Search Help
Distribution of Dissolved Pesticides and Other Water Quality Constituents in Small Streams, and their Relation to Land Use, Willamette River Basin, Oregon
Entry ID: WRIR_97_4268

[ Update this Record ]

Abstract: Water quality samples were collected at sites in 16 randomly selected
agricultural and 4 urban subbasins as part of Phase III of the Willamette River
Basin Water Quality Study in Oregon during 1996. Ninety-five samples were
collected and analyzed for suspended sediment, conventional constituents
(temperature, dissolved oxygen, pH, specific conductance, nutrients,
biochemical oxygen demand, and bacteria) and a suite of 86 dissolved
pesticides. The data were collected to characterize the distribution of
dissolved pesticide concentrations in small streams (drainage areas 2.6-13
square miles) throughout the basin, to document exceedances of water quality
standards and guidelines, and to identify the relative importance of several
upstream land use categories (urban, agricultural, percent agricultural land,
percent of land in grass seed crops, crop diversity) and seasonality in
affecting these distributions. A total of 36 pesticides (29 herbicides and 7
insecticides) was detected basinwide. The five most frequently detected
compounds were the herbicides atrazine (99% of samples), desethylatrazine
(93%), simazine (85%), metolachlor (85%), and diuron (73%). Fifteen compounds
were detected in 12-35% of samples, and 16 compounds were detected in 1-9% of
samples. Water quality standards or criteria were exceeded more frequently for
conventional constituents than for pesticides. State of Oregon water quality
standards were exceeded at all but one site for the indicator bacteria E. coli,
3 sites for nitrate, 10 sites for water temperature, 4 sites for dissolved
oxygen, and 1 site for pH. Pesticide concentrations, which were usually less
than 1 part per billion, exceeded State of Oregon or U.S. Environmental
Protection Agency aquatic life toxicity criteria only for chlorpyrifos, in
three samples from one site; such criteria have been established for only two
other detected pesticides. However, a large number of unusually high
concentrations (1-90 parts per billion) were detected, indicating that
pesticides in the runoff sampled in these small streams were more highly
concentrated than in the larger streams sampled in previous studies. These
pulses could have had short term toxicological impli cations for the affected
streams; however, additional toxicological assessment of the detected
pesticides was limited because of a lack of available information on the
response of aquatic life to the observed pesticide concentrations. Six
pesticides, including atrazine, diuron, and metolachlor, had significantly
higher (p<0.08 for metolachlor, p<0.05 for the other five) median
concentrations at agricutural sites than at urban sites. Five other
compounds--carbaryl, diazinon, dichlobenil, prometon, and tebuthiuron--had
significantly higher (p<0.05) concentrations at the urban sites than at the
agricultural sites. Atrazine, metolachlor, and diuron also had significantly
higher median concentrations at southern agricultural sites (dominated by grass
seed crops) than northern agricultural sites. Other compounds that had higher
median concentrations in the south included 2,4-D and metribuzin, which are
both used on grass seed crops, and triclopyr, bromacil, and pronamide. A
cluster analysis of the data grouped sites according to their pesticide
detections in a manner that was almost identical to a grouping made solely on
the basis of their upstream land use patterns (urban, agricultural, crop
diversity, percentage of basin in agricultural production). In this way
inferences about pesticide associations with different land uses could be
drawn, illustrating the strength of these broad land use categories in
determining the types of pesticides that can be expected to occur. Among the
associations observed were pesticides that occurred at a group of agricultural
sites, but which have primarily noncropland uses such as vegetation control
along rights-of-way. Also, the amount of forested land in a basin was
negatively associated with pesticide occurrence, suggesting that riparian
growth or runoff from forested lands helped reduce pesticide concentrations.
Estimates of pesticide application also were made for the 16 agricultural study
basins. Concentrations of pesticides in streams were significantly (albeit
weakly) correlated (p<0.05) with estimated use for only a few compounds that
are applied to a wide variety of crop types. Because of the large acreages
involved, several compounds that are applied to grass seed were better cor
related with the fraction of upstream land use in agricultural production or in
grass seed crops than with their respective estimated applications. Application
estimates for some compounds, including atrazine and metolachlor, were probably
low because of uses that are not indicated in current literature. Significant
correlations were also found among certain individual compound concentrations,
and between these and concentrations of suspended sediment. Included in both
groups were atrazine and metolachlor, suggesting that environmental factors
that mobilize atrazine and metolachlor can mobilize other compounds, and that
hydrologic conditions are as important as the specific amount and timing of
application in determining the transport of many compounds to the streams. The
suspended sediment concentration was not, however, significantly correlated
with discharge, and concentrations of only one pesticide were correlated with
discharge. Even though correlations between discharge and pesticide
concentration were poor, the similar seasonal pattern in both variables is
evidence that transport to the streams is related to discharge and consequently
to the amount of runoff. Median concentrations of atrazine, metolachlor,
diuron, metribuzin, pronamide, and suspended sediment were significantly higher
in the late fall than in the summer. Additionally, winter "baseline" sampling
for both atrazine and metolachlor confirmed that median concentrations as high
as those in the fall or spring were maintained well past any periods of initial
flushing, suggesting that a steady supply of atrazine and metolachlor is
retained in soils in the study basins. Two intensive immunoassay studies
illustrated variations in pesticide concentration over storm hydrographs.
During a large storm with localized flooding, atrazine concentration increased
on the rising limb of the hydrograph, started to decrease just prior to peak
stage (indicating dilution), and continued to decrease as the water level
decreased. Metolachlor concentrations decreased throughout the storm by a
factor of two from their concentrations prior to the storm. The future
prospects for successfully correlating the stream loads of certain pesticides
with estimates of application rates may be good if current and locally specific
rates of application to various crop types can be obtained. Alternatively,
atrazine concentration appears to be at least a rough indicator for conditions
that move several other compounds, and it was shown that it can be measured
relatively cheaply and with good accuracy and precision, with enzyme
immunoassays. However, the prevalence of atrazine in stream water throughout
the basin precludes its use for prediction of occurrence or concentrations of
specific compounds in the absence of other information.The primary purpose of
the study was to describe the distribution of dissolved pesticide
concentrations in selected small streams throughout the basin, (2) document
exceedances of water quality guidelines for the targeted pesticides, and (3)
identify the relative importance of broad measures of land use and seasonality
in determining those concentrations. Secondary objectives are to (4) describe
relations, where they exist, between selected pesticide applications and stream
concentrations or loads and, (5) for those relations identified, to further
describe their dependence on seasonality and on selected site and compound
characteristics. A final objective is to further characterize water quality at
the chosen sampling sites with respect to conventional constituents.

Discharge was measured according to standard USGS guidelines as described by
Rantz and others (1982). No sites were gauged. Discharge was measured twice
during each of spring and fall and once during summer, when a complete set of
samples were collected for analytical chemistry. However, many sites were
visited at other times for a rapid collection of samples for immunoassay
analysis of triazine (primarily atrazine) or chloroacetamide (primarily
metolachlor) herbicides. In order to estimate the relative stage of streams
when time constraints prohibited a full discharge measurement, reference points
were established at each site from which to consistently measure either the
depth of the water or the distance to the water surface. These reference point
depths were noted at the time of each discharge measurement and also at any
time that immunoassay samples were taken.

Samples were collected at each site twice during spring and fall in order to
assess stream responses to runoff, and once during summer to assess low-flow
conditions. Basinwide samplings during spring and fall were timed to
correspond to periods of rainfall runoff, with minimum intervals of
approximately 1 week of dry weather required between samplings in each season,
to allow pesticide applications to occur and stream discharges to return to
steady flows. Rain storms during spring 1996 that produced runoff were well
spaced, and samplings for successive storms were conducted in mid-April and
mid-May. During fall, basinwide samplings were conducted in mid-October and in
mid-November. Constituents collected during basinwide samplings in spring,
summer, and fall included pesticides, conventional constituents, and suspended
sediment; additional samplings were conducted at individual sites, and
basinwide during the winter, using immunoassays as a screening tool to expand
the number of samples for atrazine and metolachlor.

Water samples for pesticides and conventional constituents were collected as
grab samples from midstream. Samples for suspended sediment were collected
using the equal-width-increment method, a depth- and width-integrating
technique described by Edwards and Glysson (1988). Water temperature, DO, pH,
and specific conductance were measured in place using Hydrolab multiparameter
probes that were calibrated in the field according to the manufacturer's
suggested methods. All samples were processed prior to shipment to
laboratories for analysis.

Grab samples were generally collected at the centroid of flow by wading. When
safety considerations prevented wading, samples were collected using weighted
bottle holders suspended from a bridge or culvert above the stream. Sampling
personnel wore plastic gloves to minimize contamination. At each site,
pesticide samples were collected into cleaned and baked (350 degrees Celsius,
12 hours) amber glass (GCC) bottles, nutrients and BOD5 were collected in
polypropylene bottles, and bacteria samples were collected in autoclaved
polycarbonate bottles. Additional samples for immunoassays were collected in
GCC bottles as needed. GCC and bacterial bottles were not rinsed in the field,
whereas bottles for nutrients and BOD5 were rinsed three times with stream
water prior to filling. All sample bottles except those for suspended sediment
were stored on ice until they were returned to the Oregon District Laboratory
for processing, usually a period of 1 to 6 hours.

At the Oregon District Laboratory, pesticide samples were immediately filtered
into clean GCC bottles through 0.7 micrometer pore-sized baked glass-fiber
filters and subsequently chilled. Small aliquots of the filtrate were
subsampled for analysis of herbicides using immunoassay methods. The remaining
filtrate was extracted onto a solid-phase sorbent material (Sandstrom, 1989),
which was then shipped within 4 days of collection to the USGS National Water
Quality Laboratory (NWQL) in Arvada, Colorado, for elution and subsequent
analysis. Pesticide analysis was performed using gas chromatography/mass
spectroscopy (GC/MS-USGS schedule 2010) or high-pressure liquid chromatography
(HPLC-USGS schedule 2051). Procedures for filtration, solid-phase extraction,
elution, and analysis of pesticides by GC/MS have been detailed by Zaugg and
others (1955), and similar procedures describing sample preparation and
analysis by HPLC are described by Werner and others (1996). The suite of 86
pesticides analyzed by the two methods is listed in table 2 of the Anderson,
Wood and Morace, distribution of dissolved pesticides and other water quality
constituents in small streams, and their relation to land use, in the
Willamette River Basin, Oregon, 1996 publication and other constituents and
their methods are listed in table 3 of the same publication. Units of
concentration used are in terms of micrograms per liter equivalent to parts per
billion, or ppb) for pesticides, and milligrams per liter equivalent to parts
per million, or ppm for nutrients, BOD5 and suspended sediment.

Water samples were also collected for the measurement of atrazine and
metolachlor concentrations by enzyme-linked immunoabsorbent assays. This
method, often referred to as "immunoassay", uses antibodies selective for the
compound being analyzed for, making it possible to isolate the target compound
and determine the concentrations at low levels (less than 1 part per billion).
Immunoassay samples were collected for two purposes: (1) to assess the
agreement between the immunoassay and GC/MS methods and (2) to provide better
resolution of temporal variability, especially during storms and midwinter
"baseline" conditions. The advantages of the immunoassay method over the more
comprehensive GC/MS analysis include the lower cost and the timeliness of the
data. However, the immunoassay method is less compound specific than GC/MS.

Immunoassays for atrazine and metolachlor were chosen because these compounds
were expected to be commonly detected on the basis of previous studies
(Anderson and others, 1996) and because reliable test kits were available to
them. Immunoassay kits were used according to the manufacturer's
specifications (Ohmicron Environmental Diagnostics, Inc., written commun.,
March 1996). Analyses were performed in triplicate with the RPA-I RaPID
Photometric Analyzer TM (Ohmicron Environmental Diagnostics, Inc., 1992).

The lower limits of the NWQL's analytical capabilities are generally reported
by one of two methods. The minimum reporting level (MRL) is the lowest
measured concentration of a constituent that may be reliably reported using a
given analytical method (Timme, 1994). For methods such as nutrient analysis
that use MRLs' concentrations less than the MRL are consored, and the data are
reported as being less than the value of the MRL. The method detection limit
(MDL) is a statistically derived minimum concentration that can be identified,
measured, and reported with a 99% confidence as being greater than zero
(Sandstrom, 1989). That is, there is no more than a 1% chance that a
concentration greater than the MDL was reported for a sample that actually did
not contain the analyte (false positive). Concentrations may be reported that
are less than the MDL, but the chance of a false positive detection is greater
than 1%. In contrast, the actual concentration in a sample reported as a
nondetection has up to a 50% chance of being equal to or greater than the MDL
(false negative). Concentrations for compounds listed in table 2 (Anderson and
others, 1997) are reported using MDLs. Pesticide analysis of several stream
samples were qualified by NWQL analysts as particularly difficult due to
interferences from nontarget compounds, sometimes at relatively high
concentrations. As a special analysis, extracts from three of these "dirty"
samples were re-analyzed by custom, high-resolution electron-capture
negative-ion mass spectrometry in order to investigate the causes of the
interferences. Identification of additional compounds observed in these
samples (using a GC/MS spectral library) was provided where possible, and
inferences about their sources were made with the assistance of NWQL analysts.

Geographic Coverage
 N: 46.0 S: 43.5  E: -121.5  W: -124.0

Data Set Citation
Dataset Originator/Creator: Chauncey W. Anderson, Tamara M. Wood and Jennifer L. Morace
Dataset Title: Distribution of Dissolved Pesticides and Other Water Quality Constituents in Small Streams, and their Relation to Land Use, In the Willamette River Basin,Oregon
Dataset Release Date: 1997
Dataset Release Place: Portland, Oregon
Dataset Publisher: U.S. Geological Survey
Data Presentation Form: database

Temporal Coverage
Start Date: 1996-03-01
Stop Date: 1996-11-30

Location Keywords

Science Keywords

In order to estimate variability in sampling and laboratory techniques,
quality control (QC) samples were submitted to the NWQL for pesticides, and to
the ODEQ and ACWA laboratories for the conventional constituents. Most QC
samples were used to evaluate the potential for problems form the combination
of field and laboratory procedures. QC samples for both pesticides and
conventional constituents included (1) field and equipment blanks to test for
contamination, (2) replicate native-water grab samples to test for precision,
(3) depth and width integrated samples collected as replicates to compare with
grab samples, and (4) distinct compounds, representing relevant pesticide
families (surrogates), added in known amounds to each pesticide sample to
monitor the analytical method's ability to quantify those sample types.
Additional QC samples for pesticides included (5) native-water samples spiked
with pesticide mixtures to test for accuracy, done at a range of concentrations
(low, medium, and high)), and (6) replicate spike samples to test for accuracy
and precision. Water for blank samples was carefully selected to be free fo
the constituents of concern: organic-free water was used for pesticide,
immunoassay, and BOD5 samples; inorganic-free water was used for nutrient
samples; a peptone buffer solution was used for fecal coliform bacteria, and a
sterile saline solution was used for E. Coli bacteria blanks. QC data for this
study are presented in Appendix 1 of Anderson and others, 1997.

The study design called for sample collection from 16 randomly selected
subbasins that each had predominantly agricultural land uses upstream of the
sampling site, and 4 subbasins having predominantly urban land use. In order
to minimize inputs of water from undefined or highly varied sources, small
drainage basins ranging from approximately 3 to 15 square miles were selected.
The four urban sites were selected from a set of urban subbasins drawn on
topographic maps, and the final choice was based primarily on the desire to
sample urban drainages that had not been extensively sampled previously, the
desire to sample sites in urban areas in the northern, central and southern
Willamette Basin, and the suitability of a site for sampling. Land use
information for the urban sites was derived fromthe GIS, with coverages for
urban lands most recently updated on the basis of the 1990 census (Hitt, 1994).
Due to rapid growth in many of western Oregon's cities during the 1990's data
for urban land use taken from the GIS is expected to somewhat underestimate the
proportion of urban lands and over estimate the proportion of agricultural or
forested lands in the Phase III urban subbasins. However these data were
considered adequate for the purposes of the study and the report.

Access Constraints

Use Constraints

water quality
suspended sediment
conventional constituents
dissolved oxygen
specific conductance
biochemical oxygen demand
Willamette River Basin
Biological Data Profile

Data Set Progress

Data Center
Water Resource Division, Oregon, U.S. Geological Survey, U.S. Department of the Interior    [Information]
Data Center URL:

Data Center Personnel
Phone: (503) 251-3200
Fax: (503) 251-3470
Email: ddlynch at
Contact Address:
U.S. Geological Survey
10615 S.E. Cherry Blossom Drive
City: Portland
Province or State: Oregon
Postal Code: 97216-3159
Country: USA

Phone: (503) 251-3206
Fax: (503) 251-3470
Email: chauncey at
Contact Address:
U.S. Geological Survey
10615 SE Cherry Blossom Dr.
City: Portland
Province or State: Oregon
Postal Code: 97216
Country: USA

Phone: (301) 614-6898
Fax: 301-614-5268
Email: Tyler.B.Stevens at
Contact Address:
NASA Goddard Space Flight Center
Global Change Master Directory
City: Greenbelt
Province or State: MD
Postal Code: 20771
Country: USA

Anderson, C. W., Wood, T.M., and Morace, J.M., 1997, Distribution of Dissolved
Pesticides and Other Water Quality Constituents in Small Streams, and their
Relation to Land Use, in the Willamette River Basin, Oregon, 1996. U.S.
Geological Survey Water-Resources Investigations Report 97-4268, Portland,
Oregon. U.S. Geological Survey

Creation and Review Dates
DIF Creation Date: 2001-06-27
Last DIF Revision Date: 2016-11-18
Future DIF Review Date: 2002-06-27

[ Update this Record ]

Link to Web Site