Transport of Semivolatile Organic Compounds to the Tibetan Plateau

Feb 10, 2010 - Figure 2 includes regression lines only if a regression was significant .... is appreciated for logistic support and arrangements; we a...
0 downloads 0 Views 5MB Size
Environ. Sci. Technol. 2010, 44, 1559–1565

Transport of Semivolatile Organic Compounds to the Tibetan Plateau: Spatial and Temporal Variation in Air Concentrations in Mountainous Western Sichuan, China W E N J I E L I U , †,‡ D A Z H O U C H E N , †,§ X I A N D E L I U , * ,† X I A O Y A N Z H E N G , † WEN YANG,† JOHN N. WESTGATE,⊥ AND FRANK WANIA⊥ Chinese Research Academy of Environmental Sciences, Beijing 100012, China, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou 730000, China, National Research Center for Certified Reference Materials, Beijing 100013, China, and Department of Chemistry and Department of Physical and Environmental Sciences, University of Toronto Scarborough, 1265 Military Trail, Toronto, Ontario M1C 1A4, Canada

Received September 12, 2009. Revised manuscript received January 27, 2010. Accepted January 28, 2010.

The distribution of organochlorine pesticides and polychlorinated biphenyls in air along an altitudinal transect on Balang Mountain in western China was measured by deploying XAD-2 resin based passive air samplers in duplicate at seven sites with elevations ranging from 1242 to 4485 m above sea level for five consecutive six-month periods between 2005 and 2008. Analyzed by gas chromatography-high resolution mass spectrometry, concentrations of hexachlorobenzene were highest, followed by hexachlorocyclohexanes, DDT-related compounds and PCB congeners 28 and 52. Except for hexachlorobenzene, which had largely uniform concentrations in space and time, there were clear seasonal variations with concentrations in summer being higher than in winter. With a few exceptions, concentrations that vary little with altitude suggest that the presence of these chemicals in the area is almost entirely due to atmospheric transport, most likely from the Chengdu plain. This is supported by similarities in the relative abundance of different compounds and in the differences between summer and winter concentrations measured in the city of Chengdu and in the mountains. Furthermore, air mass trajectories during the sampling period often originate to the East, passing over the Western part of the Sichuan basin, including the Chengdu plain, prior to arriving at the sampling sites. Higher summer time values in the mountains are due to more contaminated air being blown into the region, presumably due either to higher pesticide usage in summer or due to higher temperatures leading to higher evaporation in source regions. Air and soil from the region are in equilibrium with respect to R-HCH, γ-HCH, and HCB, whereas a situation of net deposition prevails for p,p’-DDE and p,p’-DDT. * Corresponding author e-mail: [email protected]. † Chinese Research Academy of Environmental Sciences. ‡ Chinese Academy of Sciences. § National Research Center for Certified Reference Materials. ⊥ University of Toronto. 10.1021/es902764z

 2010 American Chemical Society

Published on Web 02/10/2010

Introduction Semivolatile organic compounds (SVOCs), which includes organochlorine pesticides (OCPs) and polychlorinated biphenyls (PCBs), are able to undergo long-range atmospheric transport (LRAT) and, as a result, have been ubiquitously detected at the global scale. In particular, LRAT can deliver SVOCs to high latitudes and altitudes where they may negatively affect remote ecosystems. There has been particular interest in studying SVOCs in mountainous areas (e.g., ref 1-3). The atmosphere is the environmental compartment that most directly reflects short-term changes in concentration and compositional shifts in SVOCs, but measurements of atmospheric concentrations at high altitudes are logistically challenging and often restricted to sites with access to electricity (4, 5). The passive air sampling (PAS) techniques developed in recent years (e.g., ref 6) have facilitated studies of air in remote mountainous areas where no electricity supply is available and routine maintenance is difficult (7-9). Strong elevation gradients and the close proximity to densely populated and rapidly industrializing China may render the Eastern edge of the Tibetan-Qinghai Plateau particularly vulnerable to airborne contaminants (10). Interpretation of SVOC soil concentrations sampled along the windward slope of Balang Mountain in Wolong Nature Reserve (WNR) in Sichuan province in western China has confirmed the potential for mountain cold-trapping in the region (11). In this study, air concentrations were measured along the same mountain slope. PAS were deployed for five consecutive six month periods from October 2005 to April 2008 and several OCPs and PCBs were quantified in the PAS extracts using gas chromatography-high resolution mass spectrometry (GC-HRMS). To better understand LRAT and air-soil exchange, we investigated variations of atmospheric SVOCs in WNR with altitude and season, we compared the air concentrations with those measured in the nearby city of Chengdu, and we paired them with the soil concentration data from the same transect (11).

Materials and Methods Sampling. WNR is located in Wenchuan county of Sichuan province in western China in the transition area between Chengdu plain and the Tibet-Qinghai Plateau (see Figure 1). Climate and surface cover vary along the mountain slope: temperature drops with increasing altitude, some mountain peaks have year-round snow-cover, and wind speed and insolation vary substantially. Deciduous woods dominate below 2000 m above sea level (m) with mixed forest between 2000 and 2600 m; from there to 3600 m conifers prevail; from 3600 to 4000 m alpine bush and meadow are present but beyond 4000 m only alpine meadow can be found. Between October 2005 and April 2008 duplicate XADresin based PAS were installed for consecutive six month periods at seven sites along provincial road 303, which winds its way through WNR and up the windward slope of Balang Mountain reaching the Pass at 4485 m (see Figure 1). The sampling campaign thus included three winter periods and two summer seasons. From April 2007 to April 2008, a pair of sampler was additionally installed for two six-month periods in Chengdu, a city of 11 million inhabitants located in the Sichuan basin about 80-105 km to the East of WNR. Site elevations, geographical coordinates and dominant vegetation are listed in Table 1. The lowest and highest sites within WNR were approximately 80 km apart. Most PAS were sited along the stream in the valley, whereas the highest sites VOL. 44, NO. 5, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1559

FIGURE 1. Maps showing the locations of Chengdu and the study area, the Wolong Natural Reserve, within Sichuan province in China, and eight PAS sampling sites along road 303.

TABLE 1. Elevation, Geographical Coordinates and Vegetation Cover of the PAS Sampling Sites in the Wolong Natural Reserve and the City of Chengdu site

altitude

longitude (east)

latitude (north)

pass 95km milestone Beimuping Dengsheng Sandaoqiao Panda Centre Gengda Shuijiepai Chengdu

4485 m 3619 m 3377 m 2828 m 2190 m 1847 m 1439 m 1242 m 507 m

102°53.74′ 102°57.37′ 102°58.91′ 102°58.34′ 103°05.99′ 103°13.15′ 103°18.72′ 103°23.39′ 104°02.93′

30°54.69′ 30°52.40′ 30°53.80′ 30°51.48′ 30°57.81′ 31°04.41′ 31°05.08′ 31°03.72′ 30°38.63′

were on the slope of the mountain far away from the stream. The sampling duration was approximately 182 days for all samples. The XAD-resin based PAS has been used and described in detail previously (6). Briefly, it consists of a resin-filled stainless steel mesh cylinder that is suspended in an inverted galvanized steel can with an open bottom. The PAS is deployed at 1.5 m above ground or on the roof of a building. Contaminants are taken up in the resin from the atmosphere by diffusion. The amount of a chemical sequestered in a PAS can be converted into a volumetric air concentration by dividing by the sampling duration and an air sampling rate R in units of m3 · air · day-1 (6). Extraction and Quantification. XAD resin was precleaned by Soxhlet extraction with dichloromethane (DCM) for 24 h, packed in the mesh cylinders and stored in sealed containers before use. In total 68 exposed PAS and 10 field blanks were analyzed. After 2,4,5,6-tetrachloro-m-xylene (TMX) and PCB209 (from J&K Company) were added as recovery surrogates, the retrieved XAD-resin was Soxhlet extracted for 24 h with DCM and the extracts were concentrated, solvent transferred into iso-octane, and further concentrated to a final volume of 1.0 mL. SVOCs were separated on a VT-1 capillary column (30 m, 0.25 mm i.d., film thickness 0.25 µm) and analyzed using a Finnigan MAT 900 XL GC-HRMS operating in the electron impact mode. The instrumental conditions were injector, ion source, and interface temperature 210, 220, and 1560

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 5, 2010

vegetation alpine meadow alpine meadow alpine bush-meadow cold-temperate coniferous forest mixed coniferous-deciduous forest mixed deciduous-broad leaved forest evergreen broad leaved forest evergreen broad leaved forest evergreen broad leaved woods

250 °C, respectively, filament 0.55 ma and 42 eV, electron amplifier 1.75 kV, mass resolution 10 000, temperature program: 70 °C (1 min), 70-150 at 20 °C · min-1, 150-250 at 2 °C · min-1, 250 °C (20 min). The carrier gas was helium at a constant flow rate of 1.2 mL · min-1. A 1 µL sample was injected in splitless mode, and switched to split mode one minute later. The following 11 SVOCs were quantified: R-, β-, γ-, and δ-hexachlorocyclohexane (HCH), hexachlorobenzene (HCB), PCB-28 and 52, p,p’-DDE, p,p’-DDD, p,p’-DDT, and o,p’-DDT. Before injection C-13 labeled γ-HCH, HCB, PCB-28, PCB-52, p,p’-DDE, and p,p’-DDT were added as internal standards. The selected ions for the target analytes and corresponding labeled compounds are 219 and 225 for HCHs, 286 and 292 for HCB, 318 and 330 for p,p’-DDE, 235 and 247 for p,p’-DDD, p,p’-DDT, o,p’-DDT, 258 and 270 for PCB-28 and other trichlorinated biphenyls; 292 and 304 for PCB-52 and other tetrachlorinated biphenyls. A laboratory blank showed a small interference, only for HCB, for which a correction was necessary. In the field blanks SVOCs were either absent or low and corrected if necessary. Duplicate measurements for standard-spiked (standard from National Research Center for Certified Reference Materials, Beijing, China) XAD-resins yielded an average recovery for 11 target analytes of 68 and 79%, respectively. Mean recoveries for the surrogates were TMX: 74%, 90%; PCB-209: 88%, 96%; for winter and summer samples, respectively. The detection limit ranged between 0.05 and 2 pg · m-3.

Airshed Calculation. Five-day back trajectories arriving at the coordinates of the 95 km milestone site were calculated at 50, 100, and 200 m above ground level at 6 h intervals for each day of the first year of sampling (October 2005 to October 2006) using the trajectory model of the Canadian Meteorological Centre. The more than 4000 trajectories for each of these three heights were compiled to produce back trajectory probability density maps, referred to as “airsheds” (Supporting Information (SI) Figure S1).

Results and Discussion Calculating Volumetric Air Concentrations. The amount of the target compounds sequestered in the PAS (in units of ng · PAS-1) are presented in SI Table S1. The amounts of HCB in the PAS extracts were highest, those of R-HCH, γ-HCH, and o,p’-DDT were at intermediate levels, and p,p’-DDE, p,p’-DDT, and two indicative PCB congeners occurred at lowest levels. Previous studies using the XAD-PAS had applied a constant sampling rate R to convert such amounts into volumetric air concentrations, assuming that temperature, wind speed and atmospheric pressure have a negligible influence on the kinetics of uptake (7). Some studies sought to account for such influences by quantifying the effect of temperature and pressure on molecular diffusion (8). More recent studies indicate that environmental factors may have a stronger influence on the kinetics of uptake than previously thought (12). An inspection of the data in SI Table S1 reveals that the PAS deployed at Gengda had consistently higher sequestered amounts of the target substances. The fact that it occurred for all substances to the same extent and for both winter and summer samples, suggests that local sources can not be responsible. Instead it is an indication that the kinetics of uptake at Gengda was higher than at the other sampling sites. This is consistent with the peculiar topographic circumstances, which make this part of the transect particularly windy. Compared to the other analytes, the amount of HCB sequestered in the PAS from WNR is very constant with altitude and season (SI Table S1 and Figure S2). The coefficient of variation of the HCB concentration in 64 exposed samples is 43%, which is reduced to 34%, when the samples from Gengda are excluded. HCB is very resistant to attack by photooxidants (13) and because of its relatively high volatility is not subject to efficient dry particle and wet deposition. As a result it has an exceptionally long atmospheric residence time in excess of one year, which is reflected in air concentrations in nonsource regions that are remarkably uniform in space and time, even on a hemispheric scale (14-16). We suggest that this makes it possible to use HCB to derive sampling rates R that are specific for each individual PAS. Based on ref 16, we assumed an average air concentration of HCB of 56 pg · m-3, representative of Northern hemispheric background conditions, to prevail in WNR throughout the year. This is also consistent with levels recorded at sites in Western China, for example, 38 pg · m-3 at Waliguan in Qinghai in spring 2005 (17) and 43 pg · m-3 in Gonggashan, a rural/background site by Jaward et al. (18). Levels in northeastern China are much higher (18). By dividing the amount of HCB sequestered in each PAS by this concentration and the sampling duration, we obtain a sampler-specific R, (SI Table S2) which then can be used to calculate volumetric air concentrations for the other analytes (SI Table S3). This way we indirectly account for the environmental factors (temperature, pressure, wind speed, and turbulence) that could affect R, and we can be more confident in comparing levels of the same chemical between seasons and between sites at different altitude. The choice of the constant HCB air concentration affects the numerical value of R and the calculated volumetric concentrations for the other SVOCs, but it does not affect relative differences

in concentration in space and time. The sampler specific sampling rates estimated this way average 2.83 ( 1.21 m3 · day-1. If the unusually high values at Gengda are excluded, R is 2.53 ( 0.85 m3 · day-1. This compares favorably with the R-values determined previously (12). It should be stressed that this approach assumes that all analytes have uptake rates similar to HCB, which may not be the case (12). These R values thus have to be considered approximations and differences in the concentrations between different chemicals have to be interpreted with this uncertainty in mind. The studies that yielded a circumpolar average of 56 pg · m-3 were conducted at stations situated close to sea level (16). It is presumably the mixing ratio of HCB in air that remains constant with altitude and not the actual volumetric concentration, which will be lower at lower atmospheric pressure. The approach thus yields concentrations of the SVOCs normalized to standard atmospheric pressure conditions. This facilitates comparison of the data from WNR with air concentrations reported in pg · m-3 in the literature (SI Tables S4-S6). We should note that the derived sampling rates R (SI Table S2) also apply to standard atmospheric pressure conditions, because they were all calculated assuming a volumetric air concentration of 56 pg · m-3. The actual sampling rates at the high elevation stations would therefore be somewhat higher than those listed in SI Table S2. The amounts of HCB sequestered in the PAS deployed at Chengdu (median of four samples of 134 ng · PAS-1) was notably higher than in the samples from WNR (overall median of 26 ng · PAS-1), suggesting that it would not be valid to assume the same constant HCB concentration in air at this site. We therefore used the average R obtained for the PAS at WNR for the same sampling periods to convert the amounts sequestered in Chengdu PAS to volumetric air concentrations. Mean Error of the Passive Air Sampling Technique. This study generated a large number of duplicate concentration values, which allows us to investigate the replication error of the XAD-based PAS and quantification technique. We estimated that error as the absolute difference divided by the average of the amounts sequestered in two PAS deployed at the same site at the same time. The overall median error for 264 pairs of data was 20%. However, this error was different for different chemicals (SI Table S7) and different sites (SI Table S8). The median error for different chemicals ranged from 10% for R-HCH and HCB (number of data pairs n ) 31) to 15% for γ- and β-HCH (n ) 31 and 18, respectively), and around 30% for PCB-28, PCB-52, p,p’-DDT, and p,p’-DDE (n ) 30, 30, 27, and 24). Chemicals that were only detected sporadically, such as δ-HCH and p,p’-DDE had a higher error. The error was clearly inversely related to the detected amounts of a chemical, that is, high sequestered amounts have a smaller error than small amounts. Whereas this intuitively makes sense due to the larger quantification error for trace levels close to the detection limit, it is harder to explain, why there are also differences in the error between different sites. In particular, duplicate PAS deployed at lower elevations (