Mr. Thomas Alcamo
U. S. Environmental Protection Agency

Dear Mr. Alcamo:

As requested by the U.S. Environmental Protection Agency (EPA), Tetra Tech EM Inc. (Tetra Tech) is submitting the enclosed final technical memorandum discussing stream conditions and recent fish sampling activities conducted in Conard's Branch and Richland Creek downstream of the Neal's Landfill site. This fmal memorandum incorporates minor comments received from EPA and U.S. Fish & Wildlife Service on the draft technical memorandum, which was originally submitted to you on July23' 2003. The memorandum is being submitted to assist EPA with a human health risk assessment currently being prepared for the site.

If you have any questions regarding the enclosed technical memorandum, please call me at (312) 946-6491.

This memorandum has been prepared for the U.S. Environmental Protection Agency (EPA) to describe stream conditions and fish sampling activities in Conard's Branch and Richland Creek near the Neal's Landfill site in Bloomington, Monroe County, Indiana. Descriptions are based on observations made by Tetra Tech EM Inc. (Tetra Tech) during oversight of fish electroshocking and sampling in November 2002 and May 2003. Normandeau Associates (Normandeau), a contractor for Viacom, Inc. (Viacom), conducted the fish electroshocking and sampling. Sampling was observed at the sites identified below.

In general, habitat diversity, species diversity, and overall fish biomass are expected to increase with increasing stream order (Vannote and others 1980). These increases were observed as sampling proceeded from Conard's Branch, a first-order stream, to the Richland Creek 5-mile site, a higher-order stream. Table 1 summarizes the tissue samples from Richland Creek and Conard's Branch that were submitted for analysis of polychlorinated biphenyls (PCB) in 2002 and 2003. Target species were selected to represent three feeding guilds: (1) benthic, (2) omnivorous, and (3) pelagic. Benthic species are "bottom feeders" and include fish such as suckers, redhorse, stonerollers, and catfish. Omnivorous species such as crayfish and creek chub feed in a variety of habitats and consume plants, animals, and detritus. Pelagic species such as sunfish and bass primarily inhabit open water and feed on other animals. Pelagic species are considered to be at the top of the food web because they cannot survive without other organisms as prey.

The standard length for a sampling reach in a stream is typically about 100 to 150 yards, but the actual sampling location may be longer or shorter based on field conditions and habitat availability. For example, at the Conard's Branch site, the reach identified for sediment sampling was about 150 yards long; however, fish sampling was conducted only in one large pool about 20 yards long. Fish sampling efforts focused on the pool because it provided the most appropriate habitat for the target species. At other sites, the length of the reach was extended when target species were not easily obtained. For example, at the Richland Creek 5-mile site, the first sampling effort in May 2003 covered about 150 yards. When Normandeau returned to the site the next day to obtain additional fish, the reach was extended about 300 yards downstream of the original reach, resulting in a total estimated reach length of about 450 yards.

The following sections summarize stream conditions and habitat at each of the four sites as well as observations made during electrofishing and sediment sampling conducted in November 2002 and May 2003. The Richland Creek 5-mile site was not sampled or observed in November 2002, and the Richland Creek upstream site was not sampled or observed in May 2003.

CONARD'S BRANCH SITE

Of the four sampling sites, the Conard's Branch site is closest to the Neal's Landfill site. Conard's Branch passes through the Conard farm, and most of the streamside woody vegetation has been removed to facilitate farming and other activities on the property. The sampling reach is characterized by lower- order, erosional habitat with sharp, distinct meanders and steeply incised banks 3 to 5 feet high. One large pool is present in the sampling reach. The rest of the reach was I to 3 feet wide and less than 6 to 12 inches deep. Much of the reach is scoured to bedrock and does not provide the diverse habitat necessary to support a diverse community of fish species. According to the U.S. Fish and Wildlife Service, most of the streambed habitat was removed and destroyed during sediment removal operations performed by Viacom in 1987 and 1988.

The geomorphic characteristics described above suggest that Conard's Branch experiences extreme spates, particularly during springtime high water levels. Spates are short-duration, extreme storm events accompanied by high flow velocity and significant streambed disturbance. Streambed disturbance during severe spates may cause erosional and depositional areas of the stream to vary greatly from year to year. These factors may contribute to extreme variations in sediment composition, size, and contaminant partitioning. For stream fish, the smallest size fraction of sediment is the most biologically active because it is routinely ingested during feeding and likely to be entrained in water passing over the gills. Biased sampling in depositional areas where fine matter accumulates may be necessary to reflect the greater bioavailability of the smallest sediment particles. This issue may also be relevant at higher-order sites in Richland Creek.

The sparse, shallow habitat in Conard's Branch is capable of supporting small fish and crayfish that are preyed on by herons and other wildlife. However, the habitat does not support larger fish likely to be eaten by subsistence and recreational fishermen. As a result of these stream conditions, it is not likely that anyone currently fishes or historically fished in this reach. In November 2002, Dale Conard, the farm owner, was surprised at the number of small creek chub obtained from the stream. The chub present in the stream are not considered an edible species, and subsistence crayfish trapping is not a common practice in central Indiana streams.

Continued sampling of biota at Conard's Branch raises the two concerns summarized below.

1. Many organisms in the reach may not have been present prior to the extreme regional drought of 1999. Under severe regional drought conditions, spring-fed streams may dry out in shallow rime and run areas. Ponded conditions in deeper areas would provide refuge for some fish, but many would move downstream as water levels dropped regionally, and some would die as resources became more limited. If the drought of 1999 disrupted stream flow and resulted in extensive pending of Conard's Branch, mature organisms present in 2003 may have recolonized this area after water levels rose in 2000. Creek chub have a lifespan of about 7 years, and large crayfish may be more than 10 years old. The response to the severe drought of 1999 may falsely indicate a decreasing trend in PCB concentrations detected in tissue samples for two reasons: (1) organisms historically present in the reach may have migrated downstream, and (2) organisms recolonizing the area may not have been uniformly exposed to environmental conditions in the sampling reach. The age and size of sampled specimens should be considered as factors that may bias statistical analyses.
2. Sampling efforts in November 2002 and May 2003 removed the largest creek chub present in the pool at Conard's Branch. With continued removal of the largest specimens from a small population, data may indicate a decreasing PCB trend simply because younger fish are sampled during each subsequent effort.

RICHLAND CREEK UPSTREAM SITE

The Richland Creek site upstream from the confluence with Conard's Branch was not sampled or observed in May 2003. The upstream site is predominantly depositional habitat with long pools, gentle meanders, and sluggish flow. In November 2002, the middle and upper portions of the sampling reach, upstream of the bridge at Vernal Pike, were 10 to 20 feet wide and I to 3 feet deep; riffle areas were rare. The middle and upper portions of the reach contained large piles of debris and numerous snags. Much of the debris appeared to have resulted from the removal of woody vegetation along the stream corridor. The removal of woody vegetation probably also contributed to the severe erosion along the stream bank. Downstream from the bridge at Vernal Pike, the lower portion of the reach was predominantly erosional habitat about 10 feet wide and I foot deep. Construction of the bridge at Vernal Pike probably contributed to pooling in the upper reach and increased the streambedgradient downstream of the bridge.

The Richland Creek upstream site provides appropriate habitat for most small fish species and adequate habitat for bottom feeders such as central stonerollers and white suckers. The shallow, unshaded pools provide only minimal habitat for pelagic feeders such as sunfish and bass. These fish feed in midwater and typically eat live prey; as such, these carnivores are assumed to be at the top of the food web in a stream ecosystem.

During sampling in November 2002, runoff from an upstream quarry reduced water clarity to less than 1 foot in the uppermost portion of the reach. The reduced water clarity complicated electrofishing efforts and probably degraded otherwise appropriate habitat for white suckers. The increased turbidity did not affect sample collection at the site because Normandeau extended the lower portion of the reach downstream until additional habitat was available for sampling. Although the runoff conditions did not affect sampling in this instance, similar conditions may complicate future sampling efforts. White suckers prefer clear, clean, silt-free water, and they spawn in sand and gravel beds. White suckers migrate upstream to spawn and would not be likely to pass through turbid runoff from quarries and construction sites. In May 2003, Dennis Williamson of the Monroe County Health Department stated that water quality at the site had improved after the quarry enacted best management practices to minimize suspended solids in runoff entering Richland Creek. Mr. Williamson also stated that area residents had historically fished for white suckers during their spring spawning runs in Richland Creek.

RICHLAND CREEK 1-MILE SITE

The Richland Creek l-mile site consists of highly erosional habitat with numerous side, channel, and point gravel bars; broad rime areas; long, shallow runs; and occasional deeper pools. Most of the reach was 10 to 30 feet wide and less than 2 feet deep, and areas immediately downstream of the bridge at Vernal Pike are scoured to bedrock. During electrofishing efforts, most of the creek chub and crayfish were collected from shallower habitat in the downstream portion of the l-mile site. Deeper runs and pools upstream of the bridge at Vernal Pike provide more suitable habitat for white sucker and sunfish species, but both species were collected throughout the reach.

The diverse habitat at the Richland Creek l-mile site supports a more diverse community of fish than the Conard's Branch or Richland Creek upstream sites. In November 2002, about 20 species of fish were observed at the Richland Creek l-mile site. This diversity represents the maximum typically expected for similar streams in the region.

In general, habitats that support greater species diversity generally produce more fish by weight. In conceptual terms, as stream size increases, the amount of habitat increases, and more species can coexist. As more diverse habitats are available with increasing stream order, new niches are created, and they are filled by new species. These species are expected to be larger fish that typically require more than one prey species; as the,habitat becomes richer, more prey species become available, and more top predators occur. This concept is particularly important in understanding the fish community present at the Richland Creek l-mile site. Although the site supports top predators such as sunfish, it does not provide enough pool habitat to support large numbers of large top predators. As a result, sunfish species encountered in the reach are likely to be small and numerous or large and few (see Figure 1).

Sampling results from November 2002 and May 2003 illustrate this concept. In November 2002, most long-ear sunfish observed were in two size classes: about 15 grams and about 30 grams. Neither class represents the maximum size for long-ear sunfish, which is greater than 60 grams. The smaller fish were about three times more common than the larger fish, suggesting that (1) the sunfish are segregating by size and (2) there is considerable interannualvariability in the abundance of long-ear sunfish. According to default life history models for long-ear sunfish in the White River of Indiana (Froese and Pauly 2003), long-ear sunfish reach maturity at about 3 years of age. At this age, they are about 11 centimeters long and weigh about 25 grams. Based on this information, members of the larger size class observed in November 2002 were more than 3 years old, and members of the smaller size class were about 2 years old. The default models for long-ear sunfish are based on populations in a large river expected to have a more productive food web and more abundant habitat. Sunfish populations in Richland Creek may not attain the same weights as sunfish in a larger river. As a result, the default model may slightly underestimate the ages of specimens collected from Richland Creek. The presence of only two size classes of long-ear sunfish at a ratio of 3: I suggests that the age structure of the population varies from year to year. This variability results from differences in breeding success, environmental conditions, and mortality in any given year. Because the 2002 sampling effort yielded an abundance of small long-ear sunfish but few larger specimens, the 2003 sampling plan provided for the collection of rock bass as an alternate to the sunfish species. During electrofishing efforts in 2003, Tetra Tech requested Normandeau to collect all long-ear sunfish and rock bass specimens observed, regardless of their size. Only three rock bass were observed in the entire reach, and all were large, mature specimens (greater than 100 grams). All three rock bass specimens were collected and sampled.

Interannual variability probably also occurs among other sunfish species in Richland Creek, particularly sunfish of the Lepomis genus. The Lepomis genus includes species such as long-ear sunfish, bluegill, green sunfish, and pumpkinseed fish. All these species are present in streams in Monroe County, and large specimens of bluegill and green sunfish, as well as some Lepomis hybrids, were observed during sampling in both 2002 and 2003. Fish in the Lepomis genus are so genetically similar that they often interbreed and form hybrids. Although the green sunfish may hybridize more readily and be more pollution-tolerant than the other species, from an ecosystem-level perspective, the Lepomis species is considered to be equivalent in terms of morphology, feeding guild, and ecological function. In any given year, one of the Lepomis species may outcompete the others, achieving a larger population than its competitors. Despite the interannual variability in the abundance of an individual species, the overall biomass of the Lepomis species is not expected to vary greatly from year to year. In some years, one species is expected to dominate, while in other years, other species may dominate. For this reason, analysis of trends in PCB accumulation in sunfish may need to consider sunfish collectively rather than an individual species. If it can be demonstrated that members of the Lepomis genus have a similar ecotoxicological response to PCBs, combining data for the entire genus may allow more robust statistical analyses. If this approach is adopted, future sampling plans could target any Lepomis species within a given weight range rather than a specific target species.

Figure 1 shows that white suckers, which represent the benthic feeding guild, were collected across a range of weights in both 2002 and 2003. Benthic feeders have a much simpler food web than top predators, and their abundance and size are not expected to vary as much either within or between species. Variation in the weights of white suckers collected in 2002 and 2003 is probably due to seasonal effects. White suckers collected during the spring are expected to have a greater range in weight based on (1) variable survival success during winter and (2) biomass loss as a result of spawning in females. In May 2003, the three largest white suckers collected ranged from 20.5 to 22.5 centimeters long. Two were males, and one was female. Both males weighed 120 grams, but the female weighed only 95 grams. Fish of the same length typically have comparable weights, so this female most likely lost about 20 percent of her weight during egg production and spawning. Female white suckers may produce 60,000 eggs high in lipids, possibly a continued source of PCB contamination to other fish that feed on them. White suckers collected in the fall experienced favorable temperatures and foraging conditions throughout the growing season and are expected to vary less in size. Figure 1 shows these relationships graphically and provides values for the regression coefficients (r2) of each sample set. The correlation between weight and length in fish is best predicted by a logarithmic relationship; a perfect relationship would have an r-value equal to 1.

Despite being less abundant or smaller than in downstream areas, sunfish are present in the reach in sufficient numbers to Support fishing. Many subsistence and recreational fishermen consider sunfish to be "panfish," and the sunfish present in the reach are adequate to make a meal of panfish. Panfish are typically too small to fillet and are instead fried whole in a pan. As many as a dozen or more small panfish might constitute a meal for one adult. Table 2 summarizes the target species sampled at the Richland Creek l-mile site in 2002 and 2003. The table includes only a portion of the fish encountered during electrofishing because (1) only the largest target specimens were sampled, and (2) nontarget species were not collected. In particular, other mature sunfish in the Lepomis genus were observed in both 2002 and 2003. As such, summary data in the table are biased low and represent only a portion of the biomass available to subsistence fishermen. Comparable data are not available to provide similar tables for the other sites sampled in 2002 and 2003. The Conard's Branch site does not provide sufficient habitat for edible species of fish, and too few specimens were sampled from the Richland Creek upstream site and 5-mile site to provide an adequate comparison.

Some residents near the Richland Creek l-mile site may not be aware of the potential for PCB contamination of the creek. In May 2003, property owners living adjacent to the sampling reach inquired about electrofishing activities. Tetra Tech informed them that fish were being collected and analyzed for PCBs and that fish and sediment in the creek could be contaminated. The property owners had recently moved to the area and were not aware of the PCB contamination in Monroe County. The property owners also indicated that children played in the creek.

RICHLAND CREEK 5-MILE SITE

The Richland Creek 5-mile site was not sampled or observed in November 2002. In May 2003, the sampling reach consisted of (1) an upper reach with deep, shaded pools full of woody debris; (2) a middle reach with long, shallow rime areas resulting from engineered alterations to the streambed, including a ford across the creek; and (3) a lower reach with deep, sluggish pools and engineered banks downstream of the bridge at State Road 43. During initial electrofishing efforts, only the upper reach was sampled. The reach was extended for additional fishing efforts downstream of the ford in the middle reach to below the bridge at State Road 43. In some parts of the upper reach, the tree canopy covered the entire stream corridor, providing optimal deep pool habitat needed by larger fish. Where the tree canopy had been removed or was not complete, the stream channel was steeply eroded and incised, and the banks were 12 feet high in some places. Most of the reach was 10 to 20 feet wide. One long pool just upstream of the ford was 3 to 4 feet deep; was nearly 30 yards long; and contained numerous snags, root balls, and woody debris. Additional pools farther upstream were not as long, but most were about 3 feet deep. Water clarity was excellent throughout the upper and middle reaches. Water clarity in the lower reach was diminished by earlier sampling efforts in upstream areas.

After only about 30 minutes of electrofishing in the pools in the upper reach, Normandeau had obtained 14 large, mature redhorse specimens. Smaller redhorse specimens were not observed during the initial sampling effort. Shortheadredhorse and black redhorse were collected in equal abundance, and most were in the size class of about 200 to 300 grams. All the specimens collected were at least twice as large as the average of the target weight range specified in the sampling plan. However, the large specimens were typical of mature redhorse, and redhorse of this size are most likely to be eaten by area fishermen. In addition to the numerous large redhorse, Tetra Tech observed at least 10 mature specimens of white bass and largemouth bass in the upper reach pools. Bass specimens that surfaced during electrofishing were about 8 to 12 inches long and would be considered a good catch by subsistence and recreational fishermen. Numerous white suckers and various species of sunfish were also observed during electrofishing in the upper reach. At least six of the suckers were similar in size to the redhorse specimens. The sunfish were large, mature specimens and about 5 to 6 inches long.

The abundance of large specimens in numerous feeding guilds coupled with the absence of smaller specimens suggests the points summarized below regarding the dynamics of the pool habitat in the upper reach.

• The diversity of mature fish suggests that the pools provide consistent habitat, with little year-to- year variation in flow conditions and microhabitat availability. The ford across the stream may stabilize the location of the upstream pools and allow for more consistent flow regimes.
• Fish are congregating by size, and smaller redhorse specimens are not expected to coexist with larger redhorse. Fish congregate by size for a variety of reasons, including decreased predation risks, increased success in interspecies competition, and limited availability of suitable habitat. As fish increase in size from juveniles to adults, suitable habitat typically becomes more limited, and large fish of many species may occupy different niches in the same pool.
• The location of the 5-mile site suggests that large redhorse and suckers may be exposed to (1) contaminated prey drifting downstream and (2) contaminated upstream areas encountered during spawning migrations. Deep, shaded pools in the upper reach probably trap organisms from contaminated areas upstream. Most aquatic organisms are subject to "stream drift," which results from the unidirectional flow of moving water. Most stream invertebrates and juvenile fish drift downstream at some point in their lives. During spring spates, aquatic organisms may be transported considerable distances downstream. Many aquatic species offset this drift by adopting spawning strategies that involve upstream migration. Redhorse and suckers in the upper reach pools are expected to migrate to upstream areas to spawn. Both redhorse and suckers migrate to spawn in sand or gravel beds in clean, clear water with a minimum flow velocity typical of lower- order streams.
• The tree canopy extending over the pools increases the biomass produced by the aquatic habitat. Food sources in the streamsidehabitat (such as insects and amphibians) provide additional prey for fish in the pools. These high-quality food sources may account for a significant portion of the diet of large fish and are less likely to be contaminated by PCBs at the 5-mile site. In addition, shade provided by the tree canopy keeps the water cooler; cooler water holds more oxygen, thereby reducing the likelihood of low-oxygen stress for large fish. The shaded streamside also reduces the amount of light entering the pools. Fishery research has shown that (1) some fish segregate by size under certain light conditions and (2) low light levels may contribute to increased growth rates.

The target weight range for shorthead redhorse of 50 to 100 grams specified in the sampling plan is lower than the typical weight of mature redhorse. During initial electrofishing efforts in the upper reach, all the redhorse specimens observed exceeded the target weight range. Because the sampling plan called for smaller fish than those present in the stream, Viacom instructed the electrofishing team to return to the site in an effort to collect smaller redhorse specimens. EPA instructed Viacom to ( I ) use any smaller redhorse specimens obtained as whole body samples and (2) collect fillet samples from the large redhorse that were more representative of stream conditions. Viacom agreed to collect fillet samples from one large, one medium-sized, and one small shorthead redhorse specimen that exceeded the target weight range. The largest of the 14 redhorse specimens collected was a shorthead redhorse weighing 368 grams; 117 grams of fillet tissue from both sides of this specimen was collected to meet the extra volume requirements for a matrix spike/matrix spike duplicate sample. The smallest redhorse specimen was also a shorthead redhorse weighing 137 grams, and fillets from both sides of the fish were collected as duplicate samples. The remaining 12 black and shorthead redhorse all weighed 200 to 300 grams. Tetra Tech collected a split sample fillet from a shorthead specimen weighing 244 grams.

When the electrofishing team returned to the Richland Creek 5-mile site, more than 3 hours were required to find three shorthead redhorse that weighed less than 100 grams. One such specimen was collected in the 200-yard upper reach upstream of the ford. This location was the same reach where larger redhorse were collected in abundance. The second specimen was obtained in the 250-yard middle reach between the ford and the bridge at State Road 43. The third specimen was obtained about 50 yards south of the bridge at State Road 43.

Any of the large redhorsecollected in the upper reach pools would be considered a good catch by sport or subsistence fishers. The pools apparently also support numerous large specimens of bass, sunfish, and white suckers; catfish and bullheads are also likely to occur in the reach.

During the electrofishing effort, Brian Mann, the owner of the property near the 5-mile site, arrived on site. Mr. Mann introduced himself and asked about the stream investigation. He indicated that although he and his family did not eat fish from Richland Creek, other people fished near the 5-mile site. Mr. Mann indicated that some of the people who fish in the reach may be generally aware of PCB contamination in upstream areas but may believe that they are far enough downstream to avoid catching contaminated fish.

CONCLUSIONS

Low concentrations of PCBs in sediment, dissolved in surface water, and accumulated in fish tissue represent the greatest continuous threat to organisms in Richland Creek. Fish at all sites sampled are likely exposed to continuous, low concentrations of PCBs as these compounds are partitioned in certain media and then released for uptake, dissolution, or deposition in other media. For example, a female white sucker may ingest PCBs in sediment; the PCBs are then partitioned in tissues with a high lipid content, such as eggs. When these eggs are released, the PCBs may be incorporated into the biomass of the fertilized egg, or if the egg is unfertilized, may dissolve into the sediment. The PCBs may also reenter the food chain when other fish eat the eggs or the larval fish. For stream fish, the smallest size fraction of sediment is the most biologically active because it is routinely ingested during feeding and is likely to be entrained in water passing over the gills. Biased sampling in depositional areas where fine matter accumulates may be necessary to reflect the greater bioavailability of the smallest sediment particles.

Figure 2 presents the four sampling sites on a continuum of species diversity and biomass. The figure should be considered conceptual because diversity and biomass estimates are based on best professional judgment and qualitative observations of the fish populations and habitat present in 2002 and 2003. Diversity and biomass are expected to be positively correlated over the length of the watershed; however, they may not exhibit the linear correlation shown in the dashed trend line in the figure.

Figure 2 indicates in a conceptual way that species diversity and biomass at the Conard's Branch site and the Richland Creek 5-mile site are well correlated. The relationship between diversity and biomass is less correlated at the Richland Creek upstream and l-mile sites. At the Richland Creek upstream site, habitat uniformity and degraded water quality caused by quarry operations in the area may have resulted in less diversity than at the 1-mile site. The upstream site lacks rime areas that would provide habitat for numerous small stream fish, reducing potential species diversity. Although species diversity may be reduced by the lack of rime areas, the long, shallow runs provide habitat for larger fish that account for greater biomass. At the l-mile site, habitat diversity allows greater species diversity, but the habitat is only minimally sufficient to support large numbers of mature top predators and bottom feeders. If optimal habitat is present between the l-mile and S-mile sites, fish species diversity is expected to increase with distance from the l-mile site. Diversity is also expected to increase downstream but at a much more modest rate because most species in the stream already occur at the l-mile site.

Data from fish collected at the Richland Creek l-mile and S-mile sites indicate that both sites are productive enough to support subsistence fishing. The property owner at the S-mile site indicated that people fish from the creek near his property and may not be aware of potential PCB contamination in that reach of Richland Creek. Residents living near the l-mile site were new to the area and were not aware of PCB contamination in any of the creeks in Monroe County. These residents indicated that children play in the creek at the l-mile site. Landscape features of the l-mile site might make it especially attractive to subsistence fishers. The site is easily accessible from Vernal Pike, but wooded areas in the lower reach block any view from the road.

Future sampling efforts could select target species based on their feeding guild or genus rather than focusing on a particular species. Considering a genus or a feeding guild as a whole is an approach that is well supported by the most basic theoretical assumptions of population and community ecology. The two approaches are summarized below.

1. Any species in a given feeding guild is collected, as long as it is within a specific weight range. The target weight range is a proxy measure of a fish's age and must be determined for each species potentially sampled. For example, selecting any benthic feeder at least 5 years old would provide more consistent data on PCB accumulation in bottom feeders. Likewise, providing target weight ranges for sunfish of a certain age would allow sampling efforts to obtain specimens that represent an age-based approximation of PCB uptake for fish at the top of the food web.
2. 2. Any species in a given genus is collected, as long as it is within the specified weight range. Fish species in the same genus are genetically similar and exhibit similar life histories, feeding behaviors, and physiological and toxicological responses. This approach is more conservative than selecting specimens by feeding guild. For example, combining data for sunfish in the Lepomis genus would be more conservative than combining data for a Lepomis species and a rock bass, which is in the Ambloplites genus. If data were combined based on genus, statistical analyses may be more robust, and future sampling efforts would not be affected by interannualvariation in the abundance of a particular species in a genus. This approach would allow sampling efforts to collect the most representative specimens of a genus in a given weight and age range. In practical terms, this approach would allow for the collection of specimens in a predetermined weight range that represents a particular age (for example, any redhorse in the Moxostoma genus or any Lepomis sunfish species).

REFERENCES

Froese, R., and D. Pauly. 2003. Fish Base World Wide Web Electronic Publication. On-Line Address: www.fishbase.org/ Updated June 16, 2003. Accessed on July 14 and September 25,2003. Vannote, R.L., and others. 1980. The River Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences. Volume 37. Pages 130 to 137.