JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (22)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Laabs, V.
Right arrow Articles by Zech, W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Laabs, V.
Right arrow Articles by Zech, W.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Laabs, V.
Right arrow Articles by Zech, W.
Related Collections
Right arrow Organic Compounds
Right arrow Agricultural Pesticides
Right arrow Air Pollution
Right arrow Water Pollution
Journal of Environmental Quality 31:1636-1648 (2002)
© 2002 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORTS
Organic Compounds in the Environment

Pesticides in Surface Water, Sediment, and Rainfall of the Northeastern Pantanal Basin, Brazil

Volker Laabs*,a,d, Wulf Amelunga, Alicio A. Pintob, Matthias Wantzenc, Carolina J. da Silvab and Wolfgang Zecha

a Institute of Soil Science and Soil Geography, University of Bayreuth, D-95440 Bayreuth, Germany
b Projeto Ecologia do Gran Pantanal, Instituto de Biociências, Universidade Federal de Mato Grosso, 78090-000 Cuiabá, MT, Brazil
c Max-Planck Institute of Limnology, Tropical Ecology Working Group, D-24302 Plön, Germany
d Covance Laboratories GmbH, Kesselfeld 29, 48163 Münster, Germany

* Corresponding author (volker.laabs{at}covance.com)

Received for publication July 9, 2001.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Within the last 25 years an intensive agriculture has developed in the highland regions of Mato Grosso state (Brazil), which involves frequent pesticide use in highly mechanized cash-crop cultures. To provide information on pesticide distribution and dynamics in the northeastern Pantanal basin (located in southern Mato Grosso), we monitored 29 pesticides and 3 metabolites in surface water, sediment, and rainwater of the study area during the main application season. In environmental samples, 19 pesticides and 3 metabolites were detected in measurable quantities, resulting in at least one pesticide detection in 68% of surface water samples (n = 139), 62% of sediment samples (n = 26), and 87% of rainwater samples (n = 91). Surface water samples were most frequently contaminated by endosulfan compounds ({alpha}-, ß-, -sulfate), ametryn, metolachlor, and metribuzin, although in low (<0.1 µg L-1) concentrations. Sediment samples exhibited concentrations up to 4.5 µg kg-1 of p,p'-DDT, p,p'-DDE, endosulfan-sulfate, ß-endosulfan, and ametryn. In contrast, rainwater was polluted with substantial amounts of endosulfan, alachlor, metolachlor, trifluralin, monocrotofos, and profenofos (maximum concentrations = 0.3 to 2.3 µg L-1) in the highlands. Lowland rainwater samples taken 75 km from the next application area contained 5- to 10-fold lower mean pesticide concentration than in the highlands. Cumulative deposition rates of the pesticide sum within the study period ranged from 423 µg m-2 in the highlands to 14 µg m-2 in the lowlands. The atmospheric input of pesticides to ecosystems seemed to be of higher relevance in the tropical study area than known from temperate regions.

Abbreviations: GC, gas chromatography • MS, mass spectrometry


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
PESTICIDE USE HAS increased worldwide to secure the food supply of the swelling global population. Especially in tropical regions, agricultural intensification has led to higher pesticide consumption (Racke et al., 1997). Although it is undisputed that pesticides are essential in modern agriculture, there is growing concern about possible environmental contamination from agrochemicals. In this respect, extensive research has been conducted in temperate regions, assessing the pollution of rainwater (e.g., Goolsby et al., 1997; Dubus et al., 2000), surface waters (e.g., Pereira and Hostettler, 1993; Thurman et al., 2000), and sediments (e.g., Bester and Hühnerfuss, 1996; Daniels et al., 2000) with pesticides. In contrast, studies on pesticide occurrence in environmental samples from the tropics are scarce and focus mainly on organochlorine pesticides (e.g., Caldas et al., 1999; Botello et al., 2000), which have been banned from use in most countries during the last two decades. A recent monitoring of currently used pesticides in Central American river ecosystems (Castillo et al., 2000; Castilho et al., 2000) demonstrated that these substances are also relevant contaminants in the aquatic environment. As air and soil temperatures are higher in tropical climates than in temperate regions, an increased pesticide loss to the atmosphere by volatilization has to be expected. However, rainwater contamination with pesticides in tropical environments has so far not received due attention. In view of the frequent pesticide detection in rainfall of pristine environments (Bester and Hühnerfuss, 2000) or even in surface water of the polar regions (Chernyak et al., 1996), pesticide atmospheric deposition needs to be considered as a relevant pesticide input pathway for ecosystems worldwide.

Brazil is now ranked as the fouth largest national market for agrochemicals in the world (Racke et al., 1997). The Brazilian agricultural frontier is located in the Cerrado Highlands (planalto), where an intensive, mechanized agriculture has been spreading during the last 25 yr (Resck, 1998). Many streams of the southern highlands of Mato Grosso state (central-western Brazil) drain toward the Pantanal, which is the world's largest freshwater wetland and a hotspot of biodiversity. Although pesticide application on the highlands has intensified considerably since the introduction of soybean [Glycine max (L.) Merr.] and recently cotton (Gossypium hirsutum L.) to this region, no studies on the environmental dispersion of pesticides in this region or the nearby Pantanal area exist. The objective of our pilot study was to assess environmental concentrations of currently used pesticides in the northeastern Pantanal basin. To this aim, pesticide concentrations were measured in surface water, sediment, and rainwater samples of the agricultural highland areas and in the Pantanal border regions during the main agrochemical application season.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area and Sampling Schedule
Sampling was restricted to the northeastern Pantanal basin, concentrating on the region marked off by the Cuiabá River in the west, the São Lourenço River in the south and east, and the interstate highway BR364 in the north (Fig. 1) . The climate in this area is of the savanna type, with mean annual air temperatures of 23°C in the planalto and 25°C in the lowlands, and a mean annual precipitation of 1900 mm in the planalto and 1500 mm in the lowlands. The monitoring period (November to March) was selected according to the main pesticide application season and coincided with the rainy season in the study region. Sampling locations for surface water (streams [Sites S1–S6] and rivers [Sites R1–R6]) and sediment (Sites R1–R6) targeted tributaries of the São Lourenço River (e.g., the Tenente Amaral), the São Lourenço River itself, the Cuiabá River, the mouth of the Mutum River at the Sia Mariana lake, and the Sia Mariana lake itself. Rainwater collectors were installed within the agriculturally used highlands between São Vincente and Jaciara (P1 and P2), in the Cuiabá city area (P3 and P4), and in the city area of Barão de Melgaço (P5), a border town of the Pantanal. The sampling was done on a regular basis: weekly for stream water, biweekly for river and lake water, once per month for sediment samples, and three times per week for rainwater. The sampling period began on 10 Nov. 1999 and ended on 2 Mar. 2000.



View larger version (123K):
[in this window]
[in a new window]
  		Fig. 1. Study area with sampling locations (S, streams; R, rivers; P, precipitation).

 
Spectrum of Commercialized and Analyzed Pesticides
The study region shows a dominant division between the planalto area, which is intensively used for cattle and cash crop production, and the lowland regions (Fig. 1), which are used (if at all) as pastures or, in the greater Cuiabá city area, for small-holder vegetable production. Consequently, the spectrum of pesticides in the highlands is dominated by herbicides and insecticides applied in soybean, cotton, corn (Zea mays L.), and sugarcane (Saccharum spp.) cultures (e.g., trifluralin, metolachlor, atrazine, endosulfan, monocrotofos), whereas in the lowlands mainly insecticides and fungicides used for vegetable production (e.g., parathion-methyl, chlorpyrifos, tebuconazole) are sold in small quantities (for all pesticide chemical names, see Table 1). For the monitoring we selected pesticides that were of the most frequent use in the study region, according to a survey among regional pesticide vendors, and that could be measured with gas chromatography–mass spectrometry (GC–MS) (Table 2). The spectrum of analyzed substances was completed by persistent organochlorine pesticides, which have been banned from agricultural use for more than a decade in Brazil. Due to the analytical restrictions, ionic and thermally unstable pesticides could not be covered by our monitoring study, although some of them were also frequently used in the study area (e.g., glyphosate, paraquat dichloride, 2,4-D, acifluorfen, chlorimuron-ethyl, diuron, methomyl, teflubenzuron).


View this table:
[in this window]
[in a new window]
  		Table 1. Chemical names of pesticides discussed in this paper.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Studied pesticides and metabolites, monitored ions, recovery (standard error), and routine limit of quantification (RLQ) in water and sediment samples.

 
Sample Collection
River and stream water samples were taken from 0.3 m below the water surface with amber 1-L glass bottles that had been rinsed with ethyl acetate and were heated at 250°C (4 h) before use. For sampling, turbulent midstream positions of water bodies were chosen to approximate mean concentrations of stream and river water. At Site R6, water was sampled from the middle of the lake, at 0.3 m below the surface. After collection, water samples were stored on ice during transport (<6 h) and were kept at 4°C in the laboratory (<2 d) until solid-phase extraction. Sediment samples were taken from river positions where an accumulation of fine-textured substrate (loamy to clayey texture, organic carbon content: 6 to 46 g organic C g-1 sediment dry mass) took place, with a Van Veen sampler. Composite samples were pooled from four or five subsamples (0–10 cm) of a 10-m2 area, homogenized, and wrapped in aluminum foil. Samples were immediately stored on ice (<6 h) after collection and stored at -20°C in the laboratory until analysis. Rainwater samples were collected with stainless steel funnels (110-mm diameter) mounted on 4-L amber glass bottles, which had been wrapped with aluminum foil to reduce solar heating. Due to the sampler design, the bulk deposition of pesticides was measured, consisting of wet and dry deposition. Rain collectors were installed in the open (1 m above ground), taking care to avoid influence of adjacent trees or buildings. In the agriculturally used highlands, samplers were installed on grassland areas, at least 75 m from the nearest cropped area to minimize any direct spray drift contamination. Collector flasks were emptied daily and water was pooled into samples for every two or three days; at Site P5 water was pooled into weekly composite samples, due to shortages of freezer space. After collection, rainwater samples were immediately stored in a freezer at -20°C until extraction. When less than 50 mL of rainwater were collected during a sampling interval, the sample was unified with the one of the following sampling interval.

Sample Extraction
Water samples (<=1 L) were filtered (Schleicher&Schuell GF6 glass fiber filter; Laborcenter, Nürnberg, Germany), conditioned (acidification with hydrochloric acid to pH 3, addition of 20 g L-1 KCl), and afterwards solid-phase extracted on 8-mL glass C18-cartridges (filled with 1.25 g of C18 Bakerbond; Baker Chemicals, Gross-Gerau, Germany), which had been conditioned by elution with 5 mL each of ethylacetate, methanol, and water. After extraction of the water sample (approximately 15 mL min-1) the cartridges were dried in an air stream and stored at -20°C until elution. Pesticides were eluted from the cartridges by n-hexane (8 mL) and ethyl acetate (12 mL) and the concentrated eluate was then analyzed by GC–MS. Further methodological details may be found in Laabs et al. (2000). Filter residues of samples with positive pesticide detection in the water phase were shaken for 4 h with acetone–hexane (1:1, v/v, 25 mL); the concentrated solvent extract was then analyzed for pesticides.

Sediment samples (approximately 25 g dry weight) were extracted with a mixture of acetone, ethylacetate, and water (Laabs et al., 1999). After extraction the shaking flasks were centrifuged and the sample extract was then decanted and filtrated (Schleicher&Schuell 5961/2). Following the removal of the organic solvents with a rotary evaporator, the remaining water phase was liquid–liquid extracted with dichloromethane (3 x 25 mL). The organic phase was then dried with sodium sulfate and concentrated again with a rotary evaporator. Afterwards the extract was purified by flash chromatography with an 8-mL glass column packed with 1.0 g of aluminum oxide (deactivated with 0.06 g water g-1 sorbent) on top of 1.0 g Florisil (deactivated with 0.1 g water g-1 sorbent) (Promochem, Wesel, Germany). After the sample extract had been transferred to the column in approximately 0.5 mL of toluene, the pesticides were consecutively eluted from the column with 10 mL of n-hexane and 10 mL of n-hexane–diethyl ether 1:1 (v/v). The eluate was concentrated with a rotary evaporator and then analyzed by GC–MS. All organic solvents used for analysis were of "picograde" purity (Promochem, Wesel, Germany). Glassware used during sample processing was rinsed with ethylacetate and baked at 250°C (>4 h) before every use.

Analysis of Pesticides and Quality Control
Pesticides were quantified with GC–MS (Hewlett-Packard 6890 series gas chromatograph coupled with a Hewlett-Packard 5972A mass selective detector–electron impact ionization; Hewlett Packard GmbH, Waldbronn, Germany), which was operated in the selected ion-monitoring mode at the following conditions: injector block temperature = 250°C; carrier gas = helium; oven temperature program = initial temperature of 92°C held for 2.5 min, heating up to 175°C at 15°C min-1, 175°C held for 14.5 min, heating up to 280°C at 15°C min-1, 280°C held for 9 min; and transfer-line temperature = 290°C. Quantification and identification of substances were achieved by measuring one target and two additional qualifier ions, respectively, per substance (Table 2). The calibration was performed with three-point linear functions, with external and internal standards. Internal standards ({alpha}-HCH, terbuthylazine, ditalimfos) were added (1 µg each) to samples before extraction to compensate for processing losses and to control the analytical quality. Internal standard recovery was calculated by relating their concentration to fluorene-d10, which was added (1 µg) to concentrated sample extracts prior to GC injection. Samples were injected in toluene (1 µL). The routine limits of quantification (RLQ) are listed for all compounds in Table 2.

The three internal standards were chosen to represent pesticides of different chemical and polarity classes during extraction and analysis. These pesticides were of no current use in the studied area, as terbuthylazine and ditalimfos are not commercialized in Brazil and lindane (and its potential byproduct {alpha}-HCH) was banned from agricultural use more than 10 yr ago. Eventual traces of the persistent {alpha}-HCH in the low nanogram range in samples would not interfere with pesticide quantification by this internal standard, as to every sample fixed quantities of internal standard were added in high amounts (1 µg per sample).

Individual sample processing quality was checked by calculating the recovery of internal standards (usually >75% of the spiked amount). Pesticide peaks in sample extracts were identified by retention time, presence of target and qualifier ions, and respective ratios of target and qualifier ions. Whenever possible (at concentrations greater than 0.5 µg mL-1 sample extract), full-scan spectra of substances were recorded to fingerprint pesticides and identify unknown compounds by comparison with reference spectra of spectral libraries.

During the extraction of water samples, blanks were processed regularly, in which of all pesticides only alachlor and metolachlor were infrequently detected in low concentrations (<0.005 µg L-1) in 1-L samples of distilled laboratory tap water. Malathion occurred in higher concentrations in blanks (<0.050 µg L-1) during the antimosquito spray campaign in the Cuiabá city area (January to February 2000). Blank concentrations were subtracted from calculated sample concentrations for the respective pesticides and processing lots, as they were thought to derive from condensation of pesticides on surfaces of glass vessels within the laboratory.

Pesticide Recovery
The quality of the analytical methods was assessed by pesticide recovery experiments with water and sediment matrices (Table 2). The efficiency of the water extraction procedure was tested by spiking 1-L samples of distilled tap water with 0.1 µg each of pesticide standards dissolved in acetone. Afterwards the samples were extracted and analyzed according to the procedure outlined above. Similarly, uncontaminated sediment from Site R4 (texture: clay; organic C: 9.1 g kg-1; pH 5 [in H2O]) was spiked with known quantities of pesticides before extraction (2.5 µg per substance in 25 g of sediment dry mass) and processed and analyzed as described above. Pesticide concentrations measured in environmental samples were corrected for their respective recovery values (Table 2) determined during these experiments.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Surface Water
Stream water samples of Sites S1 to S6 generally did not contain significant amounts of suspended sediments, although turbidity in these streams is known to increase strongly during rain events (Wantzen, 1998). We attributed this finding to the fact that most of the stream banks in the study area were covered by gallery forest or swampy grassland, which usually accompanied the streams on both sides in a 5- to 50-m-wide strip. Thus a continuous input of soil particles to the water body by surface runoff was inhibited, except during heavy storm events. In contrast, river water samples were mostly clouded by suspended sediments (Sites R1–R4) or were rich in dissolved organic carbon (Sites R5 and R6), as indicated by their brown color after filtration. Suspended sediment of water samples, however, never contained pesticides in detectable amounts. We attributed this on the one hand to the small mass of sediment suspended in river water (<5 g L-1), which led to pesticide concentrations below the analytical detection limit, and on the other hand to a possible desorption of the more polar pesticides from sediment particles during their riverine transport (Squillace and Thurman, 1992).

The concentrations of individual pesticides measured in surface water of the study area (Table 3) never exceeded 0.1 µg L-1 (established limit by laws of the European Community), with the exception of malathion in the Cuiabá River, and lay below the respective maximum contaminant levels of pesticides (e.g., 2 to 4 µg L-1 for alachlor, atrazine, and simazine) established for drinking water by the USEPA (2000). The current Brazilian legislature (Ministério da Saúde, 1990) does not define maximum contaminant levels of the studied pesticides in water resources, except for some formerly used organochlorine pesticides (e.g., DDT, heptachlor, lindane), which were not detected in water samples in our study.


View this table:
[in this window]
[in a new window]
  		Table 3. Detection frequency and pesticide and metabolite concentrations in surface water samples.

 
The sampled streams drained catchments of intensive agricultural use, where 40 to 60% of the land surface was used for soybean, corn, sugarcane, and cotton production. In 68% of stream water samples, one to eight pesticides were detected per sample. Individual detection frequencies and concentration spans of pesticides are presented in Table 3. The highest maximum concentrations were found for endosulfan and its metabolite endosulfan-sulfate, as well as for monocrotofos, which was only detected twice during the sampling period. Besides endosulfan, the greatest percentages of positive detections were recorded for the herbicides metolachlor, metribuzin, and ametryn (>12%). The other substances were detected in less than 6% of the analyzed samples. The concentrations of the most frequently detected pesticides at Site S2 are shown in Fig. 2 .



View larger version (37K):
[in this window]
[in a new window]
  		Fig. 2. Concentrations of selected pesticides and metabolites in stream water at Site S2.

 
Endosulfan was applied in the study area onto cotton fields by plane, which presumably accounts for the frequent detection of this nonpolar insecticide (water solubility of 0.32 mg L-1) in water samples. Probably, spray drift and wash-off from stream bank vegetation caused a major part of the observed contamination. The frequent detection of more polar substances like metribuzin and metolachlor (Fig. 2, Table 3) at low concentrations throughout the study period points to a continuous input of pesticides by subsurface flow from fields to the surface water. These assumptions were supported by the finding that these pesticides were detected in water samples free of suspended sediment, indicating that surface runoff from fields did not contribute to the measured contamination of streams with pesticides in this case. In contrast, readily degradable substances, such as monocrotofos and alachlor, or apolar substances, such as tebuconazole and trifluralin, were only detected at a few sampling dates. In general, the measured pesticide concentrations in streams were low (<0.04 µg L-1). We attributed this finding in part to the stream discharge-independent sampling strategy, which was not able to seize short-term peak concentrations related to runoff events. It is known that pesticide concentrations during discharge events in streams may increase 20-fold as compared with their concentrations at base flow (Spalding and Snow, 1989).

In comparison with monitoring data from the midwestern USA (Spalding and Snow, 1989; Schottler et al., 1994), concentrations of typical pollutants such as alachlor, atrazine, and metolachlor were at least 5- to 10-fold lower in base flow of streams in our study. This might be explained by the higher intensity and uniformity of land use in the North American region, where regionally up to 80% of the land surface are used to grow corn and soybean (Spalding and Snow, 1989), resulting in a more intensive use of the respective herbicides in these areas. In addition, effective anti-erosion measures in Brazilian agriculture and the fast dissipation of pesticides in tropical soils after application (Laabs et al., 2000) might have contributed to the low contamination level of streams with pesticides in our study area. Our results compare well with stream water concentrations of endosulfan ({alpha}-, ß-) and ametryn reported in monitoring studies from tropical environments in Central America (Castillo et al., 2000; Castilho et al., 2000).

River water was sampled only biweekly, as the dynamics of pesticides in larger water bodies are expected to be less rapid than in smaller streams (Baker and Richards, 2000). Of all samples taken, 81% were contaminated by pesticides, with up to five compounds present in one sample. The most frequently detected insecticides and metabolites were endosulfan-sulfate and malathion, while the most frequently detected herbicides in samples were ametryn, metolachlor, and simazine (Table 3). It is noteworthy that also the herbicides alachlor, atrazine, and the nonpolar trifluralin were detected in more than 15% of the analyzed samples. The frequent detection of malathion at high concentrations in river water (restricted to Sites R3 and R4 of the Cuiabá River) was attributed to its use during antimosquito campaigns in the city area of Cuiabá (January to February 2000). In Fig. 3 , the concentrations of the most frequently detected pesticides are shown for Site R1. The high detection frequency and high concentrations of ametryn in river water samples at Sites R1 and R2 was attributed to the regionally intense cultivation of sugarcane in the upper basin of the São Lourenço River. There, ametryn was exclusively used in large-sized sugarcane plantations, where it was also applied in direct vicinity to the river banks. Otherwise, the pesticide spectrum at Site R1 was characterized by the frequent detection of endosulfan compounds and sporadic occurrences of triazines (simazine, atrazine).



View larger version (36K):
[in this window]
[in a new window]
  		Fig. 3. Concentrations of selected pesticides and metabolites in river water at Site R1.

 
A smaller number of pesticides was detected in river water (10 different compounds) than in stream water (15 different compounds). We attributed this finding to an elimination of easily degradable substances such as monocrotofos and profenofos (Hornsby et al., 1996) during the riverine transport, or to dilution effects (e.g., chlorothalonil, tebuconazole). Some herbicides (e.g., ametryn, atrazine, metolachlor, simazine) were detected more often and at similar or higher concentrations in river water samples (Sites R1–R4) than in stream water samples (Sites S1–S6). These results corroborated the finding of Baker and Richards (2000) that generally higher intermediate concentrations of pesticides are found in rivers than in smaller streams of a watershed ("peak broadening"). Yet, also a more intensive use of these herbicides in other subcatchments of the greater river watershed than in the sampled ones can explain our results. The lower ratios of endosulfan ({alpha}- + ß-) to endosulfan-sulfate concentrations in water samples of rivers (mean ratio of 0.18) in comparison with stream water (mean ratio of 0.50) pointed to an increasing degradation of endosulfan during its transport toward the lowlands in surface water. Endosulfan-sulfate was shown to be the main metabolite of endosulfan degradation in aquatic systems (Navarro et al., 2000).

In comparison with river monitoring studies from temperate regions (Pereira and Hostettler, 1993; Griffini et al., 1997), measured mean concentrations in tropical river water were 5- to 10-fold lower for many compounds (alachlor, atrazine, metolachlor, simazine). This may be attributed to a generally higher land-use intensity in such agricultural centers of temperate regions, like the upper Mississippi River basin (USA) or the Arno River basin (Italy), in comparison with the still-developing tropical study region. The relatively frequent detection of some apolar pesticides (e.g., endosulfan, trifluralin), in comparison with the one of triazines and acetanilides, in tropical rivers points to an increased dispersion of these more volatile compounds in the studied tropical environment. Yet, their relative intensity of use, as well as their application mode (aerial application of cotton insecticides in the study region), may also play an important role in this respect.

In the Pantanal area only 47% of water samples contained detectable pesticide residues, mostly in low concentration (Table 3). Malathion, metolachlor, alachlor, and trifluralin were detected most frequently (>15% of sampling events) in water of the Pantanal border zone. Endosulfan-sulfate was detected at low concentrations (3–4 ng L-1); its parent compounds could not be detected any more above their limit of quantification. However, in comparison with data from sampling sites in the planalto, some pesticides occurred at similar or slightly higher concentrations in the Pantanal border zone (e.g., alachlor, metolachlor, trifluralin). Possibly, the main riverine flux of these pesticides, which are applied at the very beginning of the rainy season (October to November), already had passed on to the Pantanal area. Ametryn and endosulfan compounds were detected most frequently in the São Lourenço River throughout the study period with maximum concentrations of 18 and 82 ng L-1 (Sites R1 and R2). It seems therefore probable that also these substances would reach the Pantanal border zone to the south of our study area, warranting further investigations in this region.

Sediments
Sediment samples were only collected at Sites R1 to R6, because stream bank and bottom sediments at Sites S1 to S6 consisted of coarse sands, which do not adsorb pesticides in significant amounts (e.g., Moreau and Mouvet, 1997). It has to be noted that river sediments at Sites R1 to R3 were taken from lateral still-water zones, which were more turbulent during major storm flow events. Therefore, sediments at these locations could only accumulate during periods of low water level in the rivers. At the other sampling sites, bottom sediments of rivers and lakes were taken where a continuous sedimentation took place.

In comparison with water samples, only a small number of pesticides was encountered in sediments (Table 4). Concentrations were generally below 5 µg kg-1 sediment (dry weight), except for one detection of triazofos. This finding was in agreement with the lack of pesticide detections in suspended sediment of water samples. At Sites R1 to R3, organochlorine insecticides (p,p'-DDT, p,p'-DDE, endosulfan compounds) were detected most frequently, followed by the herbicide ametryn. In sediment samples of the Pantanal, only p,p'-DDT and p,p'-DDE were detected, again at low concentrations. Although the agricultural use of DDT was banned in Mato Grosso (Brazil) in 1985, this insecticide was still the most frequently detected pesticide in sediments of all compounds investigated. This demonstrated again the well-known persistence of this substance even in tropical environments (e.g., Castilho et al., 2000; Kidd et al., 2001) and justifies its prohibition from agricultural use in Brazil. Our results are corroborated by Matsushita et al. (1996) who measured similar background concentrations of p,p'-DDT and p,p'-DDE (0.2 to 0.6 µg kg-1) in sediments of rivers in Paraná state, Brazil.


View this table:
[in this window]
[in a new window]
  		Table 4. Detection frequency and pesticide and metabolite concentrations in sediment samples.

 
Endosulfan-sulfate and ß-endosulfan were also frequently detected in sediments of the São Lourenço River (Sites R1 and R2), indicating that this insecticide and/or its metabolite might accumulate in the sediment compartment of the studied river. Castilho et al. (2000) reported similar concentrations of endosulfan (1 to 9 µg kg-1) in bottom sediments of a Nicaraguan river system, where cotton production was intense. The absence of {alpha}-endosulfan in the analyzed samples, despite its dominance in the applied pesticide formulations (66–70% of the active ingredient mass in the formulation are comprised of {alpha}-endosulfan), pointed to a decreased persistence of this isomer in aquatic systems in comparison with ß-endosulfan. This finding corroborated results of Navarro et al. (2000) and Leonard et al. (2001) who measured a faster dissipation of the {alpha} than the ß isomer in aquatic systems.

Rainwater
Collectors for rainwater were passive pan samplers, which were shown to possess good reproducibility for the collection of pesticide bulk deposition (Cessna et al., 2000). As it is commonly assumed that the dry deposition of pesticides is of minor importance for the total pesticide deposition (e.g., Dubus et al., 2000), we regarded the measured pesticide deposition in first approximation as wet deposition. A deposition of pesticides with dust or eroded soil was thought to be negligible in our study, because no visible accumulation of particles occurred on the C18-cartridges during the solid-phase extraction of water samples. A description of sampling sites is given in Table 5. Besides the distinctive differences in land use between lowland and planalto, the higher cumulative precipitation at the highland sites (P1 and P2) also has to be mentioned. Pesticides were identified in 87% of all collected rainwater samples (n = 91), with up to nine pesticides detected in one sample.


View this table:
[in this window]
[in a new window]
  		Table 5. Characterization of rainwater sampling sites and cumulative deposition of pesticides and metabolites within the study period (10 Nov. 1999–2 Mar. 2000).

 
In planalto samples, endosulfan compounds were detected most frequently (Table 6), followed by the herbicides alachlor, metolachlor, and trifluralin (>30% of samples). The organophosphorus pesticides monocrotofos and profenofos were detected less frequently (26 and 29% of samples), but at the highest mean and maximum concentrations of all pesticides investigated (0.3 to 0.7 and 2.2 to 2.3 µg L-1, respectively). Within the study period (113 d), the highest deposition rates were measured for endosulfan ({alpha}- + ß-) and profenofos (>140 µg m-2), followed by monocrotofos (94 µg m-2) and the herbicides metolachlor and alachlor (>40 µg m-2). In Fig. 4 and 5 , the measured concentrations of selected pesticides at Sites P1 and P2 are presented. Pesticide deposition was initially dominated at both sites by selective corn and preemergence herbicides (alachlor, metolachlor, trifluralin), which are usually applied by October and November. In December, the application of endosulfan in cotton fields began, leading to continuous detection of this insecticide in rain (nearest application area: approximately 2 km to the south of Site P2, applied by plane). This period was accompanied by sporadic detections of monocrotofos, which is commonly applied to soybean. Profenofos, used in soybean and also cotton plantations, began to dominate the spectrum of deposited pesticides from the middle of January onward at both sampling sites. Yet, whereas at Site P1 alachlor was detected frequently and at high concentrations (up to 0.67 µg L-1) in rainwater, it was only detected at low concentrations (<0.07 µg L-1) in 28% of rainwater samples at Site P2. Vice versa, trifluralin was detected less frequently and at lower concentrations at Site P1 in comparison with Site P2. As in stream water and sediment samples, ß-endosulfan was detected more frequently and at higher concentrations than {alpha}-endosulfan in rainwater at both sites (Table 6). We attributed this finding in part to a higher persistence of the ß isomer in the atmosphere (Hoff et al., 1992). Yet, once volatilized, the substantially lower Henry constant of the ß as compared with the {alpha} isomer (Tomlin, 2000) would also favor the wash-out of ß-endosulfan by rainfall from the atmosphere. Although the distance between Sites P1 and P2 was more than 10 km, the pesticide deposition pattern was similar and followed the regional application sequence of pesticides in the main cash crop cultures of the planalto study area (Fig. 4 and 5). This finding suggests a large-scale dispersion of pesticide residues in the atmosphere of the highlands, resulting in a homogeneous pattern of pesticides deposited at Sites P1 and P2, in spite of sampling site–specific application history and precipitation differences. This assumption has to be verified by future studies with a grid-like sampling scheme in the planalto. The sum of pesticides deposited during the study period was about 400 µg m-2 at both sites (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 6. Detection frequency, concentrations, and deposition rates of pesticides and metabolites in rainwater (sampling period 10 Nov. 1999–2 Mar. 2000).

 


View larger version (49K):
[in this window]
[in a new window]
  		Fig. 4. Precipitation and concentrations of selected pesticides and metabolites in rainwater at Site P1.

 


View larger version (44K):
[in this window]
[in a new window]
  		Fig. 5. Precipitation and concentrations of selected pesticides and metabolites in rainwater at Site P2.

 
At least one pesticide was detected in 74% of rainwater samples (n = 46) of the lowlands. Yet, pesticide concentrations were generally below 0.1 µg L-1, with the exception of malathion, which was measured in concentrations up to 27.6 µg L-1 in rainwater of the Cuiabá city area (Table 6). Two unknown substance peaks occurred together with malathion in several rainwater samples at Site P3 and were identified by full-scan spectra as the insecticides carbaryl and fenitrothion (concentrations > 0.5 µg L-1). According to the city government, malathion had been vaporized partly in mixtures with carbaryl and fenitrothion during antimosquito campaigns from January to February 2000 to control dengue fever eruptions in the Cuiabá region. Due to the obvious origin of carbaryl and fenitrothion from local application in the public health sector, no further quantification was attempted for these substances. Besides malathion, ß-endosulfan and endosulfan-sulfate were the most frequently detected compounds (>50% of samples), followed by metolachlor, alachlor, trifluralin, and {alpha}-endosulfan (>8% of samples). Generally, pesticide mean concentrations and deposition rates were 5 to 10 times lower than in rainwater samples of the planalto. Exceptions were malathion, due to local use, and the infrequently detected triazofos and chlorpyrifos, which were detected in similar concentrations and frequencies as in the planalto. Individual deposition rates during the monitoring period were lower than 10 µg m-2 for all compounds, save for malathion. Endosulfan compounds were detected in samples since the end of December at the lowland sites, which coincides with their detection pattern at the planalto sites. All other pesticides (except malathion) were detected erratically throughout the study period.

Since all frequently detected pesticides (e.g., endosulfan, trifluralin, metolachlor, alachlor) were almost exclusively used in the planalto region, which is the nearest intensively cropped region to the sampling sites at Cuiabá and Barão de Melgaço, we deduced that the majority of pesticides detected at Sitess P3 to P5 were transported in the atmosphere for at least 75 km before deposition. The higher portion of endosulfan-sulfate found in the lowland samples as compared with the planalto rainwater samples (31.4 and 13.0%, respectively, of the mean deposited sum of endosulfan compounds) also pointed to a longer residence time of endosulfan in the atmosphere before deposition in the lowland samplers. The portion of deposited {alpha}-endosulfan in the lowlands (12.5% of the mean cumulative sum of deposited isomers) was even smaller than in the planalto (24.8%). Assuming a lesser persistence of the {alpha} isomer in the atmosphere (Hoff et al., 1992), this finding also relates to the longer transport time of endosulfan in the atmosphere until deposition at the lowland sites.

In comparison with studies on the contamination of rainwater in agricultural areas of temperate regions, the measured pesticide spectrum and concentrations compare well with data reported from Europe (Dubus et al., 2000; van Dijk and Guicherit, 1999). Yet, the most frequently detected pesticides in these studies were herbicides (alachlor, atrazine, metolachlor, simazine, trifluralin), whereas we detected in our study often insecticides (endosulfan, monocrotofos, profenofos). This was presumably due to the higher relative use of insecticides in the study area, especially in cotton plantations, as compared with the one in temperate regions. Racke et al. (1997) summarized that the use of insecticides is commonly more dominant in tropical than in temperate agriculture. In part, insecticides were applied to cotton fields by plane in the study area, which might also be responsible for the high detection frequency of some compounds (e.g., endosulfan) in our study.

The sum of pesticide deposition measured in our study in the planalto within 3.5 mo was similar to maximum annual deposition rates of pesticides (up to 400 µg m-2) found for one of the most developed agricultural regions of Greece (Charizopoulos and Papadopoulou-Mourkidou, 1999). Also, maximum concentrations and individual deposition rates of some pesticides (e.g., metolachlor, alachlor) were in a similar range as reported from Greece and the corn belt of the USA (Goolsby et al., 1997; Hatfield et al., 1996).

For tropical environments, no data on pesticide deposition with rainfall has been reported so far. Yet, the high individual maximum deposition rates of the insecticides endosulfan (sum of compounds: 168 µg m-2), profenofos (140 µg m-2), and monocrotofos (94 µg m-2) within the study period in the planalto demonstrate the need for a further monitoring of these highly toxic insecticides in the atmosphere of tropical environments. In particular, the impact of the pesticide deposition on stream–wetland ecosystems (e.g., gallery forests) in the highlands, which are important for the development and maintenance of high biodiversity in agricultural areas (Wantzen, 2000), deserves further attention. It remains to be investigated if the high deposition rates of insecticides were mainly caused by their intensive use in the planalto study area, or whether the volatilization of these compounds is generally enhanced in tropical regions as compared with temperate ones.

The deposition of typical cotton, soybean, and corn pesticides was also proven in rainwater samples of the lowlands, at least 75 km from the next application areas, although at concentrations about 5 to 10 times lower than in the planalto study area. This finding corroborated results of studies from temperate regions, where van Dijk and Guicherit (1999) also found evidence for the medium- to long-range transport of many pesticides. Pesticides deposited at the highest loads at the lowland sites (endosulfan compounds, trifluralin, metolachlor) are known to be transported thousands of kilometers in the atmosphere (van Dijk and Guicherit, 1999) and have even been detected in remote arctic regions like the Bering Sea (Chernyak et al., 1996). Consequently, the intensive use of pesticides (and especially insecticides) in the planalto might lead to the input of low, but nevertheless ecotoxicologically relevant quantities of pesticides by wet deposition to remote ecosystems (van Straalen and van Gestel, 1999), such as the Pantanal. To give a rough estimate of the atmospheric pesticide input to this wetland area (approximately 140 000 km2), we calculated, on the basis of our results (deposition of 20 µg m-2), a total of 2800 kg of pesticides deposited in the Pantanal within the study period of 3.5 mo as a worst case. In view of the strong increase of the pesticide-intensive cotton production in the state of Mato Grosso, which has been the major cotton producer in Brazil since 2000, a rising pesticide input to the Pantanal basin has to be expected.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Our pilot monitoring study showed evidence of the distribution of a broad spectrum of pesticides in the northeastern Pantanal basin. Of the 32 substances investigated (29 pesticides, 3 metabolites), 10 compounds were never detected in environmental samples: cyanazine, ß-cyfluthrin, {lambda}-cyhalothrin, deltamethrin, dimethoate, heptachlor, lindane, methyl-parathion, parathion, and quintozene. The other 22 substances were detected at least once, leading to an overall pesticide detection frequency of 68% in surface water samples (n = 139), 87% in rainwater samples (n = 91), and 62% in sediment samples (n = 26). Pesticide concentrations encountered in streams, rivers (<0.1 µg L-1), and sediments (<10 µg kg-1) were generally low when compared with studies from temperate regions. However, in planalto rainwater substantial insecticide (endosulfan, profenofos, monocrotofos) and herbicide (metolachlor, alachlor, trifluralin) concentrations were found, leading to a high pesticide deposition rate in this region. In the lowlands, at about 75 km distance from the next application areas, maximum pesticide concentrations in rainwater were about 5-fold, and deposition was about 10-fold lower than in the planalto. Pesticides of higher detection frequency and/or persistence in our study region were triazines (atrazine, simazine, ametryn), acetanilides (alachlor, metolachlor), trifluralin, and endosulfan compounds (especially ß-endosulfan and endosulfan-sulfate). Some organophosphorus pesticides (e.g., monocrotofos, profenofos, triazofos) showed higher peak concentrations in environmental samples but were rarely detected outside the application regions, presumably due to their low persistence. The absence of pyrethroids (e.g., deltamethrin, permethrin, cypermethrin, {lambda}-cyhalothrin) in environmental samples (with the exception of two cypermethrin–permethrin detections in rainwater) is attributed to their favorable physico–chemical properties (octanol–water partitioning coefficient > 40 000; vapor pressure < 2.5 x 10-3 mPa at 25°C) and their low application rates (<100 g ha-1). Formerly used organochlorine pesticides (lindane, p,p'-DDT, quintozene) could not be detected any more in water samples, and only p,p'-DDT and p,p'-DDE were found at low concentrations in sediments.

In comparison with studies from the midwestern USA (Goolsby et al., 1997), maximum concentrations in rainwater and deposition rates of pesticides were in a similar range or higher for individual compounds in the planalto area. However, measured river water concentrations of pesticides were substantially lower in our study area than in the midwestern USA (Pereira and Hostettler, 1993). This discrepancy gives rise to the speculation that in the studied tropical area the atmospheric input of pesticides to ecosystems might be more relevant than in temperate regions. We may propose that this is either caused by higher volatilization rates of pesticides after application or by an accelerated pesticide degradation in terrestrial and aquatic environments of the tropics, or by both. Clearly, pesticide dynamics in tropical ecosystem compartments are not yet sufficiently investigated to prove general differences in pesticide fate between tropical and temperate regions.


   ACKNOWLEDGMENTS
 
This research was part of the SHIFT program (Studies on Human Impact on Forests and Floodplains in the Tropics), supported by the German Bundesministerium für Bildung und Forschung (BMBF Project 01LT0003/7), the Brazilian CNPq and the Brazilian IBAMA. The authors thank Carlos A. Schneider and Bernd Reizner for assistance during the sampling period.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
M. T. Wan, J.-n. Kuo, and J. Pasternak
Residues of Endosulfan and Other Selected Organochlorine Pesticides in Farm Areas of the Lower Fraser Valley, British Columbia, Canada
J. Environ. Qual., June 7, 2005; 34(4): 1186 - 1193.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
H. Bernard, P. F. Chabalier, J. L. Chopart, B. Legube, and M. Vauclin
Assessment of Herbicide Leaching Risk in Two Tropical Soils of Reunion Island (France)
J. Environ. Qual., March 1, 2005; 34(2): 534 - 543.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (22)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Laabs, V.
Right arrow Articles by Zech, W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Laabs, V.
Right arrow Articles by Zech, W.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Laabs, V.
Right arrow Articles by Zech, W.
Related Collections
Right arrow Organic Compounds
Right arrow Agricultural Pesticides
Right arrow Air Pollution
Right arrow Water Pollution