Introductiom - There have been few published surveys of the mussels of the Mississippi River and its major tributaries in West Tennessee. First Pilsbry and Rhoads (1896, as cited in Ortmann, 1926) listed 12 species of mussels from Reelfoot Lake and five species from the Wolf River, Shelby County, Tennessee. Later Ortmann (1926) reported seven species from Reelfoot Lake and 12 species from the Obion River. Hoff (1943) and Najarian (1955) reported on unionids from Reelfoot Lake. One of us (Manning, 1989) reported 33 species of mussels from the Hatchie River. None of these surveys reported L. siliquoidea. However, relic shells of L. siliquoidea were found by one of us (DM) during a survey of Reelfoot Lake in 1985. This record was not published because of the possibility that these shells had been transported to the area by aborigines (see Hill, 1983). Starnes and Bogan (1988) did list L. siliquoidea as occurring in West Tennessee based on these relic shells. These relic shells are deposited in the Frank H. McClung Museum malacology collections, the University of Tennessee, Knoxville (Voucher # 947).

Methods - We collected living mussels by hand along the banks of the Wolf River, from Michigan City, Benton County, Mississippi, downstream to Rossville, Fayette County, Tennessee from the water's edge to a depth of approximately one meter. We sampled at Bateman Bridge in October, 1994 and spent 20 person-hours in the water sampling other sections of the river during September-October, 1995. We measured all living L. siliquoidea for length according to Hinch et al. (1989), recorded by sex based on external shell morphology, and returned to the specific location from which they were taken.

Dr. David H. Stansbery of The Ohio State University and Dr. Paul W. Parmalee of the University of Tennessee later verified the identity of L. siliquoidea shells. While Dr. Stansbery prefers L. radiata luteola (Lamarck, 1819) for the specimens from the Wolf River, our nomenclature follows Turgeon, et al. (1988). Voucher specimens are housed in the Frank H. McClung Museum malacology collections, the University of Tennessee (Voucher # 948).

We determined the length-age relationship of L. siliquoidea by observing the external annuli of 72 live individuals and from internal annuli of 16 thin-sectioned, empty shells according to Neves and Moyer (1988). Internal annuli were defined as lines crossing the nacreous layer and extending through the prismatic layer to the periostracum. Double and triple lines were frequently encountered and scored as a single annulus. Internal lines from the first four years of age were difficult to read, so, based on shell size, the first conspicuous annulus was arbitrarily assumed to indicate an animal starting its third year of growth. The age-length relationships of both internally and externally-aged individuals were fitted with the von Bertalanffy equation:

Description of the Study Site in the Wolf River

The Wolf River is in the West Tennessee Plain physiographic region or part of the western mesophytic deciduous forest (Barbour and Billings, 1988). We sampled the Wolf River between Michigan City, Benton County, Mississippi and Rossville, Fayette County, Tennessee (Fig. 1.). Approximately midway between the ends of the river is Bateman Bridge. The river at Bateman Bridge is a third to fourth order stream, and it receives water from an extensive cypress/gum swamp upstream that was formed by a natural sand dam deposited by Mount Tena Creek. The river at Bateman Bridge has a conductivity of 35 µS/cm, and during the fall of 1993, had a discharge of 3.4 x 105 m3/day and a total particulate suspended load of 3.7 x 103 kg/day (Neff and Hearnsberger, 1994).

Results - We first encountered this mussel during a Rhodes College ecology field trip to the Wolf River on October 31, 1994. We found 22 living and two dead individuals upstream from Bateman Bridge east of Moscow, Fayette County, Tennessee. We subsequently collected L. siliquoidea in the Wolf River from Michigan City, Benton County, Mississippi to Rossville, Fayette County, Tennessee. The relative abundance of this mussel and the number collected per hour per person collecting are given in Table 1.

Resulting length frequencies for three groups of living mussels (males and females) and all dead shells are given in (Fig. 2.). Sexual dimorphism in shell lengths is obvious from these distributions (mean length + sd for males was 101.7 mm + 11.0 and for females, 92.2 + 13.3 mm). These length-frequency distributions for the sexes were significantly different (Kolmogorov-Smirnov test X2= 17.01; male n=82, female n=65; P=0.039). The length-frequency distribution of dead shells reflects the lengths of the living individuals (all living mussels mean length was 97.7 mm + 12.8 and dead shells mean length was 101.2 + 11.4); these distributions were not significantly different (Kolmogorov-Smirnov test X2= 2.28; all living mussels n=147, dead shells n=28; P=0.45).

Length-at-age data for both external and internal rings are given in Fig. 3., based on internal examination of empty shells and external examination of living individuals. The internal lines were difficult to read; often multiple lines converged on the periostracum. Possibly rapid temperature changes and/or reproductive events resulted in these "disturbance" rings. These rings were counted as a single year's growth and our estimates of age from internal annuli, being conservative, may overestimate growth. Nevertheless, using external lines resulted in faster estimates of growth compared to using internal lines. Parameter estimates (and asymptotic standard errors) from the von Bertalanffy Equation, using both internal and external rings, were L_infin; = 107.0 mm (2.403), k = 0.1033 (0.024), and t_o = -4.59 (3.01). We did not observe any indication of mesoconch (sensu Clarke, 1986) shell stage.

Discussion - We are unclear why living L. siliquoidea has not been reported in Tennessee. It is common in Arkansas rivers directly across the Mississippi River from West Tennessee (Jenkinson and Ahlstedt, 1987), in oxbows of the Mississippi River to the south (Cooper, 1984), in Mississippi streams to the south of Tennessee (Cooper and Johnson, 1980; Hartfield and Rummel, 1985; Hartfield and Ebert, 1986) and has been collected in a direct tributary of the Mississippi River just north of the Tennessee state line in Kentucky (Wendell Haag, per.comm.). This mussel is a substrate generalist (Clarke, 1981). Many of its potential fish hosts (Watters,1994) are found in the Wolf River (Medford and Simco, 1971) and are widely distributed in Tennessee (Etnier and Starnes, 1993). We found L. siliquoidea to be the second most abundant mussel species. Our first and third through sixth most abundant species were also found by Pilsbry and Rhoads (1896, as cited in Ortmann, 1926) in their study of the Wolf River. If L. siliquoidea was present in the late 1800s at its present abundance, it seems unlikely Pilsbry and Rhoads would have missed this species in their survey.

Results given in Table 1 indicate that L. siliquoidea is more abundant in the Wolf River, relative to other unionids and in absolute abundance, the farther one goes upstream from Rossville, Tennessee. The patchy distribution of mussels and the uneven and often debris-covered bottom made quadrat sampling impractical. Searching efficiency and strategy did not change during collecting trips, so reporting abundances based on person-hours is presumably appropriate. According to Strayer et al. (1995), catch rates from timed searches correlate well with actual population densities if applied under carefully defined conditions.

The microhabitat that often yielded high densities of L. siliquoidea was vegetated areas (Sparganium sp.) below or along swiftly moving parts of the river. Perhaps dredging activities of the early 1970s in the Wolf River to Moscow, Tennessee significantly reduced this type of habitat in the lower reaches that we sampled (Hartfield, 1993).

The difference in size-frequency distributions between male and female shells was expected. Members of the subfamily Lampsilinae show such sexual dimorphism in shell structure (Cummings and Mayer, 1992). The agreement between the length-frequency distributions of living and dead shells suggests that mortality is not falling disproportionately at one end of the size range we observed.

The length of the smallest living male and female was 54.2 and 67.7 mm, respectively. According to Nalepa and Gauvin (1988), L. siliquoidea of these lengths in the Great Lakes are about five and eight years old. Likewise, the maximum shell length we recorded for males was 119.7 mm and for females, 115.5 mm. The longest L. siliquoidea that Nalepa and Gauvin (1988) report was approximately 80 mm and estimated to be 16 years old. Lampsilis siliquoidea from the Wolf River thus seem to grow longer and faster than individuals collected by Nalepa and Gauvin in Lake Erie (see Fig. 3). This is consistent with the work of Tevesz et al. (1985) who found that stream-dwelling L. siliquoidea are larger than lake-dwelling forms.

The population of L. siliquoidea in the Wolf River appears to be older than that sampled by Tevesz et al. (1985) in the Vermilion River in northern Ohio. The mean age of live-collected L. siliquoidea in the Wolf River was 20+ years, while those collected by Tevesz et al. (1985) was only 5.55 years. The Wolf River population may thus be dominated by very old individuals, as is indicated by the prevalence of large individuals. The mean length of all live-collected L. siliquoidea in our study was greater than that of the nine populations cited by Tevesz et al. (1985).

Based on this survey, L. siliquoidea in the Wolf River from Michigan City, Mississippi to Rossville, Tennessee appears to be most abundant immediately upstream from Bateman Bridge (see Fig. 1.). A large swamp is located just upstream from Bateman Bridge. Increased phytoplankton/periphyton production and export of organic seston from this swamp may greatly benefit the native unionids. This typical pattern of increased abundance of filter-feeders below lake outlets is discussed by Richardson and Mackay (1991). They describe an exponential decline in filter-feeder density with distance downstream from the outlet. Seston quality and quantity are important causes of this distribution. Rapid decline in mussel abundance with distance from a lake outlet was described by Bronmark and Malmqvist (1982,1984), again apparently due to food of higher quality being washed out of the lake. While the explanation of abundance based on food quality is intuitively satisfying, there may be other factors affecting mussel distribution below lake outlets. For example, lakes and wetlands may moderate stream flow and temperature, while reducing sediment load and pollution.

There are few studies dealing with mussel distribution patterns relative to stream outlets from lakes or swamps. Surveying other rivers of western Tennessee containing similar swamp/wetland habitats would determine if the pattern of high native mussel abundance downstream from swamps is an artifact to the Wolf River or a more general ecological phenomenon.

Acknowledgements
We thank Drs. D. Stansbery and P. Parmalee for confirming specific identity of these mussels, and Mark Hughes and an anonymous reviewer for improvements in the manuscript.

References Cited

Barbour, M.G. and W.D. Billings. 1988. North American Terrestrial

                           Relative Abundance               Rate
          Location                (%)            (No. collected/hour/person)

Michigan City to LaGrange         38.5%                     16.0
LaGrange to Bateman Bridge        33.1%                     10.8
Bateman Bridge to Moscow           7.0%                      3.9
Moscow to Rossville                1.0%                      0.4