like active GEM sources and show data comparable to those observed in other areas (ranging from 0.3 to 10 ng m −3 ) with no outliers [31]. The significant difference between sample medians was confirmed by the Kruskal-Wallis test (p = same = 2.18 × 10 −71 ), which is a nonparametric method for testing if there are statistically significant differences between two or more groups of an independent variable. The calculated ratio of urban to pristine site concentrations for GEM, on average, usually range from 1 to 1.8 [27], and the ratio found in this study, which is 1.52, falls within the ranges from other pristine-urban site studies.   One of the main concerns arising from GEM levels is the potential risk for local inhabitants via an inhalation pathway. According to the guidelines and safety regulations reported in Oyarzun et al. [58], it can be asserted that no risk is present for local inhabitants (WHO guideline fixed at 1000 ng m −3 ; MRL for chronic inhalation 200 ng m −3 , US OSHA and ATSDR).

GEM Time-Series
One of the goals of this study was to investigate the occurrence of daily cycles of GEM concentrations as previously reported for other contaminated sites [31]. In this context, some particular time-series, the monitoring campaigns in Fossalon (FOS3_1 and FOS1_2, 2013 and 2015), Grado (GRA2_1, 2015) and Trieste (TS2_3, 2015), are reported in Figure 3.

GEM Time-Series
One of the goals of this study was to investigate the occurrence of daily cycles of GEM concentrations as previously reported for other contaminated sites [31]. In this context, some particular time-series, the monitoring campaigns in Fossalon (FOS3_1 and FOS1_2, 2013 and 2015), Grado (GRA2_1, 2015) and Trieste (TS2_3, 2015), are reported in Figure 3. The time-series at FOS3_1 showed the presence of anomalous brief high-amplitude peaks that occurred at sunset and during the night (Figure 3a). As previously mentioned, values up to 48.46 ng m -3 were reached (Table 1). In other cases, the peaks were less high in amplitude but more frequent (Figure 3b,c). It can be hypothesised that the contaminated soils are characterised by an emission capacity, a continuous source of GEM that is generally higher during the day because of incident solar radiation. However, during sunrise the temperature decreases and the sea breeze drops; before the opposite land breeze occurs, a temporary atmospheric stable condition is created so that atmospheric dilution and mixing are not favoured, and GEM can concentrate in the lower layers of the atmosphere [30,31].
The occurrence of events of high atmospheric GEM levels is usually associated with conditions of stagnant air and low atmospheric mixing as shown in Figure 4 [17,59]. The time-series at FOS3_1 showed the presence of anomalous brief high-amplitude peaks that occurred at sunset and during the night (Figure 3a). As previously mentioned, values up to 48.46 ng m −3 were reached (Table 1). In other cases, the peaks were less high in amplitude but more frequent (Figure 3b,c). It can be hypothesised that the contaminated soils are characterised by an emission capacity, a continuous source of GEM that is generally higher during the day because of incident solar radiation. However, during sunrise the temperature decreases and the sea breeze drops; Atmosphere 2020, 11, 935 7 of 12 before the opposite land breeze occurs, a temporary atmospheric stable condition is created so that atmospheric dilution and mixing are not favoured, and GEM can concentrate in the lower layers of the atmosphere [30,31].
The occurrence of events of high atmospheric GEM levels is usually associated with conditions of stagnant air and low atmospheric mixing as shown in Figure 4 [17,59].
the lower atmosphere. Taking as an example the measurements conducted at Grado, a significant correlation of GEM with air relative humidity (r = 0.72; Table 2) occurred at GRA2_1 during six days of continuous monitoring. Moreover, GEM increases with temperature, and the behaviour of GEM differs between the Hg-contaminated soil of the Isonzo River Plain and lagoon, and the city of Trieste. In the first case, GEM showed negative correlations with solar radiation, whereas it showed the opposite trend in the urban area of Trieste. This difference is likely a consequence of the distribution of breeze strength between the considered sites; the Isonzo alluvial plain is characterised by a diurnal sea breeze stronger than a nocturnal land breeze, whereas the urban area of Trieste shows an opposite pattern, with higher speeds during the night. In this urban site, no relevant sources of contamination are known, thus GEM could largely be carried from contaminated areas by wind during the diurnal sea breeze. The influence of sea breezes on atmospheric GEM levels in urban areas is well documented [17,25,61,64].  The wind rose chart shows that when GEM is high (up to an hourly mean value of 8 ng m -3 ) with winds prevalently blowing from W and NW, in the direction of the highly contaminated Fossalon plain, it (GEM) could reach the city of Trieste via the seawater surface of the Gulf ( Figure 5). It was hypothesised that a variation in wind speed is not sufficient to cause a dilution process responsible During diurnal hours, sea breezes can dilute atmospheric GEM, resulting in lower concentrations. However, considering that the correlation between wind speed and GEM (Table 2) is below 0.3, we hypothesise that lower levels of this contaminant during the day at the Fossalon site could also be caused by an enhanced photooxidation to RGM, which can easily be removed from the atmosphere through atmospheric depositions [60]. In coastal areas, air coming from the sea driven by breezes is usually rich in oxidants such as halogen radicals (e.g., bromine), which are thought to favour the oxidation of GEM in the presence of solar radiation [61,62] and in the marine boundary layer the reaction of GEM with bromine is considered the predominant pathway of oxidation for this species [63]. On the contrary, the GEM content in the urban area of Trieste appears to be influenced by solar radiation (Figure 3d).
To assess the correlation between GEM and micrometeorological parameters, Spearman's correlation coefficients were calculated for any time-series longer than 24 h (Table 2). Generally, we found that GEM reached its highest values when air relative humidity raised, likely because they have similar behaviour with respect to wind dilution. High relative air humidity is usually associated with stagnant atmospheric conditions, which as stated above favours the accumulation of GEM in the lower atmosphere. Taking as an example the measurements conducted at Grado, a significant correlation of GEM with air relative humidity (r = 0.72; Table 2) occurred at GRA2_1 during six days of continuous monitoring. Moreover, GEM increases with temperature, and the behaviour of GEM differs between the Hg-contaminated soil of the Isonzo River Plain and lagoon, and the city of Trieste. In the first case, GEM showed negative correlations with solar radiation, whereas it showed the opposite trend in the urban area of Trieste. This difference is likely a consequence of the distribution of breeze strength between the considered sites; the Isonzo alluvial plain is characterised by a diurnal sea breeze stronger than a nocturnal land breeze, whereas the urban area of Trieste shows an opposite pattern, with higher speeds during the night. In this urban site, no relevant sources of contamination are known, thus GEM could largely be carried from contaminated areas by wind during the diurnal sea breeze. The influence of sea breezes on atmospheric GEM levels in urban areas is well documented [17,25,61,64].
The wind rose chart shows that when GEM is high (up to an hourly mean value of 8 ng m −3 ) with winds prevalently blowing from W and NW, in the direction of the highly contaminated Fossalon plain, it (GEM) could reach the city of Trieste via the seawater surface of the Gulf ( Figure 5). It was hypothesised that a variation in wind speed is not sufficient to cause a dilution process responsible for the decrease in GEM during the night. A possible explanation could instead be that there is another significant source of GEM in the area. This could be represented by the waters of the Gulf of Trieste itself, since it is well-known that elemental Hg can be transferred not only between mining-impacted marine sediments and seawater, but also from seawater to the surrounding inhabited coastal area. In a previous study, Wänberg et al. [53] found that atmospheric GEM levels in Piran follow a similar pattern, showing higher values under conditions of low wind speed blowing from the north. The authors postulated that these air masses were enriched in GEM while passing over the contaminated gulf due to GEM emission from the surface of the sea, a process potentially relevant thanks to the abundance in the water column of this area of dissolved Hg available for reduction and subsequent volatilisation [46,65]. Unfortunately, no data regarding GEM in the Gulf of Trieste are available to more strongly support this hypothesis.
Atmosphere 2020, 11, x FOR PEER REVIEW 8 of 12 for the decrease in GEM during the night. A possible explanation could instead be that there is another significant source of GEM in the area. This could be represented by the waters of the Gulf of Trieste itself, since it is well-known that elemental Hg can be transferred not only between miningimpacted marine sediments and seawater, but also from seawater to the surrounding inhabited coastal area. In a previous study, Wänberg et al. [53] found that atmospheric GEM levels in Piran follow a similar pattern, showing higher values under conditions of low wind speed blowing from the north. The authors postulated that these air masses were enriched in GEM while passing over the contaminated gulf due to GEM emission from the surface of the sea, a process potentially relevant thanks to the abundance in the water column of this area of dissolved Hg available for reduction and subsequent volatilisation [46,65]. Unfortunately, no data regarding GEM in the Gulf of Trieste are available to more strongly support this hypothesis.

Conclusions
This study provides new data regarding gaseous elemental mercury (GEM) in the atmosphere around the Gulf of Trieste, which is a historically Hg-contaminated coastal area via sediment/soilassociated metal dispersion from the Isonzo River resulting from mercury mining from Idrija. The results of the study show that GEM concentrations do not reach levels of concern for the population. However, significant variations of GEM have been observed, showing that the coastal alluvial plain

Conclusions
This study provides new data regarding gaseous elemental mercury (GEM) in the atmosphere around the Gulf of Trieste, which is a historically Hg-contaminated coastal area via sediment/soil-associated metal dispersion from the Isonzo River resulting from mercury mining from Idrija. The results of the study show that GEM concentrations do not reach levels of concern for the population. However, significant variations of GEM have been observed, showing that the coastal alluvial plain built up along the Isonzo River is an active source of mercury in the atmosphere. Anomalous peaks of GEM have been observed, especially at sunset or during the night, when the sea breeze turns into the land breeze, resulting in a decrease in wind speed and scarce low mixing of the low atmospheric layer in contact with the ground. These sites with high-amplitude plumes of GEM are also worth long-term monitoring for continuous data acquisition.
In the urban area of Trieste, the correlation between wind direction and GEM indicates a mass flow of GEM transported from the Isonzo coastal alluvial plain without excluding a contribution from coastal waters of the Gulf to the surrounding areas. The main difference between the nonurban contaminated land/lagoon and the urban area investigated is a different day-night pattern of GEM distribution. On the sites contaminated by Hg, GEM tends to be higher during the night, whereas urban areas show an opposite trend, with maximum values reached during the day. A possible explanation is that in the urban area, anthropogenic Hg emissions contributed significantly to the atmospheric GEM budget, whereas releases from contaminated soils at Fossalon are mainly driven by natural re-emission processes. Further research is needed to identify and quantify the emission sources in both environments, also including monitoring of other trace gases (e.g., CO 2 , CO, CH 4 , SO 2 ) usually used as signatures of anthropogenic influence. Future study is also required to understand the contribution of atmospheric Hg deposition to the observed daily GEM patterns, to determine whether their contribution to GEM depletion is higher than that of atmospheric mixing and transport, particularly at those sites which showed a higher daily variability (e.g., over the Isonzo coastal alluvial plain). Moreover, determination of the different atmospheric Hg species, together with measurements of atmospheric oxidants such as halogen radicals, could give further information regarding GEM oxidation to RGM, which could be particularly relevant for coastal environments and contribute to the enhancement of Hg deposition on the substrate.