Environ. Sci. Technol. 2007, 41, 1124-1130
Organochlorine Pesticides in the Soils and Atmosphere of Costa Rica GILLIAN L. DALY,† YING D. LEI,† CAMILLA TEIXEIRA,‡ DEREK C. G. MUIR,‡ LUISA E. CASTILLO,§ LIISA M. M. JANTUNEN,| AND F R A N K W A N I A * ,† Departments of Chemistry and Physical and Environmental Sciences, University of Toronto at Scarborough, 1265 Military Trail, Toronto, Ontario, Canada M1C 1A4, Aquatic Ecosystem Protection Research Division, Environment Canada, 867 Lakeshore Road, Burlington, Ontario, Canada L7R 4A6, Instituto Regional de Estudios en Sustancias Toxicas, Campus Omar Dengo, Universidad Nacional, Heredia, Costa Rica, and Centre for Atmospheric Research Experiments, Environment Canada, 6248 Eighth Line, Egbert, Ontario, Canada L0L 1N0
A survey of the contamination of the physical environment of Costa Rica with banned organochlorine pesticides (OCPs) relied on sampling air and soil at 23 stations across the country in 2004. Average annual air concentrations, determined with XAD-based passive samplers, and surface soil concentrations were generally low when compared to values reported for North and Central America, which is consistent with relatively low historical domestic use and little atmospheric inflow from neighboring countries. Statistical analysis and concentration maps reveal three types of spatial distribution: R-hexachlorocyclohexane and p,p′-DDD had a relatively uniform distribution across the country; other DDT-related species were greatly elevated over the national average at Manuel Antonio, a National Park on the Pacific coast; and dieldrin, lindane, and chlordane-related species had higher concentrations in Costa Rica’s populated Central Valley. An altitudinal transect of stations in the Central Valley shows declining air-soil concentration ratios with elevation for lindane, likely driven by atmospheric inversions and soil organic carbon content. Enantiomeric composition of chiral OCPs in air and soil was close to racemic, with slight depletion of (-)R-HCH, (-)-cis-chlordane, and (+)-trans-chlordane. Estimated air-soil fugacity fractions are highly uncertain but indicate equilibrium conditions for most OCPs, net volatilization of lindane at some sites, and net deposition for p,p′-DDE. The study demonstrates an approach for quickly evaluating the spatial distribution of OCPs in an understudied area, identifying regionally important contaminants and areas of elevated concentrations.
Introduction Organochlorine pesticides (OCPs), such as DDT, chlordanes, dieldrin, and hexachlorocyclohexanes (HCHs), are among * Corresponding author phone: (416)287-7225; e-mail:
[email protected]. † University of Toronto at Scarborough. ‡ Aquatic Ecosystem Protection Research Division, Environment Canada. § Universidad Nacional. | Centre for Atmospheric Research Experiments, Environment Canada. 1124
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 4, 2007
the substances restricted or banned globally under the Stockholm Convention on Persistent Organic Pollutants. The concern around OCPs is based on their toxicity, environmental persistence, tendency to bioaccumulate, and large spatial range. OCPs have often been used in developing countries well after they had been banned in Europe, North America, and Japan, presumably because of lack of access to effective and affordable alternatives, limited capacity for chemical regulation and enforcement, and greater need to deal with insect-borne tropical diseases. The abundance of research efforts in North America, Europe, and the Arctic have allowed the assessment of the exposure of wildlife and humans to OCPs as well as the identification of their spatial patterns, temporal trends, and key environmental pathways. Limited data availability often frustrates similar efforts in developing countries, despite often much longer lasting and more substantial use of OCPs. For example, most studies on OCPs in Central and South America have focused on only a few sampling sites in a small portion of a country and, thus, are unable to evaluate the true spatial distribution of the contaminants (1). Here we show, relying on a study conducted in Costa Rica, how a survey based on passive air and soil sampling affords a cost-effective opportunity to quickly obtain a comprehensive picture of OCP contamination of the physical environment across an entire country. Air and soil constitute the major transport and storage media for OCPs in the environment, respectively. The atmosphere is responsible for dispersing OCPs and, in most cases, also constitutes the entry point to aquatic and agricultural food chains. Air concentrations are highly responsive to changes in emissions and thus lend themselves to assessing the effectiveness of regulation. Because soils typically contain the bulk of the OCP inventory and atmospheric deposition is the only plausible source in soils of nonapplication areas, the soil burden provides a measure of historical atmospheric OCP deposition. Sampling air and soil at the same sites may allow statements on the atmosphere-surface exchange of OCPs. Costa Rica straddles the Central American isthmus at around 9-10° Northern latitude, having coastlines with both the Pacific Ocean and the Caribbean Sea. Its terrain is characterized by coastal plains separated by rugged mountains and spans in elevation from sea level to 3810 m (2). Costa Rica’s climate is tropical and subtropical with a dry season from December to April and a rainy season from May to November. Of Costa Rica’s 4 million inhabitants, 1.2 million live in the central valley, primarily in the capital city of San Jose´ and its surrounding metropolis. Approximately 10% of Costa Rica’s total land area is used for crop production. Tropical agriculture is often very chemical-intensive, reflected in pesticide imports to Costa Rica of up to 9000 metric tons annually (3). Most OCPs were used for agricultural purposes, while DDT was used for malaria control (3). DDT and a series of other OCPs (e.g., aldrin, dieldrin) were banned for use in Costa Rica in 1988, while chlordane and heptachlor were restricted in their use (3). OCP residues have been measured in a variety of samples from Costa Rica, including insect larvae (4), amphibians, turtles, mice and birds (5), human milk (6), and the aquatic environment (7, 8). Few measurements of OCPs in the abiotic environment of Central America have been reported. OCP air concentrations were measured in Belize (9) and Southern Mexico (10). A North American network of passive air samplers included four stations in Central America, including one in Costa Rica (11, 12), reporting annually averaged air concentrations for HCHs (11) and other OCPs (12). Studies 10.1021/es062349d CCC: $37.00
2007 American Chemical Society Published on Web 01/10/2007
TABLE 1. Enantiomer Fractions (EFs) of r-Hexachlorocyclohexane, trans-Chlordane, and cis-Chlordane in Soil and Air Samples from Costa Rica EF of TC chiral column
EF of r-HCH Betadex
Betadex
Manuel Antonio Maritza Cacao Palo Verde La Selva EARTH KeKo¨ ldi Belen Prusia Irazu
0.494 0.509 ( 0.005 0.504 ( 0.007c 0.501c 0.517 ( 0.003 0.507 ( 0.003 0.512
Air 0.485 ( 0.004 0.489 0.490 ( 0.008 0.484 ( 0.002 0.498 0.494 0.483 0.496a
Manuel Antonio Belen Prusia standard 95% CIb
0.509 ( 0.010 0.506 ( 0.002
0.503 ( 0.003 0.496 - 0.504
BGB
EF of CC BGB
0.491 0.500 ( 0.004c 0.501a,c
0.515
0.499 ( 0.002c 0.496 ( 0.005
0.525 ( 0.006 0.511 ( 0.008
0.495 ( 0.001 0.482 ( 0.001
0.508 ( 0.004 0.490 ( 0.006
0.514 ( 0.022 0.482 ( 0.002 0.500a 0.499 - 0.501
0.502a 0.499 - 0.501
0.530
0.468 Soil 0.498c 0.510 ( 0.017 0.482 ( 0.002 0.502 ( 0.002 0.497 - 0.503
a EFs are reported as the average of two duplicates with the associated standard deviation. Where no standard deviation is given, the chemical was only found in one of the duplicates or both duplicates had identical EFs to three decimal places. b 95% confidence interval ) mean EF of standard ( 1.96‚standard error. c EFs in bold can be considered racemic, as they fall within the 95% CI of the standard.
on OCPs in Central and South American soil have been reviewed (13), with most of the work focused on DDT-related compounds. Toxaphene has been measured in the soil and coastal environment of Nicaragua, which borders Costa Rica to the North (14).
Methods Sampling. Table S1 in the Supporting Information lists the coordinates of 23 sampling sites in Costa Rica. These sites are mostly located in protected areas, such as National Parks, Biological Reserves, and research stations, where OCPs had likely not been used in the past. Reflecting the diverse topography of the country (Figure S1), the network includes sites on the Caribbean and Pacific coast as well as sites ranging over 3400 m in elevation. The sites further differ in terms of precipitation, temperature, soil properties, vegetation cover, exposure to prevailing winds, and proximity to source regions. Air was collected with passive samplers consisting of stainless steel mesh cylinders filled with XAD-2 resin and suspended in a steel can with an open bottom (15). An evaluation and calibration of this sampler has been described in detail before (15). It has previously been used to record the spatial distribution of OCPs throughout North America (11, 12). Detail of cleaning, storage, transport, and deployment of the duplicate samplers is given in ref 16. In the vicinity of 20 air sampling sites, 10 individual soil samples were taken, penetrating with an auger to a depth of approximately 25 cm. The samples were mixed in a bucket, and two subsamples were collected for analysis. One to three aliquots from each subsample of soil were analyzed. Details are again given in ref 16. Soil water and organic carbon content (fOC) were determined as described in ref 16. Moisture contents ranged from 1 to 78%. The fOC of dry, ground soil was 0.4-19% ((16) Table S1). The average coefficient of variation for duplicate fOC measurements was 11%. Extraction and Quantification. The XAD-2 resin was transferred to an elution column, solvent-extracted, and fractionated as described previously (15). The conditions used for gas chromatography-electron capture detection and the quality assurance steps, including treatment of procedure, resin, and field blanks have been detailed before (15). Procedural blanks and field blanks for OCPs were low and are given in Table S3.
Aliquots (10-15 g) of wet soil were mixed with sodium sulfate, ground, spiked with recovery standards, Soxhlet extracted with dichloromethane, purified on an alumina column, and analyzed by capillary gas chromatography/mass spectrometry (GC/MS) using an Agilent 6890 GC-5973 MSD, based on a method described elsewhere (16, 17). The OCPs were separated on a DB-5MS column (J&W Scientific; 60 m × 0.25 mm i.d., 0.10 µm film thickness) and analyzed by negative chemical ionization in selected ion monitoring mode. The temperature program was as follows: 90 °C for 1 min, 20 °C min-1 to 160 °C, 2 °C min-1 to 255 °C, 20 °C min-1 to 270 °C, held for 5 min. Recoveries of the labeled spikes ranged from 65% to 115%. The coefficients of variation between replicates for each chemical in air and soil are given in Tables S3 and S4. Average blank values were low and are reported along with method detection limits in Table S4. Chiral Analysis. Only samples with sufficient amounts of R-HCH and trans- (TC) and cis-chlordane (CC) were subjected to chiral analysis by GC/MS using an Agilent 6890 GC-5973 MSD and Hewlett-Packard 5890 GC-59589B MS Engine, operating in electron capture negative ion mode with methane at a flow rate at 1.4 Torr (MS Engine) and 2.2 mL‚min-1 (MSD). Samples were injected splitless (2 mL, split opened after 0.5 min), helium carrier gas was 40 cm‚s-1, injector and transfer line temperature 220 °C, ion source 150 °C, and quadrupole 106 °C. The enantiomers of R-HCH were separated on a BetaDEX-120 column (20% permethylated β-cyclodextrin in polydimethylsiloxane, Supelco, 30 m × 0.25 mm i.d., 0.25 µm film thickness). The enantiomers of CC and TC were separated on Beta-DEX and BGB (30% tertbutyldimethyl-silylated betacyclodextrin in PS-086, BGB Analytik AG, Switzerland, 15 m × 0.25 mm i.d., 0.25 µm film thickness) columns. The temperature program for chiral analysis was as follows: initial 90 °C, 15 °C‚min-1 to 160 °C, 1 °C‚min-1 to 180 °C, hold for 50 min (2 min in case of R-HCH), 20 °C‚min-1 to 230 °C, hold for 5 min. Quality control issues in enantiomeric analysis are precise integration of enantiomer peak areas and elimination of interferences with the enantiomer peaks. Racemic standards were repeatedly injected on the columns to determine the reproducibility of measuring the EFs, and average values are given in Table 1. The criterion used for peak purity in samples was an agreement with the target/qualifying ion ratio within (5% of the standard values VOL. 41, NO. 4, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1125
(18). The ions monitored were 255/257/253 for R-HCH and 408/410/412 for TC and CC. Calculation of Fugacity Fractions. In order to probe the equilibrium status of OCPs between soil and air, fugacity fractions F were derived using equations provided in the Supporting Information. The equations rely on fractions of organic carbon from Table S1 and temperature-dependent octanol-air partition coefficients for OCPs from ref 19. Back Trajectory Analysis. Five-day back trajectories were calculated using the Canadian Meteorological Centre Trajectory Model for 7 stations across Costa Rica at 10, 100, and 200 m above ground level at 6 h intervals for each day the passive air samplers were deployed. The selected stations (# 3, 5, 10, 13, 15, 18, 22) represent a range of geographical locations. This information was used to produce back trajectory probability maps, referred to as “air sheds”, which can identify where the air parcels are most frequently originating.
Results and Discussion OCP Levels. Concentrations of OCPs measured in Costa Rican air and soil are given in the Supporting Information. The passive air sampler concentrations in units of ng‚sampler-1 are presented as the blank corrected average of duplicates in Table S3. Soil concentrations are reported as the blank corrected average of two to four soil aliquots in pg‚g-1 dry weight in Table S4. Multiple analyses of the soil samples yielded the reported coefficients of variation. The concentrations in Costa Rica can be compared to other data from Central and South America. In particular, comparison of the data in Table S3 with air concentrations measured by other means allows the evaluation of the reasonability of the passive air sampling measurements. Time-averaged volumetric air concentrations in pg‚m-3 can be estimated by dividing the sampler concentration (in ng‚sampler-1) by the product of deployment period (365 d) and sampling rate (0.52 m3‚d-1‚ sampler-1) (15). Caution must be employed when interpreting such concentrations, as the sampling rate is somewhat uncertain. Within Central America, average air concentrations for γ-HCH appear to be very similar, with 24 pg/m3 measured in this study, compared with 33 and 51 pg/m3 found using high-volume active samplers in Belize (9). Concentrations of p,p′-DDT, o,p′-DDT, p,p′-DDE, and o,p′-DDD in Costa Rican air are lower than those measured in Belize (9) and Mexico (10), where DDT was used for vector control much more recently than in Costa Rica. The soil concentrations for p,p′DDE and p,p′-DDD in Costa Rica are low, up to 1 ng/g, similar to soil concentrations measured in the Andean soils of Chile (20). Low concentrations are expected in nonagricultural background soils in protected areas. Back Trajectories. The air shed maps show that over the course of a year, most of the air arrives from the northeast (the Caribbean side). Annual and seasonal back trajectory probability maps are shown in Figures S2 and S3. If trajectories are analyzed on a monthly basis a strong seasonal pattern emerges. During the dry season (Nov-April) trajectories are predominantly from the northeast (Figure S3A). For the early part of the wet season (May-August) trajectories are more localized, as the intertropical convergence zone passes by and brings convective rain (Figure S3B). September and October are the months of peak rainfall, as heavy advective rains occur with cyclonic winds drawing moisture off the Pacific. It follows that the trajectories for these 2 months show localized flow as well as transport from the southwest (Pacific side, Figure S3C). The air sheds show that Costa Rica is not strongly influenced by its neighbors. This is important because adjacent Central American countries such as Nicaragua and Panama may have had higher or more recent OCP usage (3). 1126
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 4, 2007
FIGURE 1. Boxplot of OCP concentrations in air (A, ng‚sampler-1) and soil (B, pg‚g-1) measured across Costa Rica. The box is defined by the 25th and 75th percentiles, whiskers mark the 10th and 90th percentiles, and the median is represented by a horizontal line, the mean by a square, and outliers with a diamond. (TC ) transchlordane, CC ) cis-chlordane, TN ) trans-nonachlor). Relative Abundance of Various OCPs. Boxplots are used to illustrate the magnitude and spread of OCP concentrations across Costa Rica (Figure 1). The plot for air shows that the median air concentrations are all quite close, while γ-HCH and p,p′-DDE have the highest mean air concentrations. In soil, p,p′-DDE and p,p′-DDD had the highest mean, despite only being found at five stations. Although dieldrin was only detected in soil at 8 of the 20 sites, it has the next highest average soil concentration. This may seem surprising, considering that dieldrin is banned throughout Central America (13) and other banned OCPs, such as chlordane, are only found at low levels. However, the highest dieldrin soil concentration of 2.0 ng/g (station #21, Prusia) is still much lower than levels found in Columbia, Guatemala, and Panama (25-160 ng/g (13)). Levels of dieldrin in Costa Rican air are also much lower than in nearby countries. The average dieldrin air concentration in Costa Rica was 4.7 pg/m3, in contrast to Belize, where mean levels of 728 pg/m3 and 34 pg/m3 (high-volume, short-term PUF samples) have been reported (9). In both air and soil γ-HCH is relatively abundant, even though Central America has typically lower atmospheric HCH levels than North American sites at higher latitudes (11). Indeed, Costa Rican air and soil have generally low levels of HCHs when compared to the continental average, although a few stations with higher γ-HCH concentrations raise the average (Figure 1) and suggest some continued use of lindane in the country. HCH levels are higher at mid-latitudes than in the tropics because of greater source contribution, i.e., evaporation of “old” residues from Arctic Ocean water to the high latitude atmosphere (R-HCH) (11) and the recent use of lindane in Canadian agriculture (γ-HCH). Higher concentrations of photooxidants in the tropical atmosphere
FIGURE 2. Air and soil concentrations of r- and γ-HCH, trans-chlordane (TC), cis-chlordane (CC), trans-nonachlor (TN), and dieldrin measured across Costa Rica (O ) nondetect). (affecting both HCHs) may also explain why HCH levels are generally lower in the tropics. A closer inspection of Figure 1A reveals that for some substances the median and mean air concentrations are comparable (e.g., R-HCH, aldrin), whereas for others the mean is much higher, often exceeding the 75th percentile (e.g., heptachlor epoxide, p,p′-DDE). The latter is indicative of the occurrence of exceptionally high levels skewing the arithmetic mean. This can be expressed more quantitatively by calculating skewness and kurtosis values for each chemical’s distribution (Table S2). Skewness describes the degree of asymmetry of a distribution around its mean, a large absolute value denoting an asymmetric data set. Kurtosis characterizes the relative flatness or peakedness of a distribution compared with the normal distribution. The further the kurtosis value is from zero, the greater the difference from a normal distribution. The values in Table S2 indeed confirm much lower skewness ( 0.5). The samples selected for chiral analysis are spread across the country, but no strong spatial pattern emerges. If the EFs for R-HCH are plotted against the air concentrations of R-HCH or ΣHCH, no trend is apparent. Neither are the EFs for R-HCH correlated with the measured soil organic carbon content. The Costa Rican R-HCH EFs are similar to those reported earlier for Chetumal and Tapachula in Southern Mexico (0.512 and 0.504, respectively (11)). Shen et al. (11) found air over agricultural areas to be enriched with (+)-R-HCH, and air close to water from which R-HCH evaporates to be depleted in (+)-R-HCH (EF
