Environ. Sci. Technol. 2002, 36, 862-868
Temporal Trends of Organochlorine Pesticides in the Canadian Arctic Atmosphere H . H U N G , * ,† C . J . H A L S A L L , # P. BLANCHARD,† H. H. LI,| P. FELLIN,| G. STERN,§ AND B. ROSENBERG§ Meteorological Service of Canada, Environment Canada, 4905 Dufferin Street, Toronto, ON M3H 5T4, Canada, Environmental Science Department, Lancaster University, Lancaster, LA1 4YQ, U.K., AirZOne, 2240 Speakman Drive, Mississauga, ON L5K 1A9, Canada, and Freshwater Institute, 501 University Crescent, Winnipeg, MB R3T 2N6, Canada
Temperature normalization (TN), multiple linear regression (MLR), and digital filtration (DF) were used to analyze the temporal trends of an atmospheric dataset on organochlorine pesticides (OCs) collected at the Canadian high arctic site of Alert, Nunavut. Details of these techniques have been presented before (Environ. Sci. Technol. 2001, 35, 1303-1311). Both the TN and DF methods revealed that the majority of OC pesticides declined over the 5 years of study, except endosulfan I and several of the pesticide metabolites, including dieldrin and p,p′-DDE. In comparison to studies conducted in the Great Lakes, atmospheric levels in the Arctic were less dependent on temperature, although seasonal variations were apparent. Generally, levels in the winter were lower than during the rest of the year. A notable exception was p,p′-DDE. Many compounds also showed a second minimum in concentrations during June/July and possible explanations are presented to account for this. The estimated first order half-lives for the decline in OC concentrations were generally found to be comparable or slightly longer than those obtained at temperate locations, with the exception of R-HCH, which displayed a much longer half-life in the Arctic (∼17 yrs). Sporadic increases in heptachlor as well as increases in the ratio of trans- to cis-chlordane suggest episodic input of chlordanes between 1995 and 1997, especially during the winter.
Introduction The global distribution of organochlorine pesticides (OCs) has been a growing concern during the last few decades. This is mainly due to their environmental persistence and ability to be transported over great distances from source regions to remote areas, such as the Arctic (1, 2). Most of these compounds, such as DDT and chlordane, have been banned or restricted in western industrialized countries and are included in the recent United Nations Environment Program POPs protocol to eliminate their use (3). However, * Corresponding author phone: (416)739-5944; fax: (416)739-5708; e-mail:
[email protected]. † Meteorological Service of Canada, Environment Canada. # Lancaster University. | AirZOne. § Freshwater Institute. 862
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 5, 2002
many are still being used and/or produced in other parts of the world, e.g. the tropical and subtropical regions, such as parts of Africa (4) and South and Central America (5-7). It is now generally accepted that primary sources of many OC pesticides have diminished in the northern hemisphere. This has been supported by time series generated from lake sediments (8, 9), archived soils (9-11), and several longterm air studies (12-15). These air studies have shown a marked decline in air concentrations over the last 10 years for both OC pesticides and PCBs. Under the Northern Contaminants Program (NCP) of the Department of Indian Affairs and Northern Development, systematic arctic air sampling has been conducted for polychlorinated biphenyls (PCBs), OCs, and polycyclic aromatic hydrocarbons (PAHs) with data collected since 1992 for several locations in the Arctic. This has resulted in a set of measurements that is of the same duration as the temperate monitoring programs cited above, allowing a comparison of temporal trends and half-lives of decline in atmospheric concentrations. This paper is a continuation of a previous study on polychlorinated biphenyls (PCBs) (16) in which digital filtration (DF method) was used to assess the temporal trends at Alert. This station has the longest database of air measurements from the Canadian Arctic. In the previous study (16), two other common temporal trend analysis methods were employed, namely temperature normalization (TN) and multiple linear regression (MLR). Both these methods reduce the intra-annual variability by accounting for temperature-related “seasonality” in the dataset. Both the TN and MLR methods rely on the determination of a linear relationship between the natural logarithm of partial pressure in air, P (Pa), and reciprocal temperature, T (K), and have been shown to be useful in the elucidation of PCB temporal trends at temperate locations (17, 18). Application to arctic PCB data, however, proved difficult due to weak relationships observed between contaminant concentration and ambient temperature, causing many congeners to display statistically insignificant or only weakly significant regressions. In the current study, the same approach was extended to the analysis of temporal trends in the concentrations of OC pesticides measured at Alert. The objectives were to find evidence of declining trends and to compare these trends to those observed in the Great Lakes region over the same time period.
Methodology Sampling and Analysis. The sampling details and sample preparation as well as the analytical methods have been detailed in previous publications (1, 16, 19, 20). In this monitoring program weekly samples are collected at Alert (82°30′N, 62°20′W) since January 1992 resulting in 6 years of data up to the end of 1997. However, data obtained in 1992 were excluded for quality control reasons as discussed by Stern et al. (20). Quality control information, including the incidence of sample train breakthrough and number of samples below the method detection limits (MDL), are given in the Supporting Information (Table SI-1). To retain the maximum amount of information for trend analysis, samples with concentrations below MDLs were included in the analysis. Temporal Trend Analysis. To analyze the long-term trend, intra-annual variability, such as seasonality, must be reduced or eliminated in the dataset. The TN method normalizes the air concentrations to a prescribed temperature, in this case 288 K, the average global tropospheric temperature at ground 10.1021/es011204y CCC: $22.00
2002 American Chemical Society Published on Web 02/02/2002
TABLE 1. Slopes of ln P versus 1/T Plots this study a OCs
m
r2
R-HCH γ-HCH oxychlordane t-chlordane c-chlordane t-nonachlor c-nonachlor heptachlor epoxide dieldrin o,p′-DDE p,p′-DDE o,p′-DDT pentachloroanisole endosulfan I tetrachloroveratrole
227 -27 -2195 -854 -1930 -2870 -3919 -3107 -2943 473 1594 -1246 1102 -2396 690
0.00 0.00 0.24
