2009 NORTHEASTERN NATURALIST 16(3):339–354 Population Attributes of an Endangered Mussel, Epioblasma torulosa rangiana (Northern Riffleshell), in French Creek and Implications for its Recovery Darran L. Crabtree1,* and Tamara A. Smith2,3 Abstract - Baseline information on the current status of rare mussel populations in a given water body is necessary to better understand the effects of actions to protect or enhance populations. We conducted river-wide surveys in French Creek, known for its mussel abundance, to quantify population-level indicators that could be used for measuring viability of Epioblasma torulosa rangiana (Northern Riffleshell), a critically imperiled freshwater mussel, in recovering or reintroduced populations. We estimated multiple attributes of Northern Riffleshell populations, including longitudinal distribution, densities and abundances, sex-specific age structure, and mortality rates. Northern Riffleshell has been documented in French Creek since at least the early 1900s and is distributed unevenly throughout the creek (12 of 32 sites in a bimodal pattern), with no animals found in the upper third of the creek. At sites containing Northern Riffleshell, site-specific densities ranged from 0.009–6.668 m2. Maximum age of Northern Riffleshell ranged from 7–11 years at four sites with evidence of sustained recruitment (i.e., uneven age structure). Proportions of individuals in each age class were similar at each site, even though total numbers of animals differed by up to two orders of magnitude. Significantly more males than females were found in early ages (1–3), but no significant differences were found in older age classes. There were no significant differences in mortality rates (both sexes combined) at all four sites. However, mortality rates differed significantly between the sexes at older ages (ages 6–10) at the one site with enough individuals for a comparison, suggesting greater reproductive costs or selective predation for females. The population attributes of Northern Riffleshell from French Creek are important benchmarks for setting restoration goals and measuring success in other systems that share a similar biogeography to French Creek, but whose fauna has been depleted (e.g., many Ohio River tributaries). Introduction Freshwater mussels (Unionidae) are one of the most endangered groups of organisms in the United States (Stein et al. 2000, Williams et al. 1993). Up to 35 species are believed to have become extinct in the last 150 years and more than 60% are considered at risk of extinction (based on NatureServe 2007 ranking of G3 or higher; Stein et al. 2000). The genus Epioblasma is considered the most threatened genus of freshwater mussels (Stein et al. 2000). Epioblasma torulosa rangiana Lea (Northern Riffleshell) was 1The Nature Conservancy, Allegheny College, Meadville, PA 16335. 2Pennsylvania Natural Heritage Program, Western Pennsylvania Conservancy, 11881 Valley Road, Union City, PA 16438. 3Current address - US Fish and Wildlife Service, Twin Cities Field Office, 4101 American Boulevard East, Bloomington, MN 55425.*Corresponding author - dcrabtree@tnc.org. 340 Northeastern Naturalist Vol. 16, No. 3 added to the Federal Endangered Species list in 1993, and a recovery plan was produced in 1994 (US Fish and Wildlife Service 1994). The recovery plan identifies the current and historic range of Northern Riffleshell and actions that should be taken to aid in the restoration of this species. Little is known about populations of this species under relatively healthy conditions. Attributes of unaltered populations are important in estimating the success or failure of augmentation or reintroduction programs (Kuehnl and Watters 2005, US Fish and Wildlife Service 1994). River-wide population assessments of freshwater mussels in French Creek revealed high densities and large populations of many species including Northern Riffleshell (Smith and Crabtree 2005). Multiple indicators of system integrity have revealed that French Creek’s water quality and in-stream habitat quality are currently good to very good (Rankin and Armitage 2005, Smith and Crabtree 2005, Smith et al. 2003). The combination of high numbers of mussels across multiple sites and high water and habitat quality reveal a system that is relatively intact compared with other rivers in the Ohio basin. Northern Riffleshell was historically found in French Creek (Ortmann 1919) and continues to exist at many sites in the mainstem of the river (Smith and Crabtree 2005). We estimated distributions and demographic statistics to determine the current status of this species within French Creek. Population parameters such as densities, sex-specific age distributions, and instantaneous mortality rates from sites with viable Northern Riffleshell populations are presented. These attributes of viability will be useful for long-term monitoring of this species in French Creek, and also provide biological criteria for assessing recovery projects outside of French Creek. Restoration indicators that go beyond presence or absence of a species are especially important indicators for long-lived organisms because adults may be present, but reproduction may not necessarily be occurring (e.g., Plethobasus cicatricosus, (Say) [White Wartyback]; Parmalee and Bogan 1998). Field-site Description French Creek is part of the Allegheny River watershed and the greater Ohio River drainage. The entire French Creek watershed covers an area of approximately 3200 km2. Approximately 93% of the watershed is within Pennsylvania, and the remaining 7% is made up of headwater streams in New York. French Creek originates in Chautauqua County in New York State, then flows south through Pennsylvania’s Erie and Crawford counties, through the northeast corner of Mercer County, and finally into Venango County where it flows southeast to its confluence with the Allegheny River at Franklin, PA (Fig. 1). Methods Site selection and qualitative surveys We used an approach similar to Villella and Smith (2005) for selecting sites to estimate whole-river populations. The main-stem of French Creek was mapped using a Trimble GeoExplorer GPS unit from its confluence with 2009 D.L. Crabtree and T.A. Smith 341 the Allegheny River upstream to the town of Sherman, NY (small streamsize precluded sampling farther upstream). Stream reaches were classified into one of three flow regimes: pool, run, riffle, or a combination of these regimes. We eliminated all pool habitats from our sampling universe, as most of the rare mussels in this system are not often found in these slow-flowing areas. From the riffle and run habitats that remained, we used ArcView Figure 1. Survey site locations in the French Creek watershed (southwest NY, northwest PA). Starred sites are those surveyed in 2004 using quantitative methods. 342 Northeastern Naturalist Vol. 16, No. 3 3.2 to measure each discrete flow type and set a 100-m length threshold to define large lengths of similar flow types (hereafter referred to as a habitat unit). We randomly selected one habitat unit in approximately every 5.6 river km to survey. The search areas at each habitat unit were standardized to 2500 m2 per site, with a total search time of 300 min/site. This protocol yields an effective sampling fraction of 0.06, which is identical to what has been used in the Allegheny River (Smith et al. 2001). We flagged the banks dividing the 2500-m2 area into sub-areas, and one observer was assigned to each sub-area. The number and size of the sub-areas depended on the number of observers. We used masks and snorkels for all sampling, and each snorkeler collected as many freshwater mussels as possible during their allotted search time. Live mussels were identified, counted, measured, and returned to the original collection site. Catch per unit effort (CPUE) was calculated as the number of freshwater mussel individuals collected divided by personhours spent sampling. Three 5.6-km stretches had no large riffle habitat, and so pools were sampled instead, and three additional large riffles were randomly selected within three 5.6-km stretches that already had one large riffle selected. Thirty-two sites total were surveyed qualitatively in 2003 and 2005. From the 2003 qualitative sampling results, 10 sites were randomly selected, stratified by CPUE to conduct quantitative surveys (Fig. 1; Smith and Crabtree, in press). Quantitative surveys and population estimation Although qualitative surveys can be conducted relatively quickly, they provide only a relative estimate of the size of a population and can be biased toward larger animals (Strayer and Smith 2003). At a subset of sites that were searched qualitatively, a double-sampling technique using replicate quadrat samples (0.25 m2) was used (Smith et al. 2001). Double sampling yields more accurate estimates, because not all mussels can be detected at the surface. Prior to conducting the double sampling, we first obtained the total number of quadrats to be visually searched at a site and the subset of these quadrats that must be excavated. The total number of quadrats to be searched visually depends on the desired precision of the abundance estimate. Based on similar studies, a population abundance coefficient of variation (CV) of 0.30 provides acceptable precision (Smith et al. 2001). Previous mussel data from two locations on French Creek showed a wide range of densities (0.006–2.327 m-2 per species), depending on species and site. Mean total densities for these two sites were roughly 0.38 and 0.19 m-2 (Greg Zimmerman, EnviroScience Inc., Columbus, OH, pers. comm.). We used the lower total mean density and accepted a CV of 0.33. Based on this value, we determined we needed 400 quadrats per site. In order to get a sample size of 400 quadrats within our average study area (2500 m2), a grid of 4- x 4-m cells was used to partition the study area. Equally spaced transects (lead lines marked with 0.5-m intervals) were placed parallel to shore. We generated three pairs of random numbers between 0.5 and 4.0 m both across and upstream, to use as coordinates for 2009 D.L. Crabtree and T.A. Smith 343 three quadrat samples within the grid. The remaining quadrats were placed at standardized distances (4 m) upstream from the three start points. We determined the number of quadrats to be excavated by conducting a preliminary random sample of 5% of the total quadrats. Within these quadrats, we conducted paired surface observations and excavations to a depth of 10–15 cm (see below for details). The proportion of mussels found on the surface compared to the mussels found in the subsurface (excavated) was used to calculate the proportion of the total quadrats that must be excavated (Smith et al. 2001). Quadrats were sampled from downstream to upstream to maximize visibility. Divers worked in teams of two or three, and a maximum of four teams worked simultaneously. Surface surveys involved visually and tactually searching through the substrate surface within the 0.25-m2 quadrats. For subsurface samples, a metal scoop was used to excavate all substrate to a depth of 10–15 cm or hardpan. All material removed by the scoop was sieved through a 0.63-cm mesh screen and inspected for mussels. After excavation, the quadrat was visually examined for any remaining mussels. All excavated mussels were identified, measured to the nearest 0.1 mm, and placed in separate underwater mesh bags from the mussels collected at the surface. After the data were recorded, the excavated substrate and the mussels were placed back into their original quadrat. The number of surface vs. buried mussels was calculated to determine how many of the remaining 400 quadrats required excavation (see Smith et al. 2001 for thresholds). In general, if most of the mussels were found at the surface during the preliminary sampling then few excavations were required. Conversely, if few mussels were found at the surface and many were found in excavations, then potentially all of the remaining quadrats required excavation. We used the Mussel Estimation Program (Version 1.4.3) developed by the David Smith at the USGS-Leetown Science Center Aquatic Ecology Lab to estimate Northern Riffleshell density and population size at all sites where it was collected in the quantitative surveys (USGS Mussel Estimation Program 2004). Population analyses We developed length-at-age relationships from mussel shells collected from muskrat middens along the length of French Creek during 2002. A total of 22 middens was sampled in 2002. Those shells were identified and counted, but only Northern Riffleshell valves were aged and measured to the nearest 0.1 mm. Although aging mussel shells is somewhat controversial (Downing et al. 1992), similar techniques have been shown to be extremely valuable in the study of fish growth and production (Casselman 1983). We believe age-structure analysis has utility for freshwater mussels, depending on the technique used (Neves and Moyer 1988) and the species (i.e., longerlived species may be more problematic at old ages). Recently, an attempt to validate annular growth bands provided strong support for their formation in 344 Northeastern Naturalist Vol. 16, No. 3 multiple freshwater mussel species; however, Northern Riffleshell was not studied (Haag and Commens-Carson 2008) We randomly chose an equal number (117) of male and female Northern Riffleshell in each size class for aging. For the smallest (20–30 mm) and largest (70–80 mm) size classes, we had less than the number of individuals available in the middle size classes. We counted each external band as an annulus. Aging was straightforward except for possible spawning or nonwinter- related temperature change checks, and a little erosion at the umbo. Of the 224 animals aged from external bands, 25 males and 25 females were randomly selected for internal band counts to help verify external count estimates (Neves and Moyer 1988, Rogers 1999, Rogers et al. 2001, Veinott and Cornett 1996). We thin-sectioned both valves of each individual using a Buehler Isomet low-speed saw. Thin sections were placed on a microscope slide and viewed under dissecting scope. Opaque bands that originated at the umbo and extended to the periostracum (assumed as annuli) were identified following methods of Neves and Moyer (1988). Eight of the 50 shells could not be read confidently, and so our final number of paired valves was 18 females and 24 males. We tested the slopes of the paired internally and externally derived annuli for both sexes combined using an ANCOVA test for homogeneity and found no significant differences (P = 0.82, n = 42). Thus, external annuli in Northern Riffleshell appear to be formed at similar rates to internal annuli. Based on these results, we used length-at-age curves from external annuli to estimate age structure using linear regression. Male and female lengths-atage were both significantly positive (P < 0.0001, n = 117 for both; Fig. 2). The slopes of the sex-specific length-at-age curves derived from external aging were significantly different from each other (ANCOVA test for homogeneity: P = 0.029, n = 117), so we used separate equations to predict age: y = 0.1406x–1.2961 (female; r2 = 0.75), and [1] y = 0.1652x - 2.1175 (male; r2 = 0.78), [2] where x = length in mm across longest axis. We estimated frequency distributions of ages at all quantitatively sampled sites with viable Northern Riffleshell. We considered Northern Riffleshell to be viable at a site if there were at least three age classes present. Of the seven quantitatively sampled sites that contained Northern Riffleshell, only five were considered viable. Sites RKM 52 and RKM 111 had evidence of recent recruitment (individuals ≤30 mm), but few or no adults. Although site RKM 126 contained viable Northern Riffleshell, our sampling effort was inadequate to estimate their abundance with acceptable confidence. In general, age distributions showed evidence of gear selectivity, and thus we considered Northern Riffleshell to be fully recruited to the sampling gear at age-2. For the four viable sites (and without RKM 126), we contrasted female and male proportions in each age using paired t-tests. Linear regressions, paired t-tests, and ANCOVAs were all performed using SAS 9.1. 2009 D.L. Crabtree and T.A. Smith 345 We constructed catch curves of Northern Riffleshell for each site by plotting abundance vs. age. We derived instantaneous mortality rates (Z) by transforming (loge) the catch-curve data from each site and then fitting a linear equation (Gulland 1983). Analyzing the age distributions of each sex separately was problematic due to the females’ relatively low representation in early age classes, and we had some concern that early ages may have been incorrectly sexed in the field. For the one site with enough individuals (RKM 74) and focusing on older ages (ages 6 and older), we were able to compare instantaneous mortality rates for male and female. Catch curves examine age structures at a single point in time (Tokey et al. 2007). They assume that catchability is constant across ages, recruitment shows no trend but can fluctuate randomly about a stationary mean, and the population is in equilibrium (i.e., no change in mortality or no trend in recruitment). We used the abundance-at-age equations to back-calculate age-1 abundances for both sexes combined at each site. Results The longitudinal distribution of Northern Riffleshell in French Creek was uneven (Fig. 3). Northern Riffleshell was missing at all sites in the upper third of the creek. Their densities declined significantly with distance upstream from the confluence with the Allegheny River (P = 0.017, n = 32). Figure 2. Length-at-age functions for female (r2 = 0.74) and male (dashed line, r2 = 0.75) northern riffleshell from externally aged shells. Equations derived from the externally aged shells were used to predict age from length for individuals at quantitatively sampled sites. Slopes of male and female derived from externally aged shells were significantly different (P = 0.03, n = 117). Note: Xs in graph overlap and cover many Os. 346 Northeastern Naturalist Vol. 16, No. 3 However, there were two centers of high density between river km 4–29 and 52–74 among other scattered sites of lower densities (Fig. 3). It should be noted that quantitative sampling revealed an additional site for Northern Riffleshell at RKM 98. For the subset of sites where Northern Riffleshell was present and sampled quantitatively, mean density estimates ranged from 0.01–6.67 m-2, and abundance estimates ranged from 25–16,633 (Table 1). For sites where uneven ages were observed (i.e., viable sites), mean densities ranged from 0.13–6.67 m-2, and abundance estimates ranged from 336–16,633 (Table 1). The age-specific mean densities of Northern Riffleshell at these sites varied greatly (Fig. 4a). The proportions of individuals-at-age were less variable (Fig. 4b), but differences between the proportions of female and males were apparent at early ages (Fig. 4c). The mean proportions of females to males were significantly different for ages 1–3, with many more males than females (P = 0.024 for age-1, P = 0.028 for age 2, and P = 0.015 for age 3; n = 4; Fig. 4c), but by age-4 the proportions were not Figure 3. Longitudinal distribution of Northern Riffleshell in mainstem of French Creek, based on qualitative sampling only. CPUE is based on 5 people- hours 2500 m-2. CPUE was log10- transformed to illustrate general pattern. Quantitative sampling revealed two additional locations for Northern Riffleshell at RKM 89 and RKM 98. Table 1. Densities (no. m-2), population estimates, and presence of uneven age structure (at least three age classes) of Northern Riffleshell at seven sites sampled using double sampling (quantitative) in French Creek. Site numbers refer to river km and numbers in parentheses are one standard error. Population estimates are based on a survey area of 2500 m2. River km 19 23 52 68 74 89 98 Density 1.85 1.18 0.04 0.47 6.67 0.01 0.13 (1 std. error) (0.22) (0.23) (0.02) (0.11) (0.47) (0.01) (0.06) Population estimate 5026 3706 107 1232 16,633 25 336 (1 std. error) (586) (730) (75) (299) (1170) (24) (154) At least three age classes? yes yes no yes yes no yes 2009 D.L. Crabtree and T.A. Smith 347 significantly different. Although not significantly different, age-6 was the only age class with more females than males. Figure 4. Mean densities (a), mean proportions (b), and mean proportions of each sex (c) of Northern Riffleshell in each age class at four quantitatively sampled sites (+ 1 SE). Asterisks above age class indicate significant differences at α = 0.05. 348 Northeastern Naturalist Vol. 16, No. 3 Abundance-at-age data for both sexes combined at sites RKM 74, RKM 68, RKM 23, and RKM 19 were best fit by linear functions (Eqns. 3–6, respectively), as were age 6–10 males at site RKM 74 (Eqn. 7). Female ageat- abundance data for ages 6 –10 at RKM 74 were best fit by an exponential function (Eqn. 8): where x = age in years. y = -298.8x + 3578.0 (r2 = 56, P = 0.02), [3] y = -15.5x + 119.9 (r 2 = 94, P = 0.002), [4] y = -153.12x +1284.6 (r 2 = 82, P = 0.005), [5] y = -149.1x + 1442.1 (r 2 = 82, P = 0.002), [6] y = -367.4x + 3757 (r 2 = 99), and [7] y = 2.0 * 106e-1.1092x (r 2 = 99), [8] Transformation of catch curves yielded instantaneous mortality rates of 0.30 at RKM 74, 0.46 at RKM 68, 0.51 at RKM 23, and 0.33 at RKM 19 (Fig. 5a). Slopes of catch curves at all sites were not significantly different (ANCOVA, test for homogeneity: P-values ranged from 0.355 to 0.560). Slopes of catch curves for females and males at RKM 74 using the age 6–10 data were significantly different (ANCOVA, test for homogeneity: P = 0.014; Fig. 5b). We back-calculated abundances-at-age from our linear models of mortality, assuming that the model accurately describes early-life dynamics. Back-calculation yielded an abundance of 3279 age-1 animals at RKM 74, 104 age-1 animals at RKM 68, 1293 age-1 animals at RKM 23, and 1131 age-1 animals at RKM 19. Discussion Northern Riffleshell was once widespread throughout the northern tributaries of the Ohio River, occurring from eastern Illinois in the Wabash River to Conewango Creek in Northwest Pennsylvania (US Fish and Wildlife Service 1994). Northern Riffleshell also inhabited wave-swept areas in western Lake Erie and some of the western tributaries of the Lake (US Fish and Wildlife Service 1994). Northern Riffleshell may have been in parts of the Tennessee, Cumberland, and Green rivers, although not in as great an abundance (US Fish and Wildlife Service 1994). Altogether, nine states and Ontario had records of Northern Riffleshell prior to the 1950s (US Fish and Wildlife Service 1994). Currently, it is known to exist in Tippecanoe River (Indiana), Sydenham River (Ontario), Fish Creek (Ohio and Indiana), Big Darby Creek (Ohio), and the Allegheny River basin (including French and Conewango Creeks). Of all the current locations from which Northern Riffleshell has been reported, the largest populations remain in French Creek and Allegheny River (Smith and Crabtree 2005; Zanatta and Murphy 2007; R. Villella, USGS Leetown Science Center, Kearneysville, WV, pers. comm.). Quantifying the integrity of mussel assemblages is difficult when one objective is the conservation of rare species. In general, rare species population integrity treated at the community or system scale is problematic 2009 D.L. Crabtree and T.A. Smith 349 simply because their rarity confounds potential relationships with stressors. Additionally, conservation strategies for rare species are often focused at smaller scales. For imperiled mussels, these strategies include species reintroductions and/or population augmentation often at the river-reach level and smaller (Bolden and Brown 2002, Cope and Waller 1995, Villella et al. 1998). Longitudinal distribution of mussels has been investigated as a means to describe the abundance at scales beyond the reach level (e.g., Villella and Smith 2005). Lellis in Strayer and Smith (2003) used a continuous sampling technique that allowed for seamless longitudinal distributions of species in the Delaware River. However, Lellis’ results lack population specificity at the site level (a trade-off necessitated by the length of the study reach). Figure 5. (a) Catch curves for Northern Riffleshell in each age class at four sites in French Creek, and (b) catch curves for age 6–10 females and males at RKM 74. Only older ages of females and males showed significant differences in mortality rates (P = 0.014). 350 Northeastern Naturalist Vol. 16, No. 3 Our methods captured what we believe is an accurate picture of the general longitudinal distribution of Northern Riffleshell in French Creek, but also resulted in site-specific estimates of catch and abundance. Northern Riffleshell in French Creek appears to have two population centers, one near the confluence with the Allegheny River and the other approximately 1/3 of the length upstream (beginning at river km 58). This bimodal distribution may reflect the dynamic chronology of the river system in this area. Parts of the current Allegheny River are believed to have flowed through part of the current French Creek valley and drained to the north prior to the Illinoisan glacier advance (Harrison 1980). French Creek reversed its flow during successive advances and retreats of the Illinoisan and Wisconsin glaciers, eventually flowing into the Allegheny River and becoming part of the Ohio basin (Harrison 1980). Colonization of the central part of the watershed by Northern Riffleshell may have occurred early in the formation of what we now consider to be French Creek, thus creating somewhat isolated or disjunct population centers. Conversely, Northern Riffleshell distribution may reflect legacies of stream alteration. French Creek was altered to allow water into a feeder for the Erie Canal in the area between these two population centers. Although today this stretch of stream contains other mussel species, there is no true riffle habitat and no evidence of Northern Riffleshell. The maximum mean density of Northern Riffleshell, which was recorded at RKM 94 in French Creek (6.67 m-2) ranks among the highest in it’s current range (Zanatta and Murphy 2007), and compares closely with the highest mean density from the Allegheny River at West Hickory (7.57 m-2; R. Villella, pers. comm.). The minimum density for viable populations is unknown, but Zanatta and Murphy (2007), using microsatellite DNA analysis, concluded that the effective population size for Northern Riffleshell at RKM 94 was 68 individuals. Our mean population estimate for this site was 8317 females (assuming 50:50 sex ratio); the discrepancy between these numbers (more than two orders of magnitude) may indicate a locally adapted population, bottlenecks in the past, or simply an underestimate of the effective population size by Zanatta and Murphy (2007). The sites that we considered viable were based on observations of mussel populations with uneven age structure. Abundances of Northern Riffleshell at these viable sites may yield insight into the minimum viable population size, since overall numbers of animals may be a more biologically signifi- cant measure of potential viability. Mean abundances of animals (≥ age 2) at these sites ranged from 336–16,633. Abundances that have proved viable in healthy systems like French Creek should be targeted for stocking rates as resource managers begin to implement the recovery plan for this species. These estimates are for Northern Riffleshell ages 2 and up; thus, they are applicable in situations where adult or older juvenile animals are available (e.g., relocations from salvage operations). Age structure of Northern Riffleshell in French Creek was different than that of another Epioblasma species, Epioblasma florentina walkeri (Tan 2009 D.L. Crabtree and T.A. Smith 351 Riffleshell) in Virginia (Rogers 1999, Rogers et al. 2001). Tan Riffleshell numbers-at-age exhibited what appeared to be a bimodal relationship compared to the more unimodal distribution of Northern Riffleshell at sites in French Creek. The primary determinant of the age distribution of Tan Riffleshell was most likely that population estimates were based on surface counts of marked animals only. This methodology may have skewed the data to older individuals. Although Rogers (1999) and Rogers et al. (2001) make good points regarding the problems of aging mussels, populations estimates are equally, if not more problematic and have a strong influence on age-structure analyses. Understanding natural mortality rates in mussels may be critical to effective management of these species. Relocations, reintroductions, and population augmentations will need better measures of success or failure in these relatively long-lived animals if the objectives of their recovery plans are to be followed. Although a single instantaneous mortality rate is a rather static parameter to apply over the life of an organism, the general shape of the catch curve can be useful for inferring critical life stages. In French Creek, the four sites examined all exhibited similar mortality curves, suggesting that either natural processes are similar at these sites or they are experiencing similar human-induced mortalities. A similar pattern was found for different species at three separate sites in the Cacapon River (Villella et al. 2004). Due to the relatively high water and habitat quality of French Creek, we believe that the mortality patterns at these sites are close to natural rates and thus could be of value to future restoration assessments. Interestingly, when focusing on later life (ages 6–10), we found that females died at a significantly greater rate than males. Energetic costs associated with reproductive activities, such as brooding embryos and encounters with host fish (Barnhart et al. 2008), might increase the mortality of sexual mature females compared to males. Females may also experience greater rates of predation due to their greater exposure above the substrate surface while trying to attract host fish. Mark-recapture methods have been used to estimate mortality rates in other mussel species, which allowed survivorship to be explored through time (e.g., Villella et al. 2004). Catch-curve analyses yield a snapshot of mortality, and thus we must assume that mortality is not dynamic. Villella et al. 2004 found apparent annual survivorship for three species in the Cacapon River range from 0.50 to 0.99 over the three years of study. Our estimates, from 0.49 to 0.70 (instantaneous survivorship = 1- Z), suggest that mortality in Northern Riffleshell is within the range of the species studied in the Cacapon. Our back-calculated abundances of age-1 animals are meant to provide another target for managers wishing to stock this species using young, cultured individuals. Currently, efforts are underway attempting to raise juvenile Northern Riffleshell in culture (G.T. Watters, Ohio State University, Columbus, OH, pers. comm.; C. Gatenby White Sulfur Springs Laboratory, White Sulfur Springs, WV, pers. comm.). It appears possible to culture enough 352 Northeastern Naturalist Vol. 16, No. 3 animals to stock at rates consistent with the range of variation of abundance of back-calculated age-1 animals (approximately 105–3300 individuals in 2500 m2). However, as guidance for stocking, these numbers may misrepresent the total numbers, since the age-1 animals in our study may already be in suitable micro-sites and stocking may place numerous individuals into unsuitable micro-sites. Applying a catch curve to one year’s data forces us to assume that mortality rates hold for this younger age, but it is likely that age-1 animals experience increased mortality rates due to higher rates of predation from flatworms and other small predators that they quickly outgrow. Therefore, stocking prescriptions based on these data should be best viewed as an absolute minimum range for age-1 animals. In addition, since our viable sites for Northern Riffleshell in French Creek occurred near other sites with this species, sequential sites should be considered for stocking rather than single reaches. This study is intended as a resource for managers that are interested in the reintroduction or augmentation of Northern Riffleshell. Relocation or moving adult animals from one location to another may appear to offer a strategy for restoring this species, but we suggest approaching these actions with caution as the numbers of individuals found in reproducing populations is much larger than previously attempted translocations (Bolden and Brown 2002). In addition, the relatively young maximum ages observed and relatively high natural mortality rates for Northern Riffleshell may limit the ability to “jumpstart” new populations if small numbers of older animals are stocked. Propagation of juveniles in culturing facilities may offer another alternative to acquiring stock of these animals, but again our results indicate that mortality rates may make stocking enough young animals challenging if the goal is to have these animals survive and reproduce successfully. Acknowledgments Funding was provided by the US Fish and Wildlife Service through State Wildlife Grants (SWG) Program Grant T-2, administered through the PA Game Commission and PA Fish and Boat Commission. The Nature Conservancy (TNC) and the Western Pennsylvania Conservancy (WPC) provided additional funding. Work was conducted under a Scientific Collecting Permit Number 196 Type 1 granted by the Pennsylvania Fish and Boat Commission. Todd Sampsell (WPC) secured the SWG funds and provided oversight for the first year of field work. Cornell Biological Field Station, and in particular, Tom Brookings provided use of and guidance for their low-speed saw. Thanks to Alan Wolf (TNC), Krystel Bastion (Allegheny College), Dima Haliwani (Al-Quds University), and to WPC’s Megan Bradburn (also with Student Conservation Association, [SCA]), Amy Bush (SCA), Zachary Horn (SCA), Nathan Irwin, Philip Kukulski (SCA), Lucas Mattera, Erica Maynard, Elizabeth Peck, and Curtis Stumpf for invaluable field assistance. This manuscript was considerably improved by comments from James Mckenna, the Northern Riffleshell strategy group, and two anonymous reviewers. 2009 D.L. Crabtree and T.A. Smith 353 Literature Cited Barnhart, C.M, W.R. Haag, and W.N. Roston. 2008. Adaptations to host infection and larval parasitism in Unionoida. Journal of the North American Benthological Society 27(2):370–394. Bogan, A.E. 1993. Freshwater bivalve extinctions (Mollusca: Unionoida): A search for causes. American Zoologist 33:599–609. Bolden, S.R., and K.M. Brown. 2002. Role of stream, habitat, and density in predicting translocation success in the threatened Louisiana Pearlshell, Margaritifera hembeli (Conrad). Journal of the North American Benthological Society 21(1):89–96 Casselman, J.M. 1983. Age and growth of fish from their calcified structures: Techniques and tools. NOAA Technical Report NMFS No. 8. Miami, fl. Cope, W.G., and D.L. Waller. 1995. Evaluation of freshwater mussel relocation as a conservation and management strategy. Regulated Rivers: Research and Management 11:147–155. Downing, W.L., J. Shostell, and J.A. Downing. 1992. Non-annual external annuli in the freshwater mussels Anodonta grandis grandis and Lampsilis radiata siliquoidea. Freshwater Biology 28:309–317. Gulland, J.A. 1983. Fish Stock Assessment: A Manual of Basic Methods. Wiley, New York, NY. 223 pp. Haag, W.R., and A.M. Commens-Darson. 2008. Testing the assumption of annual shell-ring deposition in freshwater mussels. Canadian Journal of Fisheries and Aquatic Science 65:493–508. Harrison, S. 1980. New survey provides key to Crawford County drainage puzzle. Pennsylvania Geology 11(2):10–12. Kuehnl, K.F., and G.T. Watters. 2005. Draft augmentation and reintroduction plan for the Clubshell (Pleurobema clava) and Northern Riffleshell (Epioblasma torulosa rangiana) in Ohio and Illinois. Prepared for Region 3 US Fish and Wildlife Srevice, Fort Snelling, MN. Neves, R.J., and S.N. Moyer. 1988. Evaluation of techniques for age determination of freshwater mussels (Unioinidae). American Malacological Bulletin 6:179–188. Ortmann, A.E. 1919. A monograph of the naiads of Pennsylvania. Part III. Systematic account of the genera and species. Reprinted from the Memoirs of the Carnegie Museum Volume VIII, No. 1. Parmalee, P.W., and A.E. Bogan. 1998. The Freshwater Mussels of Tennessee. University of Tennessee Press, Knoxville, TN. 328 pp. Rogers, S.O. 1999. Population biology of the Tan Riffleshell (Epioblasma florentina walkeri) and the effects of substratum and light on juvenile mussel propagation. M.Sc. Thesis. Virginia Polytechnic Institute and State University, Blacksburg, VA. Rogers, S.O., B.T. Watson, and R.J. Neves. 2001. Life history and population biology of the endangered Tan Riffleshell (Epioblasma florentina walkeri) (Bivalvia: Unionidae). Journal of the North American Benthological Society 20(4):582–594. Rankin, E.T., and B.J. Armitage. 2005. Protection, Restoration, and Aquatic Life Potential in Nature Conservancy Areas in the Agricultural Midwest: French Creek (New York), St. Joseph River and Fish Creek (Indiana, Michigan, and Ohio), and the Mackinaw River (Illinois). A final report prepared for The Nature Conservancy. Available online at www.ConserveOnline. Accessed March 2005. 354 Northeastern Naturalist Vol. 16, No. 3 Smith, T.A., and D.L. Crabtree. 2005. Freshwater Mussel (Unionidae) and Fish Assemblage Habitat Use and Spatial Distributions in the French Creek Watershed: Reference for Western Pennsylvania Unionid Protection and Restoration. A final report to PA Fish and Boat Commission, Bellefonte, PA. Smith, T.A., and D.L. Crabtree. In press. Freshwater mussel (Unionidae: Bivalvia) distributions and densities in French Creek, Pennsylvannia. Northeastern Naturalist. Smith, D.R., R.F. Villella, and D.P. Lemarie. 2001. Survey protocol for assessment of endangered freshwater mussels in the Allegheny River, Pennsylvania. Journal of the North American Benthological Society 20(1):118–132. Smith, T.A., T. Sampsell, and E. Straffin. 2003. 1st Annual state of the stream report on the health of French Creek. A publication of the Western Pennsylvania Conservancy. XI +109 pp. Stein, B.A., L.S. Kutner, J.S. Adams (Eds.) 2000. Our Precious Natural Heritage: The Status of Biodiversity in the United States. Oxford University Press, New York, NY. Strayer, D.L., and D.R. Smith. 2003. A Guide to Sampling Freshwater Mussel Populations. American Fisheries Society, Monograph 8, Bethesda, MD. Tokey, T., N. Yochum, J. Hoenig, J. Lucy, and J. Cimino. 2007. Evaluating localized vs. large-scale management: The example of Tautog in Virginia. Fisheries 32(1):21–28. US Fish and Wildlife Service. 1994. Clubshell (Pleurobema clava) and Northern Riffleshell (Epioblasma torulosa rangina) recovery plan. US Fish and Wildlife Service, Hadley, MA. 68 pp. USGS Mussel Estimation Program. 2004. Available online at http://www.lsc.usgs. gov/aeb/2068/. Accessed March 2005. Veinott, G.I., and R.J. Cornett. 1996. Identification of annually produced opaque bands in the shell of the freshwater mussel Elliptio complanata using the seasonal cycle of δ18O. Canadian Journal of Fisheries and Aquatic Science 53:372–379. Villella, R.F., and D.R. Smith. 2005. Two-phase sampling to estimate river-wide populations of freshwater mussels. Journal of the North American Benthological Society 24(2):357–368. Villella, R.F., T.L. King, and C.E. Starliper. 1998. Ecological and evolutionary concerns in freshwater bivalve relocation programs. Journal of Shellfish Research 17(5):1407–1413. Villella, R.F., D.R. Smith, and D.P. Lemarie. 2004. Estimating survival and recruitment in a freshwater mussel population using mark-recapture techniques. American Midland Naturalist 151:114–133. Williams, J.D., M.L. Warren, K.S. Cummings, J.L. Harris, and R.J. Neves. 1993. Conservation status of freshwater mussels of the United States and Canada. Fisheries 18(9):6–22. Zanatta, D.T., and R.W. Murphy. 2007. Range-wide population genetic analysis of the endangered Northern Riffleshell mussel, Epioblasma torulosa rangiana (Bivalvia: Unionoida). Conservation Genetics 8:1393–1404.