Organochlorine pesticides in the ambient air of Chiapas, Mexico

Introduction

Despite being banned for years or even decades in Canada, the U.S. and most European countries, many organochlorine (OC) pesticides such as DDT compounds (DDTs), hexachlorocyclohexanes (HCHs), chlordanes, dieldrin and toxaphene continue being detected in the ambient air on regional and continental scales (Jaward et al., 2004;Shen et al., 2004Shen et al., , 2005Van Drooge et al., 2002). In North America, for example, OC pesticides commonly occur in air and water in the Laurentian Great Lakes on the CanadaeU.S. border (Buehler et al., 2001;Cortes and Hites, 2000;Hoh and Hites, 2004;James et al., 2001;Jantunen and Bidleman, 2003;Marvin et al., 2004;Shen et al., 2004Shen et al., , 2005, in snow and glacial ice in western Canada (Blais et al., 1998;Donald et al., 1999) and in the Canadian Arctic air (Halsall, et al., 1998;Hung et al., 2002Hung et al., , 2004). An unresolved question is whether OC pesticides detected in these environments are due to recycling from local soils or originate from areas of current or very recent use.

Many OC pesticides are known to persist in the environment long after their application and have also been shown to be susceptible to long-range atmospheric transport. Consequently, both old and new sources may account for the OC pesticides detected in North America. In order to distinguish between these contributions, we have been studying the Central AmericaeMexico region to determine its significance as a source region to North America (Alegria et al., 2000). This region has been characterised by high OC pesticide usage in both agriculture and for disease vector control (Castillo et al., 1997;Leonard, 1990;Mora, 1997). Many OC pesticides continue to be used or were used until just recently. Mexico has historically been a significant consumer of DDT. DDT was used in agriculture and for disease vector control in many parts of the country. Chiapas was the last state in Mexico legally allowed to use DDT for agriculture and was one of the largest users for disease vector control. High levels of DDT and its metabolites have been found in human milk and adipose tissue from the state of Veracruz (Waliszewski et al., 2001). Consumption of DDT in Mexico for both agriculture and public health applications was 8e9 kilotonnes (kt) year ÿ1 in 1971e 1972, declined to 3e4 kt year ÿ1 from 1974 to 1981 and reached a minimum of about 0.5e1 kt year ÿ1 from 1982 to 1984. From 1985 to 1991, DDT use increased to 1.5e 2.5 kt year ÿ1 (Lopez-Carrillo et al., 1996). The use of DDT for public health declined from 1.4 kt in 1993 to less than 0.3 kt in 1999 (NACEC, 2004). Under the North American Regional Action Plan (NARAP), Mexico agreed to phase out DDT by 2002, but reported complete phase-out in 2000 (Anonymous, 2001). In 1997, Mexico restricted the use of chlordanes to urban applications as a termiticide. Under the NARAP for chlordanes, Mexico ended their use in 2003 (Moody, 2003).

Toxaphene was legally used in the MexicoeCentral America region until the 1990s (Carvalho et al., 2003;Castillo et al., 1997). Cumulative usage in Mexico (Li and Macdonald, 2005) and Nicaragua (Rodezno, 1997) has been estimated at 71 kt and 79 kt, respectively. This represents approximately 15% of total usage in the U.S., estimated at 490 kt between 1947 and 1986 (Li, 2001). Consequently, the region may represent an important source of atmospheric contamination for North America.

In 1995e1996, concentrations of OC pesticides were measured in air at two locations in Belize, Central America, an inland one in Belmopan and a coastal one in Belize City (Alegria et al., 2000). Results indicated that concentrations of some OC pesticides at the inland location were significantly elevated compared to levels measured in the Great Lakes region (e.g. DDTs, dieldrin and aldrin), supporting the hypothesis that Central America is a source region for North America. In addition, the elevated levels of some presumably banned OC pesticides suggests that they are still being used in the Central America region. To investigate further the source region hypothesis, this study was carried out in southern Mexico, historically the largest user of OC pesticides in the region, to measure their concentrations in ambient air.

Experimental

Sampling collection and preparation

Air samples were collected from the roof of a private residence (10 m) located about 5 km from Tapachula, Chiapas, a city of approximately 250,000 inhabitants located in an area of large-scale agricultural production in the state of Chiapas ( Fig. 1). This residence is fairly new (!10 years old) and was never treated with chlordanes (termiticides), according to the owner. Sampling was carried out for 24 h every 14 days from August 2000 to June 2001. Samples were collected with a high-volume sampler in which w500 m 3 air was drawn through a 10-cm diameter glass fibre filter followed by a polyurethane foam (PUF) trap consisting of two 8.0-cm diameter ! 7.5-cm thick plugs. Filters and PUFs were stored in a freezer until analysed. Ambient temperatures during the sampling period ranged from approximately 299 to 308 K (26e35 C).

Filters and PUF plugs were soxhlet-extracted together overnight (16 h) with petroleum ether. Prior to extraction, PUF plugs were fortified with a mixture containing 20 ng each of a-HCH-d 6 , 13 C 10 -heptachlor exo-epoxide (HEPX), 13 C 10trans-nonachlor (TN) and 13 C 12 -dieldrin, and 100 ng of p,p#-DDT-d 8 which served as surrogates for assessing method recoveries for each sample. Extracts were concentrated by rotary evaporation, blown down with a gentle stream of nitrogen and exchanged into iso-octane. Extracts were cleaned up and fractionated on a column of neutral Al 2 O 3 (1 g, 6% deactivation with H 2 O) overlain with 1 cm anhydrous Na 2 SO 4 . The column was pre-eluted with 10 mL dichloromethane followed by 10 mL petroleum ether. Extracts were applied in w1 mL isooctane and the column was eluted with 15 mL 5% dichloromethane/petroleum ether. The eluate was concentrated by nitrogen blow-down and solvent-exchanged into isooctane. Volumes were adjusted to w1 mL and 13 C 10 -PCB105 was added as an internal standard just prior to analysis.

Analysis

OC pesticides (except toxaphene) were quantified by gas chromatographyeelectron capture negative ion mass spectrometry (GC-ECNI-MS) on an Agilent 6890GC-5973 MS detector with a DB-5 capillary column (J&W, 60 m ! 0.25 mm i.d., 0.25 mm film thickness). Total toxaphene was quantified on a HewlettePackard 5890 GC-5989B MS Engine, using the same type of column, as the sum of the 7-Cl, 8-Cl, and 9-Cl homologues. The detectors in both instruments were operated in the selective ion-monitoring mode to enhance sensitivity. Specific details for the monitored ions and operating conditions can be found elsewhere (Alegria et al., 2000;Jantunen et al., 2000).

Quality control

Recovery experiments were carried out two ways. First, surrogates were added to PUF plugs prior to extraction. Average recoveries of these surrogates were: a-HCH-d 6 56 G 5%, p,p#-DDTd 8 102 G 8%, 13 C 10 -HEPX 72 G 12%, 13 C 10 -TN 90 G 5%, 13 C 12 -dieldrin 76 G 4%. Second, six clean PUF plugs were spiked with a mixture containing the target pesticides and treated as samples. Average recoveries for the second method agreed with the first method, being greater than 70% for all pesticides except for a-HCH (60 G 5%). Calibration standards were also routinely run with each batch of samples.

Front and back PUF plugs were analysed separately to assess breakthrough of gas-phase OC pesticides. Breakthrough was significant only for a-HCH, g-HCH and heptachlor. a-HCH on the back PUF averaged 78% of the front PUF value (42e100%); g-HCH averaged 44% (29e76%); and heptachlor averaged 62% (38e77%). Results for these three pesticides are reported as the sum of both PUF traps.

Three field and three laboratory blanks were run. No peaks matching the target compounds were found on blank PUFs, so limits of detection (LOD) for each pesticide were calculated by finding the lowest detectable concentration of the pesticide injected on the analytical instrument (instrumental detection limit, IDL) and calculating the limit of detection as: LOD Z (average IDL) C (3 ! standard deviation). LODs ranged from 0.06 pg mL ÿ1 (trans-chlordane) to 7 pg mL ÿ1 ( p,p#-DDT). These LODs correspond to 0.1e14 pg m ÿ3 for a nominal air volume of 500 m 3 and a final extract volume of 1000 ml. The quantitation limit for total toxaphene (24 pg m ÿ3 ) was based on the lowest concentration of technical toxaphene standard employed (12 pg mL ÿ1 ). Table 1 gives the concentrations of target OC pesticides in air in Chiapas, Mexico. Because of the wide range of concentrations, both arithmetic (AM) and geometric (GM) means were calculated. GMs were lower than AMs.

Results and discussion

Air concentrations of pesticides in Chiapas

DDTs

Concentrations of total DDTs, sum of p,p#-DDT C o,p#-DDT C p,p#-DDD C o,p#-DDD C p,p#-DDE C o,p#-DDE (SDDT) ranged from 351 to 1548 pg m ÿ3 , although most values were in the upper range (11 of 18 samples had concentrations above 900 pg m ÿ3 and only 2 samples had concentrations below 600 pg m ÿ3 ). AM and GM concentrations were 997 and 927 pg m ÿ3 , respectively. These high concentrations are typical of other tropical regions with ongoing use of DDTs (Larsson et al., 1995;Ngabe and Bidleman, 1992).

The ratio of p,p#-DDT/p,p#-DDE is generally used as an indicator of aging of DDTs. This ratio calculated from mean concentrations was 1.06 based on AM and 1.02 based on GM concentrations; the ratio was O1 for half of the samples (Table 1). Shen et al. (2005) deployed passive air samplers (PAS) across North America during 2000e2001 (see below) and reported ratios of p,p#-DDT/p,p#-DDE Z 0.94 in Tapachula and 1.3 in Chetumal, Mexico, 1.2 in Belmopan, Belize, and 5.8 in Costa Rica. The Mexico and Belize ratios agree well with our results in Tapachula, while the Costa Rica composition indicates a greater input of fresh DDT. Ratios of p,p#-DDT/p,p#-DDE in the ambient air of the U.S. and Canada are generally !1, but occasionally exceed 1 (Aulagnier and Poissant, 2005;Cortes and Hites, 2000;Harner et al., 2004;Park et al., 2001;Shen et al., 2005). The p,p#-DDT/p,p#-DDE w1 in Tapachula is consistent with a mixture of contributions from recent usage and old DDT residues.

HCHs

Average concentrations of HCHs were: a-HCH AM Z 27 pg m ÿ3 , GM Z 22 pg m ÿ3 and g-HCH AM Z 76 pg m ÿ3 , GM Z 72 pg m ÿ3 . The ratios of a-HCH/g-HCH based on AM and GM concentrations were 0.38 and 0.31. This ratio in technical HCH ranges from about 4e8 (Breivik et al., 1999). The much lower a-HCH/g-HCH found here indicates the continued use of lindane (pure g-HCH) in the region. The ratio of~0.3 found here is consistent with results from a 2000e2001 passive air sampling campaign in the region in which a-HCH/g-HCH Z 0.16 in Tapachula and 0.46 in Chetumal, Mexico, 0.18 in Costa Rica and 0.27 in Belmopan, Belize (Shen et al., 2004).

The high proportion of a-HCH on the second PUF trap indicates the possibility that there may have been some breakthrough through both PUF traps and consequently incomplete collection and underestimated air concentrations, which would alter the true ratio of a-HCH/g-HCH. However, the agreement of our ratio of w0.3 with ratios in the region reported by Shen et al. (2004) noted above supports our hypothesis of continued use of lindane in the study region. Waite (2001) reported air concentrations of lindane ranging from !100e2700 pg m ÿ3 at a control site in the Canadian Prairies 2 km from a canola field that had been planted with lindane-treated seeds. The control site had not been planted with lindane-treated seeds, so the air concentrations are thought to represent background air concentrations for the region during the canola-growing season. The mean levels at our study site (103 pg m ÿ3 ) agreed with the lower end of the range, supporting our hypothesis of continued use of lindane in the study region.

Chlordanes

The AM and GM concentrations of total chlordanes, sum of heptachlor C HEPX C TC C CC C TN C CN (S chlordanes), were 201 and 140 pg m ÿ3 , respectively. In the environment, TC degrades more rapidly than CC (Eitzer et al., 2001), and a ratio of TC/CC !1 is generally taken to be indicative of aged chlordanes (Bidleman et al., 2002). In the technical chlordane used in the United States, the proportions of TC/ CC were reported as 1.00/0.85 by one laboratory and 1.00/1.05 by another (Mattina et al., 1999). The relative liquid phase vapour pressures of TC/CC at 25 C are 1.00/0.72 (Hinckley et al., 1990). Multiplying the technical chlordane composition by the relative volatility of the components gives the relative proportions of TC/CC expected in technical chlordane vapour: 1.00/0.61 or 1.00/0.76, depending on which set of technical chlordane compositional values is used. The proportions in ambient air from the AM and GM concentrations measured in this study were 1.00/.0.75 and 1.00/0.68 (Table 1). The ratio of TC/CC in air is quite similar to the expected vapour composition, indicating fresh chlordanes. Similarly, the ratio of TN/TC in technical chlordane is reported as 0.42 and 0.76 (Mattina et al., 1999), the ratio of TN/TC vapour pressures is 0.57 (Hinckley et al., 1990), and the predicted ratio of TN/TC in vapour-phase technical chlordane is 0.24e0.43. Ratios of TN/TC in air from the AM and GM concentrations in Table 1 are 0.63 and 0.53, which suggests a slight enrichment of TN. This was also found by Eitzer et al. (2003), who measured higher than expected TN proportions in air above soils containing weathered chlordane residues.

Heptachlor was detected in all samples, with AM and GM concentrations of 57 and 43 pg m ÿ3 . Concentrations showed considerable variability, ranging from 6 to 154 pg m ÿ3 . Heptachlor and Schlordane concentrations were closely correlated (r 2 Z 0.68), not surprisingly since heptachlor is a component of technical chlordane. The metabolite HEPX was quantifiable in 10 samples with AM and GM concentrations of 18 and 14 pg m ÿ3 , excluding samples below detection.

Toxaphene

Total toxaphene AM and GM concentrations were 505 and 491 pg m ÿ3 . Toxaphene is a complex mixture of several hundred compounds (Hainzl et al., 1994). In addition to quantifying total toxaphene as the sum of the Cl 7 , Cl 8 , and Cl 9 homologues, the profiles of octachlorobornane congeners in air were compared with that of technical toxaphene. Fig. 2 shows typical results. The six peaks labelled in Fig. 2 are mixtures of several co-eluting components as shown by multidimensional GC investigations of technical toxaphene and air samples collected near Lake Ontario (Shoeib et al., 2000). Prominent components of these peaks are the following congeners, identified using single congener standards (Bidleman and Leone, 2004b) and designated by AndrewseVetter nomenclature (Andrews and Vetter, 1995;Vetter and Oehme, 2000): 1 Z B8-1413, 2 Z B8-1412, 3 Z B8-531, 4 Z B8-1414 C B8-1945, 5 Z B8-806 C B8-809, 6 Z B8-2229 C unknown component. Peaks 3 and 5 are depleted in the air samples when compared to the technical mixture. This same pattern has been seen in air and soils in the southern United States (Bidleman and Leone, 2004b;Bidleman et al., 2004;Harner, 1999;Jantunen et al., 2000) and indicates that the toxaphene is from weathered residues and not from fresh sources. Toxaphene has been reported in Nicaraguan cotton field soils (Carvalho et al., 2003) at concentrations similar to those in the southern United States (Bidleman and Leone, 2004a, b), suggesting soils in the region are potential sources of toxaphene in ambient air.

Endosulfan-I

AM and GM concentrations of endosulfan-I were 367 and 287 pg m ÿ3 . Samples 1, 2 and 15 showed higher levels, reflecting a pulsed input due to usage nearby at or close to the time of sampling. Endosulfan is still used in Mexico, including the Chiapas region. It has been found in surface waters in the state (Hernandez-Romero et al., 2004). The generally high concentrations measured during the entire sampling period are indicative of the quantities used in the region.

Dieldrin and aldrin

Dieldrin was detected in all samples, but at very low levels, averaging AM Z 15 pg m ÿ3 GM Z 10 pg m ÿ3 . Dieldrin is legally banned in Mexico and all neighbouring countries, which explains the low levels in this study. Aldrin was below detection in all samples. Aldrin is also banned in Mexico and surrounding countries.

Potential sources of pesticides

In an effort to determine the potential sources of pesticides measured in air in Chiapas, air parcel back trajectories were calculated (at several altitudes) using the U.S. National Atmospheric and Oceanic Administration (NOAA) Hysplit4 model. The predominant airflow directions were from the E-NE (passing through Guatemala, Honduras, El Salvador, and sometimes other areas of southern Mexico and Nicaragua) and SE (blowing from the coast but having usually passed through Guatemala and Honduras). Some of the air masses were tracked back to the southern United States and Cuba. No clear relationships emerged to indicate a predominant source direction for the OC pesticides. The fact that air parcels for all samples passed through southern Mexico and, in most cases, parts of Central America before arriving at Tapachula makes it difficult to distinguish between local/regional sources and distant transport on the basis of air pathways alone. As noted earlier, the p,p#-DDT/p,p#-DDE ratio w1 and the relatively high p,p#-DDT concentrations suggest a combination of current usage and re-emission sources, TC/CC ratios indicate current usage and/or unweathered termiticide emissions, while elevated TN proportions and toxaphene congener profiles are consistent with soil emissions. It is likely that the pesticides measured in this study represented a mixture of fresh and weathered sources. This may reflect the fact that air masses travelled over regions where OC pesticides are still being used and emitted from weathered sources (Central America and southern Mexico) as well as regions where only weathered sources (southern United States) are expected. Our results are thus consistent with both regional and longer-range atmospheric transport.

A confounding factor in trying to determine sources of pesticides measured in air is their unauthorised use in the region. In personal conversations with farmers in the state, we ascertained that DDT and toxaphene were being used in agriculture despite official bans. Sources were variously ascribed to existing stockpiles and purchase across the border in Guatemala. We caution that we did not carry out a formal, scientific survey of farmers, but simply spoke with several farmers encountered while travelling near Tapachula. In addition, it was unclear after conversations with public health workers whether DDT was still being used in 2000e2001. Table 2 compares levels of selected pesticides in Chiapas with other measurements in potential source and receptor regions.

Comparison to other studies and regions

In 1995e1996, OC pesticide concentrations were measured in Belize, Central America, (Alegria et al., 2000). Interestingly, there are marked, statistically significant differences (unpaired t-tests at 95% confidence interval using Excel software) between Chiapas and Belize even though the spatial distance between sampling sites is not very large (w500 km) ( Table 2). Whereas in Belize levels of dieldrin and aldrin were greatly elevated (AMs of 775 pg m ÿ3 and 657 pg m ÿ3 , respectively in Belmopan), dieldrin levels in Chiapas were moderate (AM of 15 pg m ÿ3 ) and aldrin was below detection in all samples. This marked difference in dieldrin concentrations was not simply due to the 4e5 year difference between sampling at the two locations. In their 2000e2001 passive air sampling campaign, Shen et al. (2005) also found much higher levels of dieldrin in Belmopan, Belize (260 pg m ÿ3 ) compared to levels in Tapachula (AM 15 pg m ÿ3 ). In our studies, toxaphene was an order of magnitude higher (AMs 505 pg m ÿ3 vs. 35 and 29 pg m ÿ3 ) and chlordanes were 2e3 times higher (AMs 200 pg m ÿ3 vs. 76 and 78 pg m ÿ3 ) in Chiapas than in Belize. Concentrations of DDTs were statistically similar (unpaired ttests at 95% confidence interval using Excel software) in both areas.

DDTs in Chiapas and Belize were, for the most part, much higher than those measured in the southern United States. Exceptions were in Mississippi (Coupe et al., 2000) and Arkansas (Hoh and Hites, 2004) where concentrations of p,p#-DDE rivalled those in Chiapas and Belize. DDTs in Chiapas and Belize were 1e2 orders of magnitude above levels in the Great Lakes and 3 orders of magnitude above those in the Canadian Arctic (Table 2). Thus, the southern Mexicoe Central America region may be a source of DDTs to the rest of North America.

Toxaphene and chlordane were also higher in Chiapas than in the Great Lakes and Arctic receptor regions. However, concentrations of both pesticides in the air of the southern United States were of the same magnitude as those in Chiapas. In addition, levels in the southern states were about 10-fold higher than those in Belize ( Table 2). The toxaphene congener profiles in both Chiapas and the southern states suggest a weathered source, probably emission from soils. Thus, depending on the direction of air mass movement, southern Mexico and Central America may be sometimes a source and sometimes a receptor region for toxaphene and chlordanes (impacted by the southern United States). Table 2 also compares our measurements in Chiapas and Belize to results of a North American wide campaign with PAS carried out in 2000e2001 by Shen et al. (2004Shen et al. ( , 2005. The PAS consisted of XAD-2 resin contained in porous stainless steel tubes. Deployed for one year, each PAS sampled approximately 190 m 3 of air (Shen et al., 2005) and yielded one integrated air concentration for the year. The two sampling programs gave air concentrations agreeing within a factor of two for DDTs and HCHs in Chiapas and for DDTs and dieldrin in Belize. Concentrations of a-endosulfan in Chiapas also agreed excellently between our measurements (AM Z 367 pg m ÿ3 ) and the PAS campaign (350 pg m ÿ3 ). However, PAS-derived air concentrations were lower for HCHs in Belize, dieldrin in Chiapas and chlordanes in both locations.

Conclusions

Measurements in Chiapas, Mexico show that concentrations of several OC pesticides (DDTs, chlordanes, and toxaphene) are elevated in ambient air compared to those in the Great Lakes region, while other pesticides are not (HCHs and dieldrin). While this suggests southern Mexico as a source region for the former group of chemicals, comparably high levels have also been reported in parts of the southern U.S. where their suspected sources are soil emissions (DDTs, toxaphene) and termiticide usage (chlordanes). Ratios of p,p#-DDT/p,p#-DDE and TC/CC in Chiapas suggest a mixture of fresh and weathered sources, while congener profiles of toxaphene suggest emission of old residue from soils. This is supported by air parcel back trajectory analysis, which indicated that air masses over Chiapas at the time of sampling passed over areas of continuing or recent use of some OC pesticides as well as areas of past use.

The measurements presented here represent a further step in clarifying the significance of Central America and Mexico as a region for OC pesticide emission. A more comprehensive sampling programme is needed to obtain a clearer spatial picture for Mexico, to distinguish emissions from current usage and legacy sources, and to define regional differences such as those seen between Chiapas and Belize. Such information would allow emissions from the region to be placed in perspective with the emissions of old OC pesticide residues from soils in the United States and Canada.