Environ. Sci. Technol. 2001, 35, 4066-4073
Grass-Air Exchange of Polychlorinated Biphenyls H . H U N G , * ,† G . O . T H O M A S , § K. C. JONES,§ AND D. MACKAY‡ Department of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, ON, M5S 3E5 Canada, Canadian Environmental Modelling Centre, Trent University, Peterborough, ON, K9J 7B8 Canada, and Department of Environmental Science, Institute of Environmental and Natural Sciences, Lancaster University, Lancaster, LA1 4YQ, U.K.
Three field experiments were performed to assess the clearance, uptake, and exchange kinetics of polychlorinated biphenyls (PCBs) between grass and the atmosphere using mixed- and single-species grass (Holcus lanatus). In the clearance experiment, the grass was artificially contaminated by equilibration with diluted Aroclor vapor then exposed to field air, and the rates of depletion were monitored by sampling at regular intervals to determine clearance rate constants. In the uptake experiment, the uptake of PCBs from the ambient atmosphere was followed in growing grass at ambient concentrations for 3 and 6 weeks by analysis of segmented samples along the length of the sward. In the third experiment, diurnal temperaturedriven changes in grass concentrations were measured. The results indicate that the grass is behaving as a twocompartment system: (1) a fast-exchanging surface adsorption site with a response time of hours and a capacity essentially independent of KOA, the octanol-air partition coefficient and (2) a slow responding site with a response time of weeks, the capacity of which is related to KOA. The kinetic and equilibrium phenomena involved in grassair exchange are thus complex and are not adequately described by simple first-order rate constants and equilibrium partitioning coefficients.
Introduction An improved understanding of the equilibria and kinetics of atmosphere-grass exchange of “semivolatile” organic chemicals (SOCs) such as the PCBs is desirable for several reasons. Grasses can be contaminated by deposition from the atmosphere leading to subsequent transfer in the food chain to meat and dairy products as well as wildlife. Exchange between the atmosphere and surface vegetation can influence a chemical’s atmospheric concentrations, environmental persistence, and long-range transport. Vegetation can also serve as a monitor of contamination levels in the atmosphere provided that the air-vegetation concentration relationships * Corresponding author phone: (416)739-5944; fax: (416)739-5708; e-mail:
[email protected]. Current address: Meteorological Service of Canada, 4905 Dufferin Street, Downsview, ON M3H 5T4, Canada. † University of Toronto. ‡ Trent University. § Lancaster University. 4066
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 20, 2001
are understood. McLachlan (1) has proposed a general framework for assessing the exchange phenomena in which there are equilibrium, kinetic, and aerosol dominated regimes controlled by the substances’ octanol-air partition coefficient KOA. Several field and laboratory studies have probed the equilibria and kinetics of the exchange phenomena (2-11) as well as the diurnal changes in air concentrations (12). The current study comprises a set of three field experiments designed to elucidate the transport mechanisms of PCBs in an air-grass system under natural conditions in a well-characterized rural field site located in northwest England. Since PCB congeners cover a wide range of physical chemical properties, it is hoped that the results will be applicable to other substances with similar properties. In a clearance experiment, existing field grass was artificially contaminated by gaseous uptake of Aroclor mixtures under glass chambers and clearance followed for up to 36 h. In an uptake experiment, the change in PCB concentration was followed in growing grass for 6 weeks. Finally, an attempt was made to detect changes in grass concentration which occur as a result of diurnal temperature cycles. The clearance and uptake experiments were designed to be analogous to the successful corresponding bioconcentration tests with fish and water in which both processes are measured under different conditions with the aim of establishing clearance and uptake rate constants and their ratio, the equilibrium constant. An objective was to establish the applicability of this simple first-order approach to grassair exchange.
Experimental Section Field Site and Sampling. All experiments were performed on a well-characterized rural upland pasture at a field station with meteorological measurement facilities owned by Lancaster University. PCB concentrations in air have been regularly monitored at the site in recent years (2, 12-14). Clearance Experiment. The clearance experiment was performed on an open plot of mixed-species pasture (composed of approximately 30% of each of Lolium perenne and Holcus lanatus, 10% of each of Agrostis capillaris, Poa pratensis, and Cynosurus cristatus, and various other grasses and forbs) with no trees or sampling equipment in proximity. The grass was trimmed to approximately 10 cm, and a clear glass open-topped chamber was inverted over it to contaminate the grass artificially. Five microliters each of Aroclor 1242 and 1260 standards (at 1 µg/µL in hexane and cyclohexane, respectively) were placed at even intervals on the roof of the chamber at the field, and the solvent was allowed to evaporate for approximately 5 min before covering the grass. The grass was left to become contaminated for 7 days, during which time the grass showed no signs of wilting or drying. After the chamber was removed, grass samples of 3-5 g were cut by hand with a pair of scissors prerinsed in hexane, from random locations at regular time intervals from the plot but avoiding grass close to the edge to prevent inclusion of uncontaminated grass. Four trials were performed between June 25th and August 17th, 1998. Nine to 13 samples were collected in each trial which lasted for 36, 12, 3, and 6 h. Soil samples from the top 5 cm of soil underneath the grass samples were collected at specific times after exposure to field air. No soil samples were taken during the first trial. Soil samples were collected at regular intervals during the remaining trials. To avoid contamination, all grass and soil samples were sealed in labeled plastic bags immediately after 10.1021/es001820e CCC: $20.00
2001 American Chemical Society Published on Web 09/07/2001
sampling and were promptly transported in ice boxes to the laboratory and were kept frozen prior to analysis. Air samples were collected at regular time intervals (2-8 days per sample) from May 14th to July 9th. Air-borne particles were trapped on a Whatman glass microfiber filter (GFF) (grade GF/A, 10 cm diameter) and the vapor phase on a polyurethane foam plug (PP) (13). Filters were spiked with PCB recovery standards before sampling commenced. Uptake Experiment. An existing area of mixed-species pasture (same composition as that used in the clearance experiment) and an experimental plot of a single grass species (Holcus lanatus - a fast growing grass with fine hairs) were fenced and separated into 1 m2 subplots. Grass was allowed to grow for 3 weeks and 6 weeks in separate triplicate subplots (total of 12 × 1 m2 subplots), both exposure periods being harvested on the same day. Five canes were posted within each 1 m2 subplot, and the average heights of growth around each cane were marked each week. Grass harvested from the 6-week experiment was immediately segmented according to the markings on the canes (the tip of the grass being 6 weeks old and the bottom of the grass being 1 week old). Grass from the 3-week study was too short to be divided into weekly growth segments and were halved, yielding top and bottom segments. The Holcus l. subplots were hand-weeded before harvesting to yield strictly single-species samples. Samples were sealed in plastic bags in the field and were stored at -20 °C until analysis. The water content of each grass segment from the uptake experiment was determined by drying 3-10 g of samples at 100 °C overnight. Water contents were not measured for the 5- and 6-week-old segments of the 6-week samples and the top segment of the 3-week growth mixed-species samples because all grass samples were used for PCB analysis. The average dry weight percentages were used to calculate the concentration per unit dry weight for these segments. Duplicate samples were performed on mixed-species 6-week samples. The average water content of the mixed-species grass was 79% and that of Holcus l. was 80%. Diurnal Study. The diurnal study employed the same mixed-species pasture described above. Over a 3-day period in the late summer grass samples (50 g) were taken from an area of pasture which had not been cut for at least 1 month (the grass length was approximately 10 cm) at 6 hourly intervals. Samples were sealed in plastic bags and stored at -20 °C until analysis. Sampling and Analysis. Sample Preparation and Analysis. Grass samples were analyzed by previously published methods (15). Each 25 g sample of grass was frozen with liquid nitrogen, ground with anhydrous sodium sulfate, spiked with 13C-labeled PCB recovery standards, and Soxhlet extracted with dichloromethane. The extract was evaporated to dryness, resuspended in hexane, and cleaned up using silica/acid silica chromatography (eluted with hexane) followed by gel permeation chromatography (using Biobeads S-X3, eluted with 1:1 DCM:hexane). Internal standards were added, and the sample evaporated to a small volume for analysis by GC-LRMS. For the clearance studies, approximately 3 g of grass was used because of the relatively high PCB concentrations. Ten grams of soil was mixed with anhydrous sodium sulfate at room temperature, spiked with recovery standards, extracted, and cleaned up in the same way as grass samples. Air filter and PUF samples were Soxhlet extracted separately with hexane after being spiked with 13C-labeled recovery standards and cleaned up using activated silica chromatography, as reported previously (2). Extracts were analyzed by GC-MS on a Fisons MD-800 in EI mode monitoring selected ions, with separation on a CPSil-8 50 m column fitted with a retention gap, and splitless
injection. Fifty-two PCB congeners were analyzed against a series of calibration standards using an internal standard method. Quality Control. An in-house reference material was analyzed as every tenth sample, and duplicate or triplicate samples of each grass sample were analyzed, depending on sample availability. The average standard deviation for the total PCB concentration was 34% between replicates.
Results Clearance Experiment. Before performing the experiment, the evenness of contamination of the grass within the chamber was determined. The contamination procedures were performed, and grass samples were collected at random positions within the plot immediately after the chamber was removed. A 34% relative standard deviation (rsd) in ΣPCB (range from 19% to 89% for individual congeners with most rsd below 40%) was found between the five samples collected, which was considered acceptable. The original intention of collecting soil samples was to evaluate the mass balance during the clearance experiment. The ΣPCB concentrations measured ranged from 340 to 1300 pg/g f.w. (Table SI-1, Supporting Information) which are similar to the background soil levels at this location. Background soil concentrations are reported by Cousins and Jones (16). The background soil ΣPCB concentrations measured using the same method was found to be 1.3 µg/kg dry soil. With a soil moisture content of 33% (w/w), this concentration translates to 870 pg/g f.w. In other words, the soil concentration detected during the current study was not a result of the contamination period. Assuming that the background grass, soil, and air concentrations approached equilibrium, re-emission from soil and subsequent contamination of grass sward is, thus, of minor importance in this study since the grass concentrations were very high compared to the background. The only possible influence of the presence of soil in the mass balance is the transfer from grass to stagnant soil-air boundary layer to soil surface. However, no increase in soil levels was observed over time, indicating that soil plays a minor role in the grass-air-soil transport system in this particular study. The contaminated grass concentrations were initially much higher than those of natural field grass. The initial contaminated grass ΣPCB concentrations were found to be 700 ng/g fresh weight, 390 ng/g, 390 ng/g, and 270 ng/g (all f.w.) for the first, second, third, and fourth trials, respectively. The average natural field grass ΣPCB concentration was approximately 470 pg/g f.w.; thus, the initial contaminated grass concentrations were about 1000 times that of the natural field grass. Clearance results for selected congeners obtained during the first 36 h trial are presented in Figure 1. The grass concentrations decreased rapidly with time once exposed to natural field air and then dropped more slowly after 16 h of exposure during the first trial. After 1 h of exposure grass concentrations of most congeners had dropped below 70% of the initial concentrations. Subsequent trials were performed over shorter time intervals with more frequent sampling to determine the initial clearance rate. The results of the third trial, which was performed over 3 h, are presented as Figure SI-1 of the Supporting Information. The concentrations of the lower, more volatile PCB congeners, such as PCB 28 and 37, fluctuated with no discernible trends. In no case did the contaminated grass reach the same concentrations as the natural field grass within the duration of the experiment. The results suggest two significant features of the clearance process: First, there was a rapid initial decrease in concentration with a half-life of the order of an hour or two. Within approximately 10 h, the grass approached a fairly constant VOL. 35, NO. 20, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4067
FIGURE 1. Depuration rates of selected PCB congeners in the first (clearance) trial: (a) PCB 28 and 37; (b) PCB 49, 52, and 66; (c) PCB 60, 101, and 110; (d) PCB 138 and 153; (e) PCB 180, 183, and 187; and (f) PCB 193 and 206. but still slowly declining concentration. The initial half-time appears to be somewhat longer for higher congeners, but the effect was slight. This behavior suggests clearance from two compartments, a surface in which the sorbed PCB are very labile, and a less labile compartment which may be in the interior of the leaf, from which the PCBs are clearing with a half-life of weeks. Second, within the first 8 h of clearance, about 75% of the initial amount of PCBs was depleted from the grass, implying that the slowly clearing 4068
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 20, 2001
compartment contained about 25% of the initial total mass of PCBs. The corresponding concentration was still a factor of about 200 above the ambient grass concentration and represented a much higher fugacity than that of ambient air or grass exposed to it. An interpretation of these results is that during the 7-day contamination phase, PCBs rapidly saturated the surface and gradually penetrated into the interior reaching a condition with 75-80% on the surface and 20-25% in the interior. The
TABLE 1. Average Mass Transfer Rate Constants, Average Temperatures, and Windspeeds trial
av T (°C)
av windspeed (m/s)
av k2a (h-1)
2 3 4
14.4 20.8 16.5
4.8 2.7 4.5
0.148 ( 0.029 0.619 ( 0.078 0.305 ( 0.027
a
Average of all detectable PCB congeners.
clearance process from the labile surface compartment can be quantified by a first-order expression
dCS ) k1CA - k2CS dt
(1)
where CS and CA are concentrations (mol/m3) in grass surface and air, t is time in h, and k1 and k2 are the uptake and clearance rate constants with units of h-1. CA was very low compared to the high concentrations in grass; therefore, the first term in eq 1 is negligible. Solving eq 1 gives
ln
CS ) - k2t CSo
(2)
where CSo is the concentration in grass at time zero. A plot of ln CS/CSo versus time, t, of the first few hours of clearance should give a slope of k2 (h-1), the clearance rate constant from the surface. The regression coefficients, R 2, of most lower congeners were low in all four trials. For most higher congeners with log KOA at 13 °C greater than 10.5, R 2 were greater than 0.8. The difference in k2 between congeners was slight (Table SI-2), and it is believed that an average value can be assumed for all congeners with log KOA between 9 and 12 at any particular temperature. Meteorological information was only available for the last three trials. The meteorological conditions were quite stable. Wind was blowing from the southeast/east/northeast during all three trials. The windspeeds (measured at 1.75 m) ranged from 2.2 to 7.2 m/s, 1.3 to 4.5 m/s, and 2.7 to 5.8 m/s during the second, third, and fourth trials. Table 1 lists the average temperature and windspeed of the three trials and the corresponding average k2. It seems that the average k2 increased with temperature. However, a relationship between k2 and windspeed was not apparent. According to the framework for the interpretation of semivolatile organic compounds in plants developed by McLachlan (1), air-plant transfer of compounds with log KOA between 8.5 and 11 is dominated by kinetically limited gaseous deposition, i.e., air-side resistance is much greater than plant-side resistance. PCB congeners involved in the current experiment fall within this range. Although the current situation pertains to the clearance process instead of the deposition process, the same theory should apply. If air-side resistance was dominant, the clearance rate constant is presumably dependent on the magnitude of the air diffusivity of the PCB. The difference in air diffusivity between congeners is, however, small (see Supporting Information for explanation). This difference, when translated to the clearance rate constant, is judged to be insignificant under field conditions. Ko¨mp and McLachlan (17) performed a similar experiment in which contaminated grass was exposed to outdoor air and was sampled over a period of 240 h. In their experiment, a correlation between the grass clearance half-lives and log KPA (plant-air partition coefficient) was observed. In previous studies, the same group has observed linear relationships between log KPA and log KOA (18). The discrepancies between the results of their experiment and this study are believed to have been caused by the difference in study times and conditions. In this study, the clearance rate constants
observed were those of only the grass surface during the first few hours of depletion, while those studied in ref 17 were over a longer period (240 h). The clearance half-lives reported by Ko¨mp and McLachlan were therefore overall values describing clearance of both surface and inner leaf with a greater contribution from the inner leaf due to the long study period of their experiment. The different observations from these two experiments imply that sorption/desorption from the surface of plants were independent of log KOA, while uptake/clearance from the inner leaf is a function of log KOA, but the effect was buffered by the presence of the leaf surface. A plot of log KPA (plant-air partition coefficient) or log scavenging ratio (plant concentration in pg/g d.w. divided by air concentration in pg/m3) against log KOA should result in a slope of unity if plant lipid behaves similarly to octanol. However, very different slopes have been found in different studies with different plant species, and they were usually different from unity. Ko¨mp and McLachlan (18) has measured the KPA of five different pasture species using a fugacity meter. Slopes of 0.57-1.15 at 25 °C were found. Thomas et al. (3) have calculated the scavenging ratios using the same mixed species grass as in this study. Slopes of 0.161-0.516 from April to October 1996 were observed. If (de)sorption from the plant surface was independent of log KOA but that from the inner leaf was dependent on log KOA, the different slopes observed reflect different proportions of these sorbing compartments. It is highly probable that surface sorption/ desorption is dependent on the morphology of the specific plant surface, resulting in different plant species exhibiting different log KPA values and different slopes when correlated with log KOA (18). Uptake Experiment. Figure 2 shows the concentrations of selected PCBs in samples taken from 1 to 6 weeks from the single grass species Holcus lanatus. The results obtained from the mixed species grass are similar and are presented as Figure SI-2 in the Supporting Information. The water content of each segment and the yield, i.e., total mass of grass sample collected per week per m2 of grass plot, are given in Table SI-3. Statistical (t-)tests showed neither significant difference between the top and bottom segments of the 3-week segmentation samples nor in the first 4 weeks of growth of the 6-week segmentation samples. The concentrations increased significantly during the fifth week and reached the highest values on the sixth week. The total concentrations of each 6-week segment ranged from 1500 to 5700 pg ΣPCB/g DW in Holcus l. and 11006300 pg ΣPCB/g DW in the mixed-species grass. The concentrations of most of the congeners and the ΣPCB were found to be highest in the segments obtained from the grass which had been exposed longest, i.e., the tips. Since the water contents of the 5- and 6-week old segments were unknown, the average values from the 1-4-week old segments were used to calculate the dry-weight concentrations. As the water content may decrease with age, this could result in a slight overestimation of dry weight concentrations for these segments. There was no trend in water content variation between segments for Holcus l. The water content decreased slightly with age for the mixed-species grass; at about 1% per week. This slight decrease in water content cannot explain the significant increase in concentrations (Figures 2 and SI-2). From May 21st to July 2nd, the average ambient temperature was 13 °C, and the average daily run-of-wind was 284 km measured at a height of 3 m from the soil surface. The estimated average windspeed is, therefore, 3.3 m/s. The average daily rainfall was 4.4 mm. The air concentrations measured during the sampling period were fairly constant. The measured air concentrations are given in Table SI-4. Slightly elevated atmospheric concentrations of most congeners were observed over a 3-day VOL. 35, NO. 20, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4069
FIGURE 2. Holcus lanatussrelationships between dry weight concentrations and age of grass during the uptake experiment: (a) PCB 28 and 52, (b) PCB 101, (c) PCB138, (d) PCB110, (e) PCB 151 and 149, (f) PCB 187 and 183. period from May 30th to June 1st, corresponding to the growing period of the 5-week-old grass segment. Five-day Lagrangian backtrajectory analysis has been performed at 991 kPa at 6 h intervals. Results have shown that the sampling site was strongly affected by air masses originating from the North America/Arctic region, except the 3 days of recorded elevated air concentrations during which influence from northern Europe was most important (average trajectories are shown in Figure SI-3). A similar elevation in air concentrations, especially for the higher congeners, also occurred between June 14th and 23rd corresponding to the growing period of the 2-week-old grass segment. While the concentrations in the 5-week-old grass segment were high, those in the 2-week-old segment were not. In other words, the significant elevation in grass concentrations in the 5and 6-week-old grass segments cannot be explained by these slightly elevated air concentration episodes. Besides, the atmospheric concentrations of PCB 183 and 187, unlike other congeners, did not increase significantly during the two periods mentioned above. However, both congeners showed an increase in grass concentrations in both the 5- and 6-week4070
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 20, 2001
old segments; again, indicating that this increase in PCB concentrations in the oldest grass segments was not the result of elevated air concentrations. It had been expected that a steady increase in concentration with age would be detected, i.e., concentrations would increase from soil surface to tip. The sudden increase in concentration during weeks 5 and 6 was unexpected. Clearly, a simple first-order uptake relationship does not apply. Table 2 shows the results of regression analyses on the log dimensionless scavenging ratios (concentration in plant/ concentration in air), calculated assuming a grass density of 106 g/m3 fresh grass, versus log KOA for each of the segments. The slopes are considerably less than unity, as has been seen previously (3, 18) although most are statistically highly significant (99% CI, p < 0.01). The slopes are also independent of the age of the segment. The dimensionless grass-air concentration ratios are plotted in Figure 3 as a function of KOA for the 6-week samples for each sward type. Also shown in this figure is the average correlation developed by Ko¨mp and McLachlan (18) for five different pasture species using the laboratory fugacity meter
TABLE 2. Regression Results of log(CP/CA) versus log KOA Holcus lanatus
mixed species age (wk)
slope
R2
p
slope
R2
p
1 2 3 4 5 6
0.137 0.311 0.247 0.196 0.299 0.281
0.11 0.79 0.64 0.38 0.80 0.74
0.1
