Data collection during Phase II of the MMSD Corridor Study was a collaborative effort between USGS and MMSD field personnel. A total of 15 stream sites and 6 harbor sites were sampled (fig. 1, tables 1A–B) during water years 2004 and 2005². All stream sites were within the MMSD planning area. The site at the mouth of the Milwaukee River (Milwaukee River at Mouth at Milwaukee, 04087170) receives water from the Milwaukee, Menomonee, and Kinnickinnic River watersheds in addition to water from Lake Michigan (by way of reverse flow). Oak Creek and Root River both discharge directly into Lake Michigan south of the Milwaukee Harbor. Jewel Creek discharges into the Mississippi River by way of the Fox and Illinois Rivers. Of the six harbor sites sampled, three sites were inside the breakwall (inner harbor sites) and three sites were outside the breakwall (outer harbor sites). Samples were collected manually at all sites, and additional samples were collected automatically at four of the stream sites.

² Water year is a 12-month period from October 1 through September 30, and is designated by the calendar year in which it ends (for example, the 2004 water year occurred October 1, 2003 through September 30, 2004).

Table 1A. Basin characteristics and streamflow statistics for stream sites sampled during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study.

[mi², square mile; ft³/s, cubic foot per second; Q, discharge; WQ, water quality; --, not available; site locations shown in fig. 1]

   ^a Automated sample collection at this site.
   ^b Integrator site. This site receives water from the Milwaukee, Menomonee, and Kinnickinnic River watersheds in addition to water from Lake Michigan (by way of reverse flow).
   ^c (Waschbusch and others, 2006)
   ^d Instantaneous discharges were compared to the 10-percent-exceedence discharges at this site to identify high-flow event conditions for the MMSD Planning Area.

Table 1B.  Harbor sites sampled during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study.

[site locations shown in fig. 1]

return to top

Land Use and Phase II Sites

Land use for Phase II stream-site drainage basins was determined by means of a Geographic Information System (GIS). Drainage-basin boundaries were primarily derived from the Southeastern Wisconsin Regional Planning Commission (SEWRPC) Subbasin layer (2002a). For sites not on SEWRPC Subbasin boundaries (Southeastern Wisconsin Regional Planning Commission, 2002a), USGS 1:24,000 topographic maps were used to delineate the most downstream boundary. Resulting polygons were intersected with land-use layers. Land-use information was derived primarily from the SEWRPC digital land-use layer (2000). This layer did not incorporate the most upstream parts of the Milwaukee River drainage basin; therefore, upstream parts of this drainage basin were obtained from the WISCLAND land-cover layer (Wisconsin Department of Natural Resources, 1998); resultant data were combined with data obtained from the SEWRPC digital land-use layer to yield land-use information for these sites (Milwaukee River near Cedarburg, Milwaukee River at Milwaukee, and Milwaukee River at Mouth). Data for each drainage basin were summarized into basic categories (appendix 1).

The 15 stream sites were selected to fill spatial, temporal, and technological data gaps identified in Phase I analysis. Sites spanned a wide range of drainage areas and land uses (table 1A, fig. 2). Drainage areas ranged from 6.33 to 872 mi², while land use ranged from predominantly urban to predominantly agricultural and natural areas (fig. 3). Land-use types were highly correlated with one another; therefore it was not possible to observe the direct effects of individual land-use types (especially the types of urban land use). Urban land-use types were grouped together into a single total urban land-use category which was made up of transportation, residential, commercial, industrial, and other urban land uses. Because land-use types were highly correlated with one another, including natural area and agricultural land uses, all comparisons were made to total urban land use. Two Superfund sites in the Milwaukee area are discussed in this report in conjunction with Phase II site results:

1. Moss-American Co., Inc. (Kerr-McGee Oil Co.)(U.S. Environmental Protection Agency ID: WID039052626), Milwaukee. The types of contaminants at this site are polycyclic aromatic hydrocarbons (PAHs) and volatile organic compounds (VOCs) in bed sediment. (U.S. Environmental Protection Agency, 2006c). This site is upstream of the Little Menomonee River site sampled during Phase II.
2. Cedar Creek (U.S. Environmental Protection Agency ID: WID988590261) at Cedarburg. This is a Superfund alternative site; primary contaminants at this site are PCBs in bed sediment (U.S. Environmental Protection Agency, 2006b). Cedar Creek is a tributary to the Milwaukee River and is upstream from the Phase II Milwaukee River sites.

Figure 2.  Land use/land cover and stream-site drainage-area boundaries for stream sites sampled during Phase II of the Milwaukee Metropolitan Sewerage District (MMSD) Corridor Study. Site abbreviations listed in table 1.

Figure 3.  Land use for Phase II sites of the Milwaukee Metropolitan Sewerage District Corridor Study. Urban land use consists of residential, transportation, industrial, commercial, and other urban land use. Site abbreviations listed in table 1.

There are three wastewater treatment plants (WWTPs) in the planning area: the Jones Island Wastewater Treatment Plant, the South Shore Wastewater Treatment Plant, and the City of South Milwaukee Wastewater Treatment Facility. None of these WWTPs discharge into streams sampled in Phase II.

return to top

Collection of Stage and Discharge Data

Stage and discharge data were collected continuously (every 5, 15, or 60 minutes, depending on gage) at 10 of the 15 stream sites at the beginning of Phase II sampling through the USGS stream-gage network (table 1A). One site (Jewel Creek at Muskego) was dropped from the network shortly thereafter. Records from water year 2005 were not published for the Lincoln Creek site; however, the gage remained operational (Waschbusch and others, 2006); data provided for this time period are considered provisional and have been subjected to all quality-control checks and adjustments except those necessary for ice-affected records. Gages at three additional sites were added in April 2004: Honey Creek at Wauwatosa (04087119), Little Menomonee River at Milwaukee (04087070), and Root River at Grange Avenue at Greenfield (04087214). Standard USGS stream-gaging techniques were used to determine discharge at all of the stream gages (Rantz and others, 1982). Current-meter measurements of discharge at the three new sites were made every 4 to 6 weeks and more frequently during high flows to define the stage-discharge relation for each site. Current-meter discharge measurements at the remaining two sites (Willow Creek at Maple Road near Germantown and Jewel Creek at Muskego) were made during water-quality sampling.

Manual Water-Quality Sampling

Water-quality samples were collected manually at all 15 stream sites and 6 harbor sites (table 1A–B). USGS personnel collected samples at the 14 wadeable stream sites. MMSD personnel collected samples by boat at Milwaukee River at Mouth and the six harbor sites. Water-quality sample-collection protocols differed by the manner of collection. USGS field personnel used standard USGS depth-width-integrated collection methods for wadeable streams; samples were collected from nearby bridges when streams could not be waded because of high flows (U.S. Geological Survey, variously dated). MMSD field personnel collected composite samples from a boat using a Kemmerer sampler deployed at three depths; sampling techniques were modified from standard USGS sampling methods (appendix 2A)(U.S. Geological Survey, variously dated).

Sample collection was designed to encompass a range of flows. From February 2004 through September 2005, personnel from both agencies simultaneously sampled all 21 sites over a few days in an effort to obtain a snapshot of water quality. Samples were collected quarterly over a 2-year period, resulting in eight fixed-interval samples per site. Additionally, personnel collected samples during four targeted high-flow events over the 2-year period: one spring snowmelt event and one summer-storm event per year. Targeted high-flow events for sampling were generally defined as events in which the instantaneous discharge was greater than the 10-percent-flow-duration discharge at gages representing the major basins draining to the harbor (Milwaukee River at Milwaukee, Menomonee River at Wauwatosa, Kinnickinnic River at S. 11th Street at Milwaukee, and Root River near Franklin). Because MMSD sampling boats were not operated during winter months, no samples were collected from the harbor sites during those times; however, samples were collected at Milwaukee River at Mouth (as grab samples) so that all 15 stream sites could be sampled during winter months. Site locations are shown in figure 4.

Figure 4.  Locations and characteristics of Phase II sampling sites, and locations of National Weather Service precipitation gages in the Milwaukee Metropolitan Sewerage District (MMSD) planning area, Wis. Site abbreviations listed in table 1.

Ambient water temperature, pH, specific conductance, and dissolved oxygen were measured at all sites during sampling using a calibrated multiparameter meter. All water samples were brought back to a central processing area (vehicle or laboratory) where samples were processed (split, filtered, and preserved). All water-quality samples were delivered to the appropriate laboratory for analysis. All manually collected water-quality samples were analyzed for a variety of water-quality constituent groups: nutrients, major inorganics, carbon, wastewater compounds (WWCs) and microbes (table 2 and appendix 3). The laboratories used for analyses were: USGS laboratories (National Water-Quality Laboratory (NWQL) in Denver, Colo.; Wisconsin Water Science Center Mercury Research Laboratory in Middleton, Wis.; Iowa Water Science Center Sediment Laboratory in Iowa City, Iowa; Kentucky Water Science Center Sediment Laboratory in Louisville, Ky.) and the Wisconsin State Lab of Hygiene (WSLH) in Madison, Wis.

Table 2.  List of properties and constituents analyzed during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study.

[BOD, biochemical oxygen demand; COD, chemical oxygen demand; PCBs, polychlorinated biphenyls]

The general timeline for hydrologic, water-quality, and biologic data collection is given in figure 5. Samples were analyzed for most constituents during each sampling event, however, a few constituents were investigated less frequently during selected target conditions. Samples were collected for pesticide and water-column-toxicity analysis from all 21 sites during the spring-quarterly sampling (before pesticides were being actively applied) and during the summer-storm events (fig. 5). Samples for mercury analysis were collected at all sites during the summer-quarterly and event sampling.

Figure 5.  Timeline of data-collection activities for Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study. [WQ, water quality; X indicates a pesticide sample was collected during this event]

return to top

Mercury Sampling

Because of the generally low mercury concentrations in natural waters, extra care is required when sampling. To accurately quantify the concentration of total mercury (Hg) and methylmercury (MeHg) in water samples, trace-metal clean techniques were used to minimize sample contamination during collection, handling, and analysis. Analytical methods used for constituent detection were highly sensitive and operated with a reporting level of 0.04 ng/L for both Hg and MeHg. A brief description of sampling procedures is in appendix 2B; a more complete description can be found in Olson and DeWild (1999).

Microbiological Sampling

Samples for microbiological analysis were collected at each of the 21 sites during every sample visit and were analyzed by WSLH. WSLH holding times (for monitoring purposes) for fecal coliform and Escherichia coli (E. coli) were 24 hours; the microbiological samples (14 L) were transported to the WSLH on the day they were collected in order to adhere to this holding time. For 10 of the 12 sampling events, samples were also collected for analysis by MMSD. These samples (500 mL) were collected and delivered to MMSD in compliance with Wisconsin Administrative Code NR 219, which specifies that analyses for fecal coliform and E. coli must be analyzed within 6 hours of sample collection. Sample results from the two laboratories were compared to determine the degree of variability and the absolute difference between concentrations reported.

The WSLH data set was used for analyses of fecal coliform and E. coli in this report since all sampling events were represented and all other microbiological constituents were analyzed by this laboratory.

Bed-Sediment Sampling

Streambed-sediment traps were used in Phase II as a means to qualitatively assess the current transport rates of metals and polychlorinated biphenyl (PCBs) in specific stream segments. Bed-sediment analyses for metals and PCBs were less expensive than equivalent tests on water-column samples; moreover, because many metals and organic chemicals sorb onto fine sediment particles, bed-sediment analyses were an effective means of integrating contamination episodes in stream segments over extended time periods. Sediment-trap samples were analyzed for total PCBs, total phosphorus, particle grain size, arsenic, cadmium, chromium, copper, lead, mercury, nickel, and zinc. Sediment traps were deployed at each of the 15 stream-sampling locations. Two traps were deployed at each site during the 2004 and 2005 field seasons (approximately June–October 2004 and approximately April–August 2005). During each field season, sediment traps were deployed for periods ranging from 1 to 4 months.

These sediment traps were based on designs used by WDNR at Lincoln Creek (Baird and Associates, 1997). The traps were made from concrete blocks, plywood, PVC pipe, and acrylic plastic sheet stock (fig. 6). The traps primarily served to anchor and protect the glass jars used to collect the sediment samples. The jars used in the sediment traps were rinsed with acetone before deployment at the sample sites.

Figure 6.  Sediment traps used for collection of bed-sediment samples during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study.

If an insufficient sediment volume was collected at a site by the time of the midseason sample collection, the sediment that was collected was bottled and stored at 4°C, and the sediment trap was cleaned and redeployed. At the end of the season, the stored sample and the end-of-season sample were combined and submitted for analysis.

Water-Column-Toxicity Sampling

Acute toxicity of water-column samples was analyzed by WSLH using the Microtox Acute Toxicity Test (Strategic Diagnostics Inc., Newark, Del.). This analysis quantifies differences (expressed in percent effect) in the fluorescence values of bioluminescent marine bacteria exposed to sample water when compared to a laboratory control. Decreases in fluorescence (positive percent effect values) are attributed to the toxic effects of sample water on the viability of test bacteria; the magnitude of the change in luminescence relates positively to the toxicity of the sample water. Changes in luminescence were reported after 5 minutes and 15 minutes of exposure to environmental sample water (AZUR Environmental, 1998). Samples submitted for toxicity analysis were collected from all 21 Phase II sampling sites during summer-quarterly and event sampling.

Biological, Habitat Assessment, and Fish-Tissue Toxicity Sampling

One-time surveys were conducted in late summer and early autumn 2004 for fish-, macroinvertebrate-, and algal-community data, as well as habitat data at the 14 wadeable stream sites that were sampled as part of Phase II (table 1A). Fish, macroinvertebrates, and algae samples were collected according to the methods documented in Moulton and others (2002): fish—fish sampling protocols; algae—qualitative multi-habitat sampling protocols; macroinvertebrates—semi-quantitative targeted habitat-sampling protocols. Fish were collected, identified, and enumerated in the field. Fish-tissue samples were collected during fish-community surveys and analyzed for accumulated toxics (metals and PCBs) at three sites (Milwaukee River at Milwaukee, Menomonee River at Wauwatosa, and Root River near Franklin). Identification and enumeration of macroinvertebrate samples was done by the University of Wisconsin–Stevens Point Aquatic Entomology Laboratory, except for samples from the Little Menomonee River and Honey Creek sites, which were done by the USGS National Water-Quality Laboratory. Algae samples were identified and enumerated by the Academy of Natural Sciences in Philadelphia, Pa. Habitat assessments were performed using methods employed by the NAWQA program, as documented in Fitzpatrick and others (1998).

Semipermeable Membrane Device Sampling from the National Water-Quality Assessment Program

Semipermeable membrane devices (SPMDs) mimic biological membranes, such as the gills of fish, and can be used to predict contaminant exposure and accumulation in fish. These devices contain a synthetic lipid solution similar to that found in fish. SPMDs are used to gather time-integrated information on the presence of dissolved (biologically available) hydrophobic organic contaminants in water. Toxicity tests followed by chemical analyses were done on extracts from the triolein in the SPMDs. The Cytochrome P450RGS test assessed toxicity from PAHs, planar PCBs, dioxins, and furans in the water. The Fluoroscan test estimated the concentration of PAH compounds and was expressed in pyrene equivalents, and the Microtox test screened for acute toxicity from synthetic organic compounds.

As part of the USGS NAWQA Effects of Urbanization on Stream Ecosystems Topical Study in 2004, SPMDs were deployed at 30 sites in the Western Lake Michigan Drainages study unit. SPMDs were placed for one month at 15 Milwaukee-area sites, 7 of which were at or within a few miles of MMSD sample sites on these streams. Four NAWQA sites were at the same location as Phase II sites (Lincoln Creek, Oak Creek, Menomonee River at Menomonee Falls, and Little Menomonee River), and three NAWQA sites were near MMSD sites: Honey Creek near Portland Avenue at Wauwatosa (04087118, about 1 mi upstream from the Honey Creek at Wauwatosa Phase II site), Root River at Layton Avenue at Greenfield (04087213, about 2 mi upstream from the Root River at Grange Avenue Phase II site), and Underwood Creek at Watertown Plank Road at Elm Grove (040870856, about 3 mi upstream of the Underwood Creek at Wauwatosa Phase II site).

Toxicity tests were done by the U.S. Army Corps of Engineers in Vicksburg, Miss. (P450RGS test) and the USGS Columbia Environmental Research Center in Columbia, Mo. (Fluoroscan and Microtox tests). Chemical analyses of the SPMD extracts were done at the USGS National Water-Quality Laboratory in Denver, Colo.

return to top

Automated Water-Quality Sampling

Automated samplers were installed and maintained by USGS personnel and used to collect water-quality samples so that loads of suspended sediment, total phosphorus, and chloride could be computed. The four sites where the samplers were installed were Milwaukee River near Cedarburg (04086600), Milwaukee River at Milwaukee (04087000), Kinnickinnic River at South 11th Street at Milwaukee (04087159), and Menomonee River at Wauwatosa (04087120).

Approximately 8 to 10 samples were collected per storm. More samples were collected during times of increasing discharge, when the constituent concentrations were expected to have the greatest variation. Samples were analyzed for suspended sediment, total phosphorus, and chloride by the Wisconsin State Laboratory of Hygiene (WSLH). Storm, daily, and annual loads were computed for each site. Constituent loads were determined by multiplying constituent concentration by stream discharge and a conversion factor (Porterfield, 1972). Daily loads were determined by use of the integration method described by Porterfield (1972). Daily and annual loads were compared to historical data at these sites.

Quality Assurance and Quality Control

Quality-assurance and quality-control (QA/QC) samples collected during Phase II of the MMSD Corridor Study made up about 15 percent of the water-quality samples. QA/QC samples included both field blanks and replicates. Field-blank results were evaluated to estimate how analytical results might be biased by contamination of the sample from the sampling equipment, equipment cleaning, and sample processing. Replicate samples were evaluated to estimate the degree of variability in sample results. Results from the blanks are summarized in table 3, and replicate results are summarized in table 4.

Table 3.  Results of constituent detections in field blanks during Phase II of the Milwaukee Metropolitan Sewerage District (MMSD) Corridor Study.

[mg/L, milligram per liter; µg/L, microgram per liter; e, estimated; M, detected but not quantified]

¹ Field-blank results where constituent concentrations were less than 25 percent of the minimum environmental sample concentration were determined to be within data-quality limits, and were considered insignificant with respect to result interpretation (MacCoy, 2004).

Field blanks were collected by passing certified, analyte-free blank water through the cleaned sampling apparatus. Field-blank results where constituent concentrations were less than 25 percent of the minimum environmental sample concentration were determined to be within data-quality limits, and were considered insignificant with respect to result interpretation (MacCoy, 2004). Field-blank results where constituent concentrations exceeded 25 percent of the minimum environmental sample concentration were examined further to determine how contamination affected data interpretation. Some constituents were detected but not quantified in field blanks; detections of this type also are noted in table 3.

For most constituents, blank detections were found to be within data-quality limits (table 3). However, concentrations in blanks exceeded 25 percent of the minimum environmental sample concentration for silica (15 of 17 field blanks), biochemical oxygen demand (BOD) (10 of 17 field blanks), chlorophyll a (3 of 17 field blanks), and chemical oxygen demand (COD) (13 of 17 field blanks). Equipment-cleaning solutions may have been responsible for the high number of detections of BOD and COD in field blanks. Laboratory variability may have been responsible for the blank detections in chlorophyll a (a point that will be further illustrated in the sample-replicate result discussion).

Analysis of wastewater compounds (WWCs) involved new techniques whose methods for detection are challenged by issues of contamination and low reporting levels. Due to the small number of samples with concentrations above the reporting level, percent detections were used to discuss their occurrence. Phenol was detected in 9 of 17 field blanks as well as NWQL laboratory blanks; therefore, it was dropped entirely from analysis. N,N-diethyl-meta-toluamide (DEET) was not detected in NWQL laboratory blanks but was detected in field blanks. The majority of DEET field-blank contamination was at a level that could only be qualitatively detected; since this type of sample data were considered nondetections for the purposes of analyses, the impact of the contamination on the results discussed in this report was felt to be negligible (table 18, page 83). Any constituent result that was detected but not quantified was counted as a non-detect for the purposes of data analysis. DEET field-blank contamination was not fully understood, but may have been due to an atmospheric source, a proximity to field gear/sampling equipment containing residual insect repellent, or some other undetermined source (Kingsbury and others, 2006). NWQL laboratory blank results also indicated contamination of naphthalene and 1-methylnapthalene. With respect to field blanks, it is likely that phenol, naphthalene and 1-methylnapthalene contamination were also affected by the proximity of sample processing in or near gasoline-powered vehicles.

Of the constituents analyzed during Phase II, the majority of replicate-constituent results were within the quality-control limits. As a mathematical rule, relative percent differences (RPDs) between 2 small numbers are higher than the RPD between 2 larger numbers with the same value difference; in order to adjust for that phenomenon, replicate samples were evaluated at 3 levels based on the concentration of the analyte above the reporting level (MacCoy, 2004). Level 1 evaluation was defined for constituents whose concentration was between 0 and 5 times the reporting level, in which case a difference of 1 reporting level was allowed between replicate pairs. Level 2 evaluation was defined for constituents with detection between 5 and 20 times the reporting level. In this case, a maximum difference of the greater of 2 times the reporting level or 20 percent RPD was allowed. Level 3 evaluation was defined for constituents with concentrations more than 20 times the reporting level. Constituents at this level were not to exceed an RPD of 10 percent. Replicate pairs where 1 result was above and 1 result was below the reporting level were not considered where the higher result was within a reporting level difference of the reporting level; BOD and particulate inorganic carbon each contained a single replicate pair showing a difference greater than one reporting level. Very few replicate-pair results failed to meet quality-control limits (table 4). More than 1 sample pair for each of the following constituents exceeded quality-control limits: chlorophyll a (5 sample pairs), particulate organic carbon (3 sample pairs), particulate total carbon (2 sample pairs), phosphorus (2 sample pairs), BOD (2 sample pairs), COD (2 sample pairs), and 3,4 Dichloroaniline (2 sample pairs). One sample pair for each of the following constituents exceeded quality-control limits: particulate inorganic carbon, total nitrogen, particulate nitrogen, nitrate, pH, atrazine, 2-Cholor-4-isopropylamino-6-amino-s-triazine, tebuthiuron, and phenol.

Table 4.  Results of replicated sample issues during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study. The following analytes did not meet one of the three levels used for replicate-pair analysis based on parameter concentration (MacCoy, 2004). Each X represents a single replicate pair.

[mg/L, milligram per liter, µg/L, microgram per liter; °C, degrees Celsius; --, not applicable]

return to top

Special Cases of Quality Control for Selected Constituents

Due to the unique nature of the data, quality-control analyses of microbiological, bed sediment, and water-column toxicity were not subject to the same types of analyses as other water chemistry constituents.

Microbiological

Quality-control (QC) results for microbiological samples are known to have higher levels of variability than those for chemical-quality constituents. Analytical methods used for quantification of indicator and pathogenic microorganisms differed in strategy. Methods for indicator organisms generally analyzed sample volumes similar to what was reported for the result and generated data sets of continuous values. Methods for pathogenic organisms generally analyzed sample volumes that were quite different from the reported volume, and when the number of organisms detected were adjusted to the reported volume, it generated data sets of discrete values with many concentrations in partial organisms (for example, 33.3 oocysts/100 L). As a result of the difference in these data sets, QC analyses for indicator organisms and pathogenic organisms were evaluated separately.

Replicate-pair comparisons for indicator organisms were analyzed in much the same manner as other water-quality constituents; results were divided into three levels based on the concentration of the analyte above the reporting level (MacCoy, 2004). Replicate-pair analyses performed by WSLH are summarized in table 5, along with the criteria used for water-quality replicate-pair evaluation. However, given that microbiological samples are known to have higher variability than most water-quality constituents, data were not evaluated for violations of replicate criteria associated with each level. Variability was generally low to moderate for indicator organisms, with a few exceptions in the data sets for each indicator organism. The most notable difference was a non-matching replicate pair in the fecal coliform data set, where members of the replicate pair had concentrations of less than 10 and 200 col/100 mL. In addition, one of the replicate pairs in Level 3 of this data set differed by an RPD of 125 percent. Replicate pairs with the highest RPDs in the E. coli and coliphage data sets represented relatively small differences in actual concentrations.

Table 5.  Replicate results for indicator organisms analyzed by Wisconsin State Laboratory of Hygiene during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study. Replicate pairs were grouped into levels based on concentration above the reporting level (MacCoy, 2004). Criteria used for most water-quality constituents at each level are provided for general comparison; microbiological samples are known to show greater variation, however, and were not analyzed on the basis of these criteria. Difference in concentrations between replicate pairs is expressed as the difference and (or) relative percent difference, depending on the recommended criteria; these values are expressed as ranges.

[RL, reporting level; RPD, relative percent difference; col/100 mL, colonies per 100 milliliters; MPN/100 mL, most probable number per 100 milliliters; plaques/100mL, plaques per 100 milliliters, --, not applicable]

Fecal coliform and E. coli analyses were also performed by MMSD and replicate analyses for these samples generally indicated slightly less variability when compared to the WSLH data sets. All replicate pairs for MMSD fecal coliform and E. coli analyses (9 and 10, respectively) grouped into Level 3 and indicated RPDs of 9 to 59 col/100 mL (median of 30 col/100 mL) and 0 to 32 most probable number per 100 milliliters (MPN/100 mL)(median of 11 MPN/100 mL).

In addition to QC analysis of traditional replicate-pairs analyzed by each laboratory, analyses were performed on results for all (regular and QC) samples analyzed by both laboratories to determine the degree of variability and the absolute difference between results reported. Comparison of fecal coliform and E. coli results from WSLH and MMSD are illustrated in figure 7. Slightly more variability was observed among fecal coliform results (R2 value of 0.71), than among E. coli results (R2 value of 0.91). With relation to absolute difference, a line of one-to-one correspondence on these graphs would indicate that results reported by these laboratories were identical for concurrently-collected samples. The distribution of fecal coliform results for this comparison indicates that the body of data lies below the line of one-to-one correspondence, indicating that WSLH concentrations were, on average, lower than MMSD concentrations. The differences in concentrations may have been due to the variation in holding times between the two laboratories. More specifically, the longer holding times at the WSLH may have caused a loss of viability of the microorganisms, thereby causing these concentrations to be consistently lower than MMSD concentrations for fecal coliform. The distribution of E. coli results indicated that, on average, the two laboratories found very similar concentrations. In order to maintain consistency within the rest of the microbiological data set, analyses for fecal coliform and E. coli are limited to analytical results from WSLH.

Figure 7.  Inter-laboratory correlation between Phase II results for A, fecal coliform concentrations, and B, Escherichia coli concentrations from samples analyzed at the Wisconsin State Laboratory of Hygiene (WSLH) and Milwaukee Metropolitan Sewerage District (MMSD).

The discrete nature of the results in the pathogen data set required a different approach for replicate analyses; replicate pairs were compared in a manner reflective of analytical and reporting methods. E. coli O157:H7 data were reported as presence/absence; of the 16 replicate pairs analyzed, 1 replicate pair had mismatched results. Results reported for Salmonella were an interpretation of the minimum volume of sample that indicated the presence of the bacterium. Data were limited to values of 0.1, 0.2, 1, and 10 MPN/100 mL, and are reflective of positive results at analytical volumes of 10, 20, 100, and 1,000 mL, respectively. Of the 16 replicate pairs analyzed, 12 replicate pairs had identical results. In 3 of the remaining pairs, 1 member was not detected and the other member was detected in volumes ranging from 10 to 100 mL. Volumes of detection for the final pair were 10 mL and 100 mL. Giardia and Cryptosporidium replicate analyses were performed using 1.5 to 3.3 L of sample water; volumes were similar within each replicate pair. The numbers of detections in the analytical volume were extrapolated to the reporting volume of 100 L, and yielded discrete numbers of organisms based on the original sample volume. Replicate comparisons were performed based on differences in individual organism counts in the analytical volume (table 6). Differences were lower for Giardia than Cryptosporidium, but were low for both. The maximum difference observed was three organisms (for Cryptosporidium), but the majority of differences ranged from 0 to 1 organism.

Table 6.  Replicate results for pathogenic organisms during Phase II of the Milwaukee Metropolitan Sewerage District Corridor Study. Differences between replicate pairs are expressed as the difference between the actual numbers of organisms counted in the analyzed sample volume.

[RL, reporting level; MPN/100 mL, most probable number per 100 milliliters; cysts/100 mL, cysts per 100 milliliters; oocysts/100 mL, oocysts per 100 milliliters; --, not applicable]

Variability in the replicate analyses for indicator and pathogen data sets may have stemmed from sample-collection techniques (replicates were collected as sequential grab samples), or analytical subsampling techniques. Attempts should be made to limit variability in future studies; however, for the purposes of this study the affect of sample variability for both indicator and pathogenic organisms were thought to be negligible.

All blank QC results for microbiological constituents were within data-quality limits.

return to top

Bed Sediment

Seven duplicate bed-sediment samples were collected from six sites during the study. In 2004, duplicate samples were collected from the Milwaukee River near Cedarburg, Milwaukee River at Milwaukee, Menomonee River at Wauwatosa, and Oak Creek sites. In 2005, duplicate samples were collected from the Willow Creek, Honey Creek, and Oak Creek sites. The duplicate samples were obtained by collecting a completely separate field sample from a second sediment trap deployed near the primary sediment trap. This approach was a way to qualitatively assess the combined effects of in-stream, sample-processing, and laboratory variability. The range of the RPDs between duplicate samples ranged from near 0 to 130 percent. Table 7 lists the variability in sediment trap results for duplicate samplers.

Table 7.  Relative percent differences in sediment-trap results for six sites in Phase II of the Milwaukee Metropolitan Sewerage District (MMSD) Corridor Study.

[all values given in percent; --, not detected]

The largest source of variability associated with the sediment-trap samples appears to have been related to the variability of sediment-deposition patterns in the stream and the capture efficiency of the traps themselves. For example, duplicate samples at the Milwaukee River at Milwaukee site yielded two samples with recovered masses of 217 g and 1,714 g. Such a difference between recovered sample masses suggested that the traps were placed within differing sediment-deposition zones. Alternatively, one trap at this location may have been covered by debris for part of the deployment time. For the Milwaukee River at Milwaukee site, RPDs between analytical results for the duplicate samples ranged from 10 to 62 percent. Duplicate samples at Willow Creek yielded recovered sediment masses of 903 g and 921 g, suggesting that the traps were placed within very similar deposition zones. Correspondingly, the RPD between duplicates ranged from 4 to 14 percent.

Generally, the RPDs decreased for analytes as the mean concentration of the analyte increased relative to the reporting level. For the Milwaukee River at Milwaukee site, the RPD for mercury was 62 percent, with the mean concentration only about two times the reporting level. By contrast, the RPD for total PCBs at Milwaukee River at Milwaukee was about 10 percent, with the mean concentration of total PCBs greater than 40 times the reporting level.

For all sites, the pattern of contaminant concentrations was similar between duplicates for a given site; therefore, the relative site comparisons presented later in this report most likely reflect environmental conditions and are not the result of a single, irreproducible sample result.

Water-Column Toxicity

Six blanks and five replicate pairs were analyzed using the Microtox Acute Toxicity Test. Microtox results were reported as a percent difference in luminescence (compared to a laboratory control), therefore, members of replicate pairs could not be compared in the same manner as the majority of water-quality constituents.

For replicate-pair comparisons, percent effects were converted into percentages of laboratory-control fluorescence (that is, 100 percent minus the observed percent effect); variability in fluorescence percentages between members of replicate pairs was considered negligible, with RPDs ranging from 2 to 8 percent at the 5-minute reading and 3 to 8 percent at the 15-minute reading.

Results for blanks ranged from -3.98 to 29.50 percent effect. Although one blank had higher percent effects than many environmental samples (at 26.99 and 29.50 percent effect at 5- and 15-minutes, respectively), overall, the percent effects observed in blanks were generally higher than those in environmental samples. Given the consistency of the results and the general lack of water-chemistry analytes found in blank water used for this study, these higher values were not thought to be indicative of contamination with toxic compounds, and may instead be indicative of low osmotic strength in the water (which could potentially decrease the viability of the marine bacterial species used in this analysis).

return to top
continue to Evaluation of Data Collected During Phase II