Scientific Investigations Report 2006–5101–D
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2006–5101–D
Stevens-Greenspan Model PS310 pressure transducers, each with an internal data logger, were used to measure stream-stage fluctuation during the study (Greenspan Technology, 2006). Transducers were placed instream in pools or runs to ensure consistent response of hydrologic stage and to minimize the potential for dewatering. The transducer model was not vented to the atmosphere; therefore, changes in recorded pressure reflected changes in stream level and atmospheric pressure. Data were corrected for fluctuations in atmospheric pressure using hourly barometric pressure data from nearby airports because continuous barometric pressure records were not available at the study sites. The stage data from the transducers had a precision of ±0.036 m, which did not meet USGS requirements for stage data precision (±0.003 m) (Sauer, 2002); however, it was deemed acceptable for the purposes of this study.
For logistical reasons, the pressure-transducer deployment periods varied among the sampling sites, with the most complete record obtained between March and November 2004. This 9-month period included a number of “typical” spring and autumn rain events, and provided adequate data to characterize the hydrologic variability among sites. Hydrologic variables calculated from the stage data included more than 35 hydrologic variables. These variables included measures of stage variability (regularity in streamflow), estimates of streamflow magnitude (amount of water moving past a given point per unit of time), stream flashiness (how quickly streamflow changes from one magnitude to another), duration (length of time associated with specific streamflow conditions), and frequency (how often streamflows greater than or less than a certain magnitude recur). Calculations were based on equations outlined in McMahon and others (2003) using SAS version 8 (Delwiche and Slaughter, 1998).
Stevens-Greenspan Model PS310 pressure transducers also monitored continuous water temperature data (30-minute intervals) during the study. Twenty percent of the transducers were tested for accuracy (within ± 0.01°C, verified by comparing readings in a temperature bath with a traceable National Institute of Standards and Technology [NIS] thermometer prior to field deployment). Temperature data were stored in the Automated Data Processing System (ADAPS), a part of the National Water Information System (NWIS) (U.S. Geological Survey, 2003). Summary statistics for various water temperature measures included daily minimum, maximum, mean, range, and standard deviation, which were calculated for each stream using hourly data.
Occasionally, short periods of temperature record were lost due to transducer failure or dewatering, such as when stream levels dropped during low flow in late summer. Temperature data for the missing intervals were reconstructed using an extraction-correlation technique, which used the 30-minute data to extract daily mean values from March through November 2004 data. Sites then were correlated with each other. Linear regressions based on these correlations were used to estimate temperature values for days without a daily mean value. At sites with missing data, an average of the regressions was used to estimate missing values.
Watershed-level characterization of habitat provided information on the upstream geologic, climatic, hydrologic, morphologic, and biologic influences at a site. Watershed-level habitat variables defined in this study included drainage area, drainage density, watershed length, mean watershed elevation, drainage shape, watershed relief, drainage texture, and cumulative perennial stream length. Other watershed-level information included land cover, surficial geology, soil, and riparian variables. Segment-level characterization of habitat provided information on finer scale influences in the relatively homogenous segment stream length. Actual segment length varies among streams depending on the distance between significant tributaries and/or point source inputs (Fitzpatrick and others, 1998). Segment-level variables determined in this study included sinuosity, slope segment length, and channel gradient. Watershed-level and segment-level characteristics were derived by GIS.
Reach-level characterization based on site visits was the principal means for describing local-scale influence within a segment (Fitzpatrick and others, 1998). Reach length was determined by multiplying the mean wetted channel width by 20 to ensure that all habitat types (pools, riffles, and runs), were represented within the reach. Reach-scale habitat data were collected during low-flow conditions in July and August 2004. Stream depth, width, bed substrate, habitat cover, bank morphology, canopy closure, stream velocity, and bank vegetation were measured at 11 or 12 equally spaced transects along each reach (mesoscale characterization). At one site—Curtin Creek—only nine transects were completed due to channel reach constraints. In addition, point velocity, substrate, and depth were measured where richest targeted habitat algae and benthic macroinvertebrate samples were collected (micro-scale characterization). A complete list of habitat variables used in this study is given in the appendix (tables A3 and A4). Detailed information on methods of habitat data collection and variables is available in Fitzpatrick and others (1998).
GIS variables, additional to those originally used in site selection and UII genesis, were gathered for analytical purposes. Hydrologic variables describing stream segment, riparian buffer, and road/stream intersection were examined, as well as associated dams, reservoirs, and waterway diversions. In addition, the program FRAGSTATS (McGarigal and others, 2002) was run for each final watershed to evaluate spatial land use patterns. FRAGSTATS variables quantified the degree of fragmentation, such as size, configuration, and connectivity, of urban and nonurban areas in a watershed (Sprague and others, 2006). As Alberti and Marzluff (2004) noted, this disruption of continuous land can affect ecosystem health by limiting or interrupting the natural movement of organisms. All additional GIS variables are available in Sprague and others (2006).
Water samples for chemical analysis were collected from all 28 sites twice during the study. Samples were collected from all sites in May 2004 (spring sampling) and in late August or early September 2004 (summer sampling) to bracket the biological sampling during July through September 2004. Water-chemistry conditions during these months were more likely to have a more direct effect on the biological communities in the streams than conditions earlier in the study. To document the seasonal variability in water chemistry, 10 of the 28 sites were sampled 4 additional times: November 2003, and January, March, and June 2004. These 10 “high frequency” sampling sites spanned the full range of the UII to determine whether the degree of urbanization affected the seasonality of water chemistry (table 1).
Sulfate, chloride, nutrients, pesticides, dissolved and particulate organic and inorganic carbon, and suspended sediment samples were collected at all sites (table A7). Field measurements of water temperature, dissolved oxygen (DO), pH, specific conductance, and streamflow also were made during sampling. Samples were collected using standard protocols as outlined in the USGS National Field Manual (U.S. Geological Survey, variously dated). Nutrient and pesticide samples were analyzed at the USGS National Water-Quality Laboratory (NWQL) in Lakewood, Colorado, using methods developed by Fishman (1993) and Zaugg and others (1995). Suspended-sediment samples were analyzed at the USGS Cascade Volcano Observatory (CVO) sediment laboratory in Vancouver, Washington. Quality-control samples, including field blanks, replicates, and laboratory spikes were collected throughout the study and analyzed at the NWQL and CVO. About 10 percent of the total number of field samples was collected for quality assurance. All quality-control, or quality-assurance, samples analyzed indicated that sample collection, processing, or laboratory analysis were acceptable.
Semipermeable membrane devices (SPMDs) are passive sampling cartridges that were deployed in each stream to sample and concentrate hydrophobic organic contaminants from the water (Huckins and others, 1993; Bryant and others, 2007). In this study, SPMDs were designed to mimic the fatty tissues of fish, and used to indicate the potential for bioaccumulation of polychlorinated dioxins and furans, polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), and organochlorine and pyrethroid insecticides.
SPMDs were deployed in each of the 28 streams for about 4 weeks beginning in July 2004. At the end of the deployment period, they were removed and sent to multiple locations for analysis. Contaminant residues were recovered and separated at Environmental Sampling Technologies in St. Joseph, Missouri, as described in Huckins and others (1990). An ultraviolet fluorescence scan to quantify total PAHs (Johnson and others, 2004) and a Microtox® bioassay (Johnson, 1998) was run at the USGS Columbia Environmental Research Center in Columbia, Missouri. U.S. Army Corp of Engineers environmental laboratory in Vicksburg, Mississippi, ran an additional assay, the P450 RGS test that screens for aryl hydrocarbon receptor (AhR) type compounds that include PCBs, PAHs, dioxins, and furans (Murk and others, 1996).
Because SPMDs integrate chemical conditions over time and, sometimes, during variable flow conditions, they provide a more complete representation of chemical exposure than periodically collecting water samples (Huckins and others, 1993). SPMDs also eliminated the problem of determining chemical exposure in aquatic organism tissue by eliminating concern whether organisms metabolized compounds or if organisms migrated from exposure sources. Limitations of the SPMDs include:
All final data used in analysis were blank corrected/time normalized according to procedures outlined by Bryant and others (2007) to address these limitations and allow for better comparability of data among sites.
Algal, benthic macroinvertebrate and fish assemblage samples were collected once during the study period in each of the 28 streams following protocols described in Moulton and others (2002). Algal and benthic macroinvertebrate samples were collected during September and October 2004, respectively, and fish communities were sampled in July and August 2004.
Quantitative algal samples were collected at each site from riffles at the richest targeted habitat (RTH) using methods described in Moulton and others (2002). RTH algal samples were collected from 5 to 15 representative rocks per site and combined into a single sample. Rocks were removed, and algal material was collected using the pipe-scribe top rock scrape method described in Carpenter (2003). A round plastic scribe (short length of PVC pipe) with an outside diameter ranging between 4 and 10.4 cm was placed on each rock, and algal material outside the scribe was removed with a plastic-bristle brush or scraped off with a knife, and discarded. The circular patch of algae remaining on the rock was scraped into a small washbasin, and rinsed into a 1-L sample bottle using stream water. Samples were briefly homogenized with an electric blender, and subsamples were collected for chlorophyll-a, ash-free dry mass (AFDM), and species identification. Chlorophyll-a and AFDM subsamples were collected on 45-micron glass-fiber filters, wrapped in foil, sealed, packed on dry ice, and shipped to the NWQL for fluorometric chlorophyll-a (Arar and Collins, 1997) and gravimetric AFDM analyses (Britton and Greeson, 1987). All algae water samples were preserved in 5 percent buffered formalin solution and shipped to the Academy of Natural Sciences of Philadelphia for taxa identification and enumeration following protocols described by Charles and others (2002).
One semiquantitative RTH sample for benthic macroinvertebrates was collected from five riffle areas in each stream. Each of the five subsamples were collected using a 500-micron mesh Slack sampling net (modified Surber design), which was placed in the stream, and rocks were cleaned of benthic organisms from a 0.25 m2 sampling area into the net. The five subsamples were combined in the field at each site. Additionally, one qualitative multihabitat (QMH) sample was collected using a 500-micron mesh dip net, which was used to collect and composite organisms from a diversity of microhabitats present at each site (for example, riffles, runs, pools, grasses, woody debris) into a 19-L bucket. Microhabitats were sampled equally for a maximum of 1 hour. Both individual sample types underwent a field elutriation process to clean and remove large organic debris, excess rocks, and sand. The composited macroinvertebrate sample was transferred into a 1-L plastic bottle, preserved with 10 percent buffered formalin, and shipped to the NWQL for taxa identification and enumeration (Moulton and others, 2002).
Fish were collected using a Smith-Root Model BP2 backpack electro-shocker, with two separate upstream passes from the start to the end of the reach. Fish were caught using 6 mm mesh nets and stored in aerated live wells. All fish were identified and enumerated in the field after each sampling pass. The first 30 fish of each species were weighed to the nearest 0.1 g, measured to the nearest millimeter, and checked for external anomalies (Moulton and others, 2002). The remaining individuals of each species were enumerated and checked for anomalies. Representative specimens of difficult to identify species were labeled and preserved in 10 percent buffered formalin solution and sent to the Department of Fish and Wildlife Ichthyology Museum at Oregon State University, Corvallis, Oregon, for identification verification.
U.S. Department of the Interior | U.S. Geological Survey
URL: https://pubs.usgs.gov/sir/2006/5101-D
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:24:16 EST
