Appendix L-15

Mussel Survey Protocols

These protocols are currently being prepared and will be included in this MSHCP

when available from the Service. These protocols will be based, in part, on the specifications

provided in Smith 2006, Survey design for detecting rare freshwater mussels (attached).

i

J. N. Am. Benthol. Soc., 2006, 25(3):701-7110 2006 by The North American Benthological Society

Survey design for detecting rare freshwater mussels

David R. Smith'Aquatic Ecology Laboratory, Leetown Science Center, US Geological Survey, 11649 Leetown Road,

Kearneysville, West Virginia 25430 USA

Abstract. A common objective when surveying freshwater mussels is to detect the presence of rarepopulations. In certain situations, such as when endangered or threatened species are potentially in the areaof a proposed impact, the survey should be designed to ensure a high probability of detecting speciespresence. Linking survey design to probability of detecting species presence has been done for quantitativesurveys, but commonly applied designs that are based on timed searches have not made that connection. Ipropose a semiquantitative survey design that links search area and search efficiency to probability ofdetecting species presence. The survey can be designed to protect against failing to detect populationsabove a threshold abundance (or density). I illustrate the design for surveys to detect clubshell (Pluerobemaclava) and northern riffleshell (Epioblasma torulosa rangiana) in the Allegheny River. Monte Carlo simulationindicated that the proposed survey design performs well under a range of spatial distributions and lowdensities (<0.05 m2) where search area is sufficient to ensure that the probability of detecting speciespresence is predicted to be >0.85.

Key words: unionid, probability of species detection, detectability, qualitative sampling, rarepopulations, species presence, timed search, occupancy.

A common objective of surveys of freshwatermussels is to detect the presence of rare populations,e.g., when assessing site-specific impacts on endan-gered or threatened species (Wilcox et al. 1993, Smithet al. 2001a) or when delineating the range of a rarespecies (Strayer et al. 1996). An important applicationof this objective is determining the presence of anendangered or threatened species in an area of aproposed impact. In that case, confirmation of speciespresence would halt or influence the activity thatwould cause the impact, whereas failure to detect aspecies when it was in fact present (analogous to aType II error) could permit an adverse impact to occur.Thus, a survey designed to achieve this objectiveshould ensure a high probability of detecting speciespresence.

Intuition tells us that the probability of detectingspecies presence is related to species abundance andspatial distribution, sampling effort, search efficiencywithin the area sampled (i.e., detectability), and thedistribution of sampling effort within a study site.McArdle (1990) and Green and Young (1993) relateddetection of rare species to the number of samplingunits taken in a quantitative, quadrat-based surveyassuming perfect search efficiency. Near-perfect search

1 E-mail address: drsmith@usgs.gov

efficiency would be achieved in a freshwater musselsurvey by sediment excavation (Hornbach and Deneka1996, Smith et al. 2001b). Green and Young (1993)provided guidelines for designing a quantitativesurvey that would ensure a high probability ofdetecting rare species. However, their guidelines havenot been widely adopted for freshwater musselsurveys, in part because quantitative sampling isperceived as time-consuming and expensive (Ober-meyer 1998), and timed-search surveys result in morespecies detections per unit time than quadrat-basedsurveys (Strayer et al. 1997, Vaughn et al. 1997,Obermeyer 1998).

Timed searches are qualitatively more efficient thanquadrat-based surveys, but an explicit method torelate search time to the probability of detectingspecies presence does not appear to exist. Strayer etal. (1997) calculated probability of detection for timedsearches for Elliptio complanata, but cited high varianceof catch per unit effort statistics as a limitation on thegenerality of a timed-search-based detection curve.Metcalfe-Smith et al. (2000) found that >50% of speciespresent are missed when typical search times are usedand that increased search time resulted in more speciesdetections. However, the essential question of howmuch search time is enough to ensure a high

701

[

702 D. R. Sronrrx

probability of detecting a rare species remains unan-swered.

I propose an alternative survey design that isintermediate between a timed search and Green andYoung's (1993) quadrat-based sampling. The designrelates probability of detecting species presence tosearch area and search efficiency. The semiquantitativeapproach does not require sediment excavation, butdoes require a priori information on search efficiency.Search effort is constrained to defined areas (i.e.,sampling units), so the survey design can be linked toprobability of detecting species presence. I describe anexample survey designed to detect clubshell (Pluer-obema clava) and northern riffleshell (Epioblasma tor-ulosa rangiana) in the Allegheny River. Last, I evaluatethe design using a Monte Carlo simulation thatincludes spatially clustered populations because thesurvey design relies on assumptions about spatialdistribution of rare populations.

Survey Design

I developed the survey design by specifying thesurvey objective and applying a model to link theobjective to elements of the design. In particular, Iconsidered factors that affect search efficiency (e.g.,detectability) because it is an important element inmussel survey design. I also considered relevantstatistical principles that could guide how best todistribute the area to be searched within a site.

Survey objective

Clear, specific, and quantifiable objectives are centralto successful survey design (Strayer and Smith 2003,McDonald 2004). The primary objective of our surveywas to detect the presence of a rare population, but asurvey objective should be defined further and statedquantitatively to allow for evaluating whether aproposed design will meet the objective. For example,the objective might be stated quantitatively: "To detectthe presence of any of the endangered or candidatespecies in a site with probability >0.85 given thatspecies abundance is >100 individuals." This state-ment has 2 important elements: 1) the minimumthreshold for the probability of detecting presence ofa species, and 2) a species abundance or density that isdeemed biologically meaningful. I used an abundanceof 100 individuals only as an example. The determi-nation of a biologically meaningful threshold shouldinvolve multiple considerations including legal man-dates, life history, population viability, and compar-isons of densities throughout a local watershed, region,or range.

[Volume 25

Modeling the sampling process

A model of the sampling process is needed to relatethe proposed objective to the survey design. Themodel represents the expected survey results (countsof mussels) as a function of the controlling factors-mussel abundance, search area, and search efficiency.Search efficiency, which is also termed detectability, isthe probability of detecting an individual mussel giventhat it is within the search area.

The expected number of individuals counted in asurvey of a site can be represented as

E(C) = a(3T [1J

where C is the count of individuals, E(C) is the

expected count based on a repeatable sampling

process, a is the fraction of the site that is searched,

(3 is the probability of detecting an individual given

that it is in the search area , and T is the total numberof individuals in the site (Williams et at. 2002:244).

The expected number of individuals in the search area

is a T =aµ where a is the search area and µ is species

density. Note that the fraction of the site that is

searched is a = a /A where A denotes the area of the

site . The search area is the sum of the areas of eachnunit in the sample, i.e., a = Tai where n is the

sample size and a; is the area of the i`h sampling unit(typically a; is the same or nearly the same for allsampling units).

Search efficiency, which refers to the probability ofdetecting an individual given that it is in the searcharea, is a function of search rate ( time per unit area)and search area (Fig. 1). In eq. 1, search efficiency isdenoted by p. In theory, if one spends enough time andeffort searching an area , all individuals that are presentwithin the search area will be detected, in which case R= 1. However, in actual sampling situations, searchtime and effort are restricted so that not all individualsin the sample area are detected and (3 < 1.

Mussel sampling techniques have been classified asquantitative, semiquantitative, or qualitative (Strayerand Smith 2003). This classification can be related tothe parameters in eq. 1 (Table 1). Quantitative andsemiquantitative sampling are distinguished fromqualitative sampling by a. a is known when samplingis quantitative or semiquantitative, but a is not knownwhen sampling is qualitative. Quantitative and semi-quantitative sampling are distinguished by Q Quanti-tative sampling is the case where (3 =1 or p < 1 and isestimated. In either case, p can be accounted for in eq.1. Semiquantitative sampling is the case where R isunknown. Unbiased estimation of abundance ordensity is possible only when a and (3 are known or

I

2006] DETECTING RARE FRESHWATER MUSSELS

and lotic habitats. The Downing and Downing (1991)formula indicates that the variance-to-mean ratioapproaches 1 (spatial randomness) as the meanapproaches 0.10, the threshold for rarity used by Greenand Young (1993). I used data from Smith et at. (2001b)and found variance-to-mean ratios for 60 species/sitecombinations (31 species at 14 sites) that indicatedmussel distributions were statistically spatially randomfor p < 0.10/m2. The same relationship betweendensity and spatial distribution has been found inother populations (McArdle 1990, Welsh et al. 1996).Therefore, I propose eq. 3 as a useful approximation forguiding survey design, and I evaluate its use in asimulation that includes spatially clustered populationsand sampling units other than quadrats (see below).

Equation 3 can be revised to account for searchefficiency by including the parameter a therebymaking a connection to the sampling-process modelin eq. 1. The expected number of individuals detected

TABLE 1. Contrast of sampling techniques (classified as qualitative, semiquantitative, or quantitative) based on fraction of sitesearched (a), search efficiency or detectability (R), and which parameter(s) are known or estimated. C = count of mussels in a sampleat a site, T = total number of individuals in the population at a site, T= estimated total number of individuals at a site, Q = estimatedprobability of detecting an individual given that it is within the search area.

Sampling technique at II Survey result

Qualitative Unknown Unknown Incomplete countSemiquantitative Known Unknown Incomplete count within searched area

Quantitative Known Known or estimated Abundance estimate: T = Cl (a(3)

estimated, i.e., t = C/(a]i) where t is the estimatedtotal number of individuals in the study site (i.e., theabundance estimate) and 03 is an unbiased estimate ofthe probability of detecting an individual given that itis in the search area.

Detecting the presence of a rare species within a siteis equivalent to detecting at least one individual of thatspecies, and it follows from eq. 1 that this event is afunction of a, p, and T. That is:

Prob(detecting at least one individual)

= Prob (C > 0) = f(a(3T). [2]

Green and Young (1993) considered sampling rarepopulations of freshwater mussels in quadrats andderived a formula for the probability of detecting thepresence of a low-density population (i.e., µ < 0.10/mz)using a Poisson probability distribution:

Prob(detecting at least one individual) = 1 - e-'n"

(3]

where m is the number of individuals within asampling unit and n is the number of random samplingunits searched. The Poisson assumption implies thatmussels at very low density have a spatially randomdistribution. This assumption does not imply anabsence of underlying ecological relationships, suchas habitat associations and dispersal mechanisms,which affect distribution (Downing and Downing1991). Rather, it indicates that when mussels aregeographically rare at a site (i.e., µ < 0.1/m 2), theirlow density masks underlying ecological relationshipsand their spatial distribution is random from astatistical perspective. Green and Young (1993) pre-sented empirical data to support this contention. Inaddition, Smith et al. (2003) found that low-densitymussels on the Cacapon River, West Virginia, hadrandom distributions as evidenced by variance-to-mean ratios. A variance-to-mean ratio of 1 indicates aPoisson distribution (Elliott 1977). Downing andDowning (1991) presented a formula for variance as afunction of the mean number of individuals collectedthat was developed empirically from surveys in lentic

is Rnm = (3aT. Thus:

Prob(detecting at least one individual)

=1-e-PaT =1-e-f`T/A=1-a-p"p.

703

[4]

Equation 4 can be used to examine the effect ofsearch efficiency (0), search area (a), and density (µ) onthe probability of detecting at least one individual or,analogously, the probability of detecting speciespresence. Figure 2 shows the probability of detectingspecies presence for t = 0.01, 0.05, and 0.10/ms, p =0.2, 0.4, 0.6, and 0.8, and a =100 to 1000 m2. Equation 4also could be used to examine the effect of abundance(T) for a given study site area (A) instead of p. Table 2shows probability of detecting species presence for T100 to 500 and A = 16,000 and 32,000 m2.

Factors that affect search efficiency

Search efficiency is a function of search area andsearch time ( Fig. 1 ). The exact form of that relationshipis not known and will vary over time and area. For agiven search area , the more time spent searching, thehigher the search efficiency. It is likely that searchefficiency will increase quickly as search time is

i

704 D. R. SknTH [Volume 25

L

Search area

FIG. 1. Search efficiency (p; legend) as a function of searchtime and search area (a). The axes are not labeled because theexact form of the relationship is determined by a variety offactors involving mussel biology, physical environment, andobserver capabilities.

increased from low to moderate levels and the rate ofincrease in search efficiency will slow as it approachescomplete detection, exhibiting a point-of-diminishing-returns-type phenomenon. These relationships be-tween search time and search efficiency also havebeen shown empirically (Metcalfe-Smith et al. 2000).

The exact form of the relationship between searchefficiency and search time will depend on a number offactors (Strayer et al. 1997), some of which are inherentto the biology and natural history of the musselspecies. For example, some species are more crypticthan others by virtue of their size, coloration, orreproductive behavior (Miller and Payne 1993, Ober-meyer 1998, Haag and Warren 2000). Mussels exhibitseasonal patterns in vertical migration associated withday length and water temperatures (Amyot andDowning 1991, Wafters et al. 2001, Perles et al. 2003).Other biological factors include gender and demo-graphics. For example, female northern riffleshell(Epioblasma torulosa rangiana) are more visible thanmales (Smith et al. 2001a), and small mussels aredifficult to detect (Miller and Payne 1988, Hornbachand Deneka 1996, Richardson and Yokley 1996, Smithet al. 2001b). Other factors, such as turbidity, hydro-logic variability, substrate, and vegetative cover, areassociated with the physical environment (Di Maioand Corkum 1997, Smith et al. 2001b). Last, somefactors, such as observer experience, visual acuity, andfatigue, are associated with the observer (Strayer et al.1997).

Only those mussels that are epibenthic or not buried

can be found in a search restricted to the substratesurface (Amyot and Downing 1991). If an area issearched thoroughly so that all mussels on thesubstrate surface have been found, then searchefficiency will be capped at the proportion of musselsthat are on the surface. Beyond that level of effort,excavation would be required to increase searchefficiency to the point that all or nearly all musselswithin the searched area are found (Smith et al. 2001b).

Impact of search efficiency on survey design

Because search efficiency directly affects the prob-ability of detecting species presence, it should beconsidered when designing a survey. Two approachescould be used to incorporate search efficiency insurvey design. First, one could be conservative andassume that search efficiency (0) was low. Then therelationship from eq. 4 (Table 2, Fig. 2A-D) could beused as a guide to find the search area (a) that wouldensure that the probability of detecting speciespresence is sufficiently high (Fig. 2A-D). For example,if 0 were assumed to be <0.2, then a would have tobe >1000 m2 to have a probability of detecting atleast one individual = 0.85 for p = 0.01 /m2 (Fig. 2A).This a would be equivalent to ten 1-m-wide X 100-m-long transects (distribution of search effort through-out the site is discussed below). The assumed 0 couldbe based on life-history traits, such as likelihood thatan individual would be endobenthic (Amyot andDowning 1991). This approach would be precau-tionary.

Second, (3 could be estimated at another time andplace where the rare species was numerous or by apilot survey based on a related, but more common,species. For example, 0 could be estimated bysearching the surface of quadrats before excavatingsediment (cf. Haukioja and Hakala 1974, Smith et al.2001b). In this case, the estimate of (3 and eq. 4 could beused to predict the a that would result in the desiredprobability of detecting species presence. For example,if R for a search rate of 2 min/m were estimated as 0.4,then a = 500 m2 would ensure a probability ofdetecting species presence = 0.85 for µ = 0.01 /m2(Fig. 2B), and 1000 min (16.67 h) of search time wouldbe required. The shortcoming of using an estimatefrom another time and place is that 0 would beestimated under one set of conditions and appliedunder a similar, but not identical, set of conditions. Ifan overestimate of p were used in survey design, thenthe probability of detecting species presence alsowould be overestimated, and the design would notbe precautionary. The number of quadrats needed toestimate R would depend on p at the site and the

i

2006] DETECTING RARE FRESHWATER MUSSELS 705

TABLE 2. Probability of detecting species presence given the study site area (A), search efficiency (p), abundance (T), and searcharea (a). Bold font indicates probability of species detection >0.85.

a (m2)

A (m2) P T 100 200 300 400 500 600 700 800 900

16,000 0 . 2 100 0 .12 0.22 0.31 0.39 0.46 0.53 0.58 0.63 0.68200 0.22 0.39 0.53 0.63 0.71 0. 78 0.83 0 .86 0.89300 0.31 0.53 0.68 0.78 0.85 0.89 0 .93 0.95 0.97400 0.39 0.63 0.78 0.86 0 .92 0.95 0 .97 0.98 0.99500 0 .46 0.71 0.85 0.92 0 . 96 0.98 0 .99 0.99 1.00

0.4 100 0.22 0.39 0.53 0.63 0.71 0.78 0. 83 0.86 0.89200 0.39 0.63 0.78 0.86 0.92 0.95 0.97 0.98 0.99300 0 .53 0.78 0.89 0.95 0.98 0.99 0.99 1 .00 1.00400 0 .63 0.86 0.95 0.98 0.99 1.00 1.00 1 .00 1.00500 0 . 71 0.92 0.98 0.99 1 .00 1.00 1 .00 1.00 1.00

0.6 100 0.31 0.53 0.68 0.78 0.85 0.89 0.93 0.95 0.97200 0.53 0.78 0.89 0.95 0.98 0.99 0.99 1.00 1.00300 0. 68 0.89 0.97 0.99 1 .00 1.00 1 .00 1.00 1.00400 0.78 0.95 0.99 1.00 1 .00 1.00 1 .00 1.00 1.00500 0.85 0.98 1.00 1.00 1 .00 1.00 1 .00 1.00 1.00

32,000 0.2 100 0.06 0.12 0.17 0.22 0.27 0.31 0.35 0.39 0.43200 0.12 0.22 0.31 0.39 0.46 0.53 0.58 0 .63 0.68300 0.17 0.31 0.43 0.53 0.61 0.68 0.73 0.78 0.82400 0.22 0.39 0.53 0.63 0.71 0.78 0.83 0.86 0.89500 0.27 0.46 0.61 0.71 0.79 0.85 0 .89 0.92 0.94

0.4 100 0.12 0.22 0.31 0.39 0.46 0.53 0.58 0 .63 0.68200 0.22 0.39 0.53 0.63 0.71 0.78 0. 83 0.86 0.89300 0.31 0.53 0.68 0.78 0.85 0.89 0 .93 0.95 0.97400 0.39 0.63 0.78 0.86 0 .92 0.95 0 .97 0.98 0.99500 0.46 0.71 0.85 0.92 0 .96 0.98 0 .99 0.99 1.00

0.6 100 0.17 0.31 0.43 0.53 0.61 0.68 0.73 0.78 0.82200 0.31 0.53 0.68 0.78 0 .85 0.89 0 .93 0.95 0.97300 0.43 0.68 0.82 0.89 0 .94 0.97 0 .98 0.99 0.99400 0.53 0.78 0.89 0.95 0 .98 0.99 0.99 1.00 1.00500 0 .61 0.85 0.94 0.98 0 .99 1.00 1 .00 1.00 1.00

proportion of individuals on the substrate surface(Smith et al. 2001b, Strayer and Smith 2003). Therefore,the environmental conditions in the pilot surveyshould be as close as possible to the conditions likelyto be encountered at the site where species presencewill be determined. Information on species-specificdensities and search efficiencies are available in theliterature in some cases (e.g., Smith et al. 2001a), andunpublished agency surveys are likely to providerelevant data.

Statistical principles guiding the distribution of search effortwithin the site

Two statistical principles, in particular, are useful forguiding distribution of search effort. First, spatiallybalanced sampling has been recognized as efficient forsampling natural resources (Christman 2000, Stevensand Olsen 2004). A spatially balanced sample is onethat is distributed throughout a site or population.Various systematic or grid sampling methods qualify

as spatially balanced. Second, it is generally moreefficient (reduces sampling error) to distribute effortamong many small units than a few large units. Thisprinciple is particularly relevant when the populationis spatially clustered (Elliott 1977). The mitigatingfactor is the effort required to move among units.Many small units require more between-unit travelthan few large units. Thus, the challenge is to find asampling-unit size that represents a compromisebetween cost and sampling error. These principlescan be combined with stratification to allocate effortefficiently and to ensure that sampling is done in allhabitats. For example, a site can be stratified bymacrohabitat (e.g., riffle, run, pool) and search area canbe allocated proportionately or according to antici-pated habitat value (i.e., more effort in better habitat).On the other hand, the survey could be conducted inphases as suggested by Kovalak et al. (1986) andimplemented recently by Villella and Smith (2005).During the 1St phase, an informal search or surveil-lance can be conducted to delineate mussel beds or

706 D. R. SMITH

1.0

0.5

0.0

1.0

0.5

0.0

A

C

200 500 800

-o-- 0.01 ind./m2- 6 0.05 ind./m2---o- 0.10 ind.1m2

200 500 800

[Volume 25

Search area (m2)

FIG. 2. Probability of detecting species presence as a function of search area (a) and density (0.01, 0.05, 0.10 individual/m2) ofmussels when search efficiency (1) was 0.2 (A), 0.4 (B), 0.6 (C), and 0.8 (D).

habitat. During the 2nd phase, the semiquantitativeapproach can be applied after the search area (a) hasbeen determined to ensure a sufficiently high proba-bility of detecting species presence. The predetermineda should be allocated so that most, but not all, of thearea occurs within the bed or habitat identified duringthe 16' phase.

A cautionary note is warranted regarding thedistribution of sampling effort according to an explicitor implied habitat model. If the habitat model is agood approximation, then it can be helpful in distrib-uting search area. Depth and hydrological variabilityare useful predictors of mussel density (Haukioja andHakala 1974, Strayer and Ralley 1993, Di Maio andCorkum 1995). However, if the model is a poorapproximation, as Strayer and Ralley (1993) foundfor microhabitat variables, then model-based distribu-tion can be inefficient at best and misleading at worst.A poor habitat model could lead to omission of theactual habitat from the area searched.

Detection of Clubshell (Pleurobema clava) andNorthern Riffleshell (Epioblasma torulosa rangiana)

in the Allegheny River

I used data from the clubshell and northern riffle-shell in the Allegheny River to illustrate the design of a

survey to detect their presence. In previous surveys onthe Allegheny River, Smith et al. (2001a) reported thata thorough search of the substrate surface required 2min/m2 of search time. At the West Hickory bridgesite, -30% and 50% of clubshell and northern riffle-shell were found at the substrate surface, respectively.

Suppose the goal was to protect a site against

adverse impact if either species was present at t >

0.01 /m2 with a probability of detecting species

presence _> 0.85. (Tolerance for risk is a subjective

decision that often would be set during the regulatory

process.) To protect either species, the 0 corresponding

to the least detectable species, the clubshell, would be

used. In this case, we assume that the substrate surface

within a will be searched thoroughly so that (3 is the

proportion of mussels on the substrate surface. Given

this information, we can design a survey using eq. 4:

0.85 = 1 - e-0.30ao.01

and solve for a:

a In (I - 0.85)-0.003

= 632m2.

Based on the principle of spatially balanced sam-pling, at least 632 m2 of search area should be

i

2006] DETECTING RARE FRESHWATER MUSSELS 707

distributed throughout the site. A reasonable design

would be to search within transects oriented perpen-

dicular to shoreline or the thalweg. Following the rule

that more small units are better, use of 0.5-m-wide

transects would allow greater spatial dispersion of

sampling effort; however, logistics and tradition might

favor 1-m-wide transects, especially at sites where

SCUBA is required. Transect length would depend on

site dimensions. For example, if the site was 100 m

across the river, then seven 1-m-wide transects would

be required. Good spatial balance and coverage would

be achieved by selecting a random start and placing

transects at equal intervals. An improvement on that

plan would include 2 random starts. To increase

probability of detecting species presence to 0.95, ten

1 X 100 m transects would be required.

After a has been determined based on R and a p. thatis to be protected, the time required to conduct thesurvey can be calculated. Based on 2 min/m2 to searchthe surface substrate thoroughly, searching seven I X100-m transects would require -23 h, which could bedivided among multiple observers. The survey couldbe accomplished in -1 d with a crew of 4. This timeand effort does not seem to be an unreasonable surveycost when the objective is to detect a rare orendangered species before an adverse impact occurs.Budgets for construction projects, for example, canamount to hundreds of thousands to millions ofdollars. The cost to conduct a rigorous mussel surveyis trivial by comparison.

Monte Carlo Simulation

To evaluate the proposed survey design, a computerprogram was used to generate locations for individualmussels within a site of 16,000 m2 (100 M X 160 m),apply search efficiencies so that different proportionsof the mussels were detectable, and count detectablemussels within systematically placed 1-m transects.Abundance at the site was a Poisson random variablewith means of 100, 300, and 500 mussels representingpopulation densities of 0.006, 0.02, and 0.03 (individ-uals/m2). Individual mussels were in clusters withmean sizes of 1, 3, or 5 individuals (a cluster size of 1represented complete spatial randomness). The loca-tion of the cluster center was random within the site,and individuals were distributed from the clustercenter at a uniform random angle and exponentialrandom distance, with mean distance of 1 m. Searchefficiencies of 0.2, 0.4, or 0.6 were applied to determinewhether each individual in the population wasdetectable. Detectable individuals were counted with-in 1-m transects oriented across the short axis of thesite (100 m). Areas searched were 400, 600, 800, and

TABLE 3. Abundance (T), search efficiency (p), and clustersize for the populations used to simulate the proposedsurvey design. The study site was 16,000 m2 (160 m X 100m). Variance-to-mean ratios were calculated for individualswithin I m X 100 m transects.

Variance-to-mean ratio

T 13Cluster

sizeEntire

populationDetectable portionof the population

100 0.2 1 1.19 0.913 2.03 1.115 2.33 1.04

0.4 1 1.21 1.173 1.93 1.405 2.17 1.29

0.6 1 1.05 1.043 1.91 1.585 2.84 1.77

300 0.2 1 0.87 0.853 2.44 1.085 2.53 1.20

0.4 1 0.83 1.033 1.85 1.395 2.11 1.36

0.6 1 0.94 1.083 1.77 1.315 2.62 1.92

500 0.2 1 1.13 1.153 1.84 1.195 2.91 1.44

0.4 1 1.03 1.043 1.61 1.115 2.19 1.39

0.6 1 1.03 0.803 2.01 1.455 2.81 2.04

1000 m2. The probability of detecting species presencewas calculated as the proportion of 1000 replicationswhere at least one individual was counted. Computa-tions were done in SAS (version 9.1 SAS Institute,Cary, North Carolina).

The populations showed differing degrees of spatialclustering (Table 3, Fig. 3). Variance-to-mean ratiosincreased with cluster size were lower when calculatedusing detectable individuals only. Thus, the detectableportion of the population appears less spatiallyclustered than the actual population.

Simulated probabilities of detecting species presencegenerally tracked the probabilities predicted from eq. 4(Table 4). Variability in the simulated probabilities wascaused by variability in abundance, search efficiency,cluster size, and sample selection. This result isrelevant because abundance, search efficiency, andspatial distribution would not be known exactly whenusing eq. 4 for survey design. The simulationsindicated that eq. 4 is a useful guide under a rangeof conditions. Most important, the survey design

i

708 D. R. SMITH

A

80 -

60 -

40 -

20 -

0

60

40

20

06

00 50 100 150

Position along shore (m)

FIG. 3. Example spatial distributions of detectable mus-sels used to evaluate the survey design when simulatedabundance was 100, search efficiency was 0.2, and clustersizes were 1 (A) and 5 (B). Detectable mussels were a randomsubset of the abundance determined by the search efficiency.There were 23 detectable mussels in A and 29 in B.

performed well when a was predicted to result in ahigh probability of detecting species presence. Simu-lated probabilities of detecting species presence were>0.85 in 92% (77 of 84) of cases where eq. 4 predictedthe probabilities would be >0.85 (Table 4).

Discussion

Clear, specific, and quantitative objectives are pre-requisites to a successful survey design (McDonald2004). For example, the objective for a pre-dredgingsurvey could be to detect the presence of anyendangered or candidate species with probability>0.85 given that species density is >0.01/m2. Animportant question to ask when designing a survey iswhether the proposed design will meet the statedobjective (Strayer and Smith 2003). The survey designdescribed here provides a method for answering thatquestion by linking survey elements, i.e., search areaand search efficiency, to the probability of detecting

species presence.The proposed survey design, which is intermediate

between timed search and quadrat methods, requiresthat the search area be constrained within sampling

[Volume 25

units, but excavation is not required because searchefficiency is assumed to be less than perfect. Distribu-tion of the search area within the site is flexible withinguidelines. Based on well-established principles ofsampling natural resources, it is best to distributesampling effort throughout a study site in relativelysmall sampling units. The size of the sampling units ismitigated by logistic considerations with transectsrecommended in some cases because of ease of fieldapplication. A Monte Carlo simulation confirmed thatuse of systematically placed transects is a goodapproach for the objective of species detection. How-ever, use of transects would not be a good approachwhen the objective is to estimate abundance or densitybecause some amount of excavation would be requiredand, therefore, quadrats would be required (Smith etal. 2001b, Strayer and Smith 2003). Information onhabitat or mussel beds can be used to stratify the siteand to allocate the search area within strata eitherproportionately or with more of the search effortallocated to better habitats. More complex sample-selection procedures, such as unequal probabilitysampling, could be applied. However, ease of appli-cation should be an overarching concern, and simpleselection procedures, such as systematic sampling,would be preferable.

Some population abundances or densities areunlikely to be detected without substantial samplingeffort by increasing search efficiency or search area(Table 2). This constraint is unavoidable in anyprotocol. The proposed survey design incorporatessampling techniques (i.e., transect-based, semiquanti-tative sampling) that are part of many existingprotocols. However, the user of the proposed designcan be fully aware of population sizes that are likely tobe detected by explicitly stating the probability ofdetecting species presence for given population sizeand sampling effort. As one reviewer noted, a mainadvantage of the proposed design is that the user hasan answer to the question: "How much sampling effortis enough?"

A reasonable concern with the proposed design isthe cost to survey a site. The recommended samplingeffort is likely to exceed the costs associated withcurrently applied protocols. Few protocols for rarespecies detection have been published; however,Young et al. (2001) recommended at least 2 person-hours of search time in optimal habitat beforeconcluding that a species was absent if no individualswere detected. At a search rate of 2 min/m2, a 2-hsearch would be equivalent to <100 m2 of search area,which appears to be an insufficient effort for detectingrare species. A search area of 100 m2 resulted in aprobability of detecting species presence as low as 0.12

20061 DETECTING RARE FRESHWATER MUSSELS 709

TABLE 4. Probabilities of detecting species presence observed from a computer simulation and predicted by eq. 4. Abundance (T),search efficiency (p), and cluster size are mean values used in the simulation, but were random variables in the simulation. Clusterlocations were random within a 16,000-m2 study site. Searches were conducted within 1 m x 100 m transects. The search area (a)was the sum of the transect areas. Bold font indicates combinations with predicted probabilities >0.85.

a (m2)

400 600 800 1000Cl ster

T 13

usize Simulated Predicted Simulated Predicted Simulated Predicted Simulated Predicted

100 0.2 1 0.45 0.39 0.78 0.53 0.55 0.63 0.51 0.713 0.36 0.39 0.73 0.53 0.61 0.63 0.89 0.715 0.43 0.39 0.57 0.53 0.70 0.63 0.75 0.71

0.4 1 0.76 0.63 0.95 0.78 0.82 0.86 0.91 0.923 0.63 0.63 0.89 0.78 0.79 0.86 1 .00 0.925 0.49 0.63 0.65 0.78 0.82 0.86 0 .92 0.92

0.6 1 0.89 0.78 0.97 0.89 1 . 00 0.95 0 .95 0.983 0.67 0.78 0.87 0.89 0 . 96 0.95 0.94 0.985 0.72 0.78 0.86 0.89 0 . 95 0.95 0 .73 0.98

300 0.2 1 0.75 0.78 0.97 0.89 0 . 95 0.95 1 .00 0.983 0.73 0.78 0.78 0.89 1.00 0.95 1 .00 0.985 0.79 0.78 0.69 0.89 0.97 0.95 0 .96 0.98

0.4 1 0.92 0.95 1.00 0.99 1 .00 1.00 1.00 1.003 0.83 0 .95 1.00 0.99 1.00 1.00 1.00 1.005 0.94 0 .95 1.00 0.99 0 .93 1.00 1.00 1.00

0.6 1 1.00 0.99 1.00 1.00 1 .00 1.00 1.00 1.003 0.88 0 .99 1.00 1.00 1.00 1.00 1.00 1.005 0.95 0 .99 0.97 1.00 1.00 1.00 1 .00 1.00

500 0.2 1 0.92 0.92 1.00 0.98 1.00 0.99 1.00 1.003 0.90 0.92 0.92 0.98 1 .00 0.99 1 .00 1.005 0.95 0.92 0.96 0.98 1 .00 0.99 1 .00 1.00

0.4 1 1.00 0.99 1.00 1.00 1.00 1.00 1.00 1.003 0.96 0.99 0.99 1.00 1 .00 1.00 1 .00 1.005 1.00 0.99 1.00 1.00 1 .00 1.00 1.00 1.00

0.6 1 1 .00 1.00 1.00 1.00 1.00 1.00 1 .00 1.003 1.00 1.00 1.00 1.00 1.00 1.00 1 .00 1.005 1.00 1 .00 1.00 1.00 1 .00 1.00 1 .00 1.00

and <0.85 for all but one combination of abundanceand search efficiency in Table 2. If this result is anyindication, using the proposed survey design wouldlead to increased sampling effort and higher surveycosts than currently practiced. A legitimate andreasonable question is whether the added cost isworthwhile and affordable. Ultimately, that questionwill have to be answered on a case-by-case basis by theorganizations that are funding the survey. Onecounterbalancing consideration is the cost of failingto detect the presence of a rare population within thearea of a pending adverse impact. Cost would bereduced if searching stopped as soon as one individualof the rare species was detected; however, that practicewould limit the utility of the survey. There certainlyare circumstances when designing a survey to achievea high probability of detecting species presence will beworthwhile. Surveys of federally endangered speciesin areas of proposed adverse impacts would probablybe one of those circumstances.

Acknowledgements

This paper was motivated by my participation in areview of survey protocols for detecting rare popula-tions in the Allegheny River prior to sand and graveldredging. The review was organized by Bob Carlineand sponsored by the Pennsylvania Fish and BoatCommission. I appreciate the informative andthoughtful discussions on mussel sampling with thereview panel members: Bob Carline, Drew Miller, DickNeves, and Gerald Dinkins. I thank Paul Geissler, JimNichols, Penelope Pooler, Jennifer Brown, Rita Villella,Caryn Vaughn, and 3 anonymous referees for helpfulcomments on the paper.

Literature Cited

AMroT, J.-P., AND J. A. DOWNING . 1991 . Endo- and epibenthicdistribution of the unionid mollusk Elliptio complanata.Journal of the North American Benthological Society 10:

280-285.

I

710 D. R. SMITH

CHRISTMAN, M. C. 2000. A review of quadrat-based samplingof rare, geographically clustered populations. Journal ofAgricultural, Biological, and Environmental Statistics 5:168-201.

Di MAIO, J., AND L. D. CORKUM. 1995. Relationship between thespatial distribution of freshwater mussels (Bivalvia:Unionidae) and the hydrological variability of rivers.Canadian Journal of Zoology 73:663-671.

Di MAIO, J., AND L. D. CORKUM. 1997. Patterns of orientation inunionids as a function of rivers with differing hydro-logical variability. Journal of Molluscan Studies 63:531-539.

DOWNING, J. A., AND W. L. DOWNING. 1991. Spatial aggrega-tion, precision , and power in surveys of freshwatermussel populations. Canadian Journal of Fisheries andAquatic Sciences 49:985-991.

ELLIOTT, J. M. 1977. Some methods for the statistical analysisof samples of benthic invertebrates. Freshwater Bio-logical Association Scientific Publication 25. FreshwaterBiological Association, Ambleside, UK.

GREEN, R. H., AND R. C. YOUNG. 1993. Sampling to detect rarespecies. Ecological Applications 3:351-356.

HAAG, W. R., AND M. L. WARREN. 2000. Effects of light andpresence of fish on lure display and larval releasebehaviours in two species of freshwater mussels. AnimalBehaviour 60:879-886.

HAUKIOJA, E., AND T. HAKALA. 1974. Vertical distribution offreshwater mussels (Pelecypoda, Unionidae) in south-western Finland. Annales Zoologici Fennici 11:127-130.

HORNSACH, D. J., AND T. DENEKA. 1996. A comparison of aqualitative and a quantitative collection method forexamining freshwater mussel assemblages. Journal of theNorth American Benthological Society 15:587-596.

KovALAK, W. P., S. D. DENNIS, AND J. M. BATES. 1986. Samplingeffort required to find rare species of freshwater mussels.Pages 34-45 in B. G. Isom (editor). Rationale forsampling and interpretation of ecological data in theassessment of freshwater ecosystems. ATSM STP 894.American Society for Testing and Materials, Philadel-phia, Pennsylvania.

MCARDLE, B. H. 1990. When are rare species not there? Oikos57:276-277.

MCDONALD, L. L. 2004. Sampling rare populations. Pages 11-42 in W. L. Thompson (editor). Sampling rare or elusivespecies: concepts, designs, and techniques for estimatingpopulation parameters . Island Press, Covelo, California.

METCALFE-SMITH, J. L., J. Di MAIO, S. K. STATON, AND G. L.MACKIE. 2000. Effect of sampling effort on the efficiencyof the timed search method for sampling freshwatermussel communities. Journal of the North AmericanBenthological Society 19:725-732.

MILLER, A. C., AND B. S. PAYNE. 1988 . The need for quantitativesampling to characterize size demography and densityof freshwater mussel communities . American Malaco-logical Bulletin 6:49-54.

MILLER, A. C., AND B. S. PAYNE. 1993. Qualitative versusquantitative sampling to evaluate population andcommunity characteristics at a large-river mussel bed.American Midland Naturalist 130:133-145.

[Volume 25

OBERMEYER, B. K. 1998. A comparison of quadrats versustimed snorkel searches for assessing freshwater mussels.American Midland Naturalist 139:331-339.

PERLES, S. J., A. D. CHRISTIAN, AND D. J. BERG. 2003. Vertical

migration, orientation, aggregation, and fecundity of the

freshwater mussel Lampsilis siliquoidia. Ohio Journal of

Science 103:73-78.

RICHARDSON, T. D., AND P. YOKLEY. 1996. A note on sampling

technique and evidence of recruitment in freshwatermussels (Unionidae). Archie fur Hydrobiologie 137:135-140.

SMITH, D. R., R. F. VILLELLA, AND D. P. LEMARti. 2001a. Survey

protocol for assessment of endangered freshwatermussels in the Allegheny River. Journal of the NorthAmerican Benthological Society 20:118-132.

SMITH, D. R., R. F. VILLELLA, AND D. P. LEMARIE. 2003.

Application of adaptive cluster sampling to low-density

populations of freshwater mussels. Environmental and

Ecological Statistics 10:7-15.

SMITH, D. R., R. F. VILLELLA, D. P. LEMARIE, AND S. VON

OETTINGEN. 2001b. How much excavation is needed to

monitor freshwater mussels? Pages 203-218 in R. A.

Tankersley, D. I. Warmolts, G. T. Watters, B. J. Armitage,

P. D. Johnson, and R. S. Butler (editors). Freshwater

Mollusk Symposium Proceedings. Ohio Biological Sur-

vey, Columbus, Ohio.

STEVENS, D. L., AND A. R. OLSEN. 2004. Spatially balancedsampling of natural resources . Journal of the AmericanStatistical Association 99:262-278.

STRAYER, D. L., S. CLAYPOOL, AND S. J. SPRAGUE. 1997. Assessing

unionid populations with quadrats and timed searches.Pages 163-169 in K. S. Cummings, A. C. Buchanan, C. A.Mayer, and T. J. Naimo (editors). Conservation andmanagement of freshwater mussels II. Initiatives for thefuture. Upper Mississippi River Conservation Commit-tee, Rock Island , Illinois . (Available from: Illinois NaturalHistory Survey, 607 East Peabody Drive, Champaign,Illinois 61820 USA.)

STRAYER, D. L., AND J. RALLEY. 1993. Microhabitat use by anassemblage of stream-dwelling unionaceans (Bivalvia),including two rare species of Alasmidonta. Journal of theNorth American Benthological Society 12:247-258.

STRAYER, D. L., AND D. R. SMITH. 2003. A guide to samplingfreshwater mussel populations. American FisheriesSociety Monograph No. 8. American Fisheries Society,Bethesda, Maryland.

STRAYER, D. L., S. SPRAGUE, AND S. CLAYPOOL. 1996. A range-wide assessment of populations of Alasmidonta heterodon,an endangered freshwater mussel (Bivalvia : Unionidae).Journal of the North American Benthological Society 15:308-317.

VAUGHN, C. C., C. M. TAYLOR, AND K . J. EBERHARD . 1997. Acomparison of the effectiveness of timed searches vs.quadrat sampling in mussel surveys. Pages 157-162 in K.S. Cummings, A. C. Buchanan, C. A. Mayer, and T. J.Naimo (editors). Conservation and management offreshwater mussels II. Initiatives for the future. UpperMississippi River Conservation Committee, Rock Island,Illinois. (Available from: Illinois Natural History Survey,

2006] DETECTING RARE FRESHWATER MUSSELS 711

607 East Peabody Drive, Champaign, Illinois 61820USA.)

VILLELLA, R. F., AND D. R. SMITH. 2005. Two-phase sampling toestimate river-wide populations of freshwater mussels.Journal of the North American Benthological Society 24:357-368.

WATrERS, G. T., S. H. O'DEE, AND S. CHORDAS. 2001. Patterns of

vertical migration in freshwater mussels (Bivalvia:Unionida). Journal of Freshwater Ecology 16:541-550.

WELSH, A. H., R. B. CUNNINGHAM, C. F. DONNELLY, AND D. B.

LINDENMAYER. 1996. Modelling the abundance of rarespecies: statistical models for counts with extra zeros.Ecological Modelling 88:297-308.

Wncox, D. B., D. D. ANDERSON, AND A. C. MILLER. 1993.Survey protocol and decision criteria for estimating thelikelihood that Lampsilis higginsi is present in areaswithin the Upper Mississippi River System. Pages 163-

167 in K. S. Cummings, A. C. Buchanan, and L. M. Koch

(editors). Conservation and management of freshwater

mussels. Upper Mississippi River Conservation Com-

mittee, Rock Island, Illinois. (Available from: Illinois

Natural History Survey, 607 East Peabody Drive,

Champaign, Illinois 61820 USA.)

WILLIAMS, B. K., J. D. NICHOLS, AND M. J. CONROY. 2002.

Analysis and management of animal populations.

Academic Press, New York.

YOUNG, M. R., P. J. COSGROVE, L. C. HASTIE, AND B. HENNIGER.

2001. A standardised method for assessing the status of

freshwater mussels in clear , shallow rivers . Journal of

Molluscan Studies 67:395-405.

Received: 25 August 2005

Accepted: 17 March 2006

J. N. Am. Benthol. Soc., 2007, 26(2):253–260� 2007 by The North American Benthological Society

PIT tags increase effectiveness of freshwater mussel recaptures

Jennifer Kurth1

Department of Wildlife Ecology, 5755 Nutting Hall, University of Maine, Orono, Maine 04469-5755 USA

Cynthia Loftin2AND Joseph Zydlewski3

Maine Cooperative Fish and Wildlife Research Unit, US Geological Survey, 5755 Nutting Hall,University of Maine, Orono, Maine 04469-5755 USA

Judith Rhymer4

Department of Wildlife Ecology, 5755 Nutting Hall, University of Maine, Orono, Maine 04469-5755 USA

Abstract. Translocations are used increasingly to conserve populations of rare freshwater mussels.Recovery of translocated mussels is essential to accurate assessment of translocation success. We designedan experiment to evaluate the use of passive integrated transponder (PIT) tags to mark and track individualfreshwater mussels. We used eastern lampmussels (Lampsilis radiata radiata) as a surrogate for 2 rare musselspecies. We assessed internal and external PIT-tag retention in the laboratory and field. Internal tagretention was high (75–100%), and tag rejection occurred primarily during the first 3 wk after tagging. Athin layer of nacre coated internal tags 3 to 4 mo after insertion, suggesting that long-term retention is likely.We released mussels with external PIT tags at 3 field study sites and recaptured them with a PIT pack(mobile interrogation unit) 8 to 10 mo and 21 to 23 mo after release. Numbers of recaptured musselsdiffered among study sites; however, we found more tagged mussels with the PIT-pack searches with visualconfirmation (72–80%) than with visual searches alone (30–47%) at all sites. PIT tags offer improvedrecapture of translocated mussels and increased accuracy of posttranslocation monitoring.

Key words: PIT tags, freshwater mussels, survival, recapture, Lampsilis radiata radiata, translocation.

A goal in the national strategy for the conservation

of native freshwater mussels is to ‘‘develop, evaluate,

and use the techniques necessary to hold and

translocate large numbers of adult mussels’’ (National

Native Mussel Conservation Committee 1997). Suc-

cessful recovery of translocated mussels is essential for

accurate assessment of translocation success. Previous

studies of freshwater mussel translocation used visual

searches to recover mussels with varied success

(Layzer and Gordon 1993, Havlik 1995, Bolden and

Brown 2002, Cope et al. 2003). Survival estimates of

translocated mussels often are based on the number of

mussels recaptured or found dead, and mussels that

are not recaptured are assumed to have emigrated

from the study site (Dunn and Sietman 1997, Hamilton

et al. 1997, Dunn et al. 2000). A review of 33 musseltranslocation studies found a mean estimated survivalrate of 51% (but mortality was not reported in 27% ofthe studies); the average recapture rate was 43%(range: 1–97%) (Cope and Waller 1995).

Passive integrated transponder (PIT) tags may be aneffective tool for tracking translocated mussels toincrease accuracy of survival estimates. PIT tags areelectronic glass-encased microchips that are activatedby an inductive coil. They can be attached to anorganism internally or externally. The tag is passiveuntil activated by a fixed or portable reader with anantenna. When activated, the tag transmits a uniquecode to the reader, identifying the individual organism(Gibbons and Andrews 2004). Tag longevity isindefinite because an internal power source is notneeded. In aquatic systems, PIT tags have been usedextensively to study fish passage past stationaryantennae or readers (Zydlewski et al. 2001). PortablePIT-tag systems are used in shallow waters to assessspatial distributions of local fish populations, fine-scale

1 E-mail address: jennifer_kurth@umit.maine.edu2 To whom correspondence should be addressed. E-mail:

cyndy_loftin@umenfa.maine.edu3 E-mail addresses: jzydlewski@usgs.gov4 judith_rhymer@umit.maine.edu

253

movements, and microhabitat preferences (Roussel etal. 2000, Hill et al. 2006). This mobile application isideally suited to freshwater mussel translocationstudies because mussel movements often occur overshort distances.

Traditional mussel recapture methods depend onvisual encounters and excavation to locate burrowedmussels. PIT tags may enhance mussel recapture atsites where visibility is poor (e.g., turbid water) orwhen mussels are burrowed in sediments. Reliabilityof any tagging method depends on tag retention. Thetagging method selected for freshwater musselsdepends on shell thickness and the type of habitatinto which the tagged mussels will be placed. Internaltagging may be best for thick-shelled species, whereasexternal PIT-tag placement may be more appropriatefor thin-shelled species. In a fast-flowing environmentwith a rocky substrate, an external PIT tag might bedislodged, whereas an internal PIT tag would beprotected from abrasion.

We designed an experiment to evaluate the use ofPIT tags to mark and track individual freshwatermussels as part of a larger study to determine thefeasibility of translocations of 2 state-listed threatenedmussel species (tidewater mucket [Leptodea ochracea]and yellow lampmussel [Lampsilis cariosa]) in responseto an impending dam removal. The objectives of ourstudy were to evaluate internal and external PIT-tagging methods, retention, and posttagging survivalin freshwater mussels and to determine the effective-ness of PIT-tag technology for mussel recaptures. Weused the relatively common eastern lampmussel (Lamp-silis radiata radiata) as a surrogate for the listed speciesto develop the method. We tested internal taggingmethods for future use with thick-shelled species (e.g.,yellow lampmussel) and external attachment for usewith thin-shelled species (e.g., tidewater mucket).

Methods

Internal PIT tagging: mantle separation

We used 2 methods to place internal PIT tags. Formethod 1 (mantle separation), we placed the musselsin sandy substrate, waited until they were activelysiphoning and slightly gaped, and then inserted amicropipette tip between the valves to separate themby ;5 mm. We teased the mantle tissue away from theshell and inserted the PIT tag (Digital Angel, South St.Paul, Minnesota) between the mantle and shell alongthe midventral margin. We also marked all musselsexternally with numbered bee tags (The Bee Works,Orillia, Ontario) cemented (GC Fuji I Glass IonomerLuting Cement; Henry Schein, Melville, New York) tothe posterior end of the left valve. We sealed the bee

tags with Delton Light Curing Pit and Fissure Sealant(Henry Schein). Control mussels received only thenumbered bee tags. We were able to tag ;20 mussels/h with this method. Most of our time was spentwaiting for mussels to gape so we could insert themicropipette tip.

In October 2004, we collected eastern lampmussels(55–101 mm length, n ¼ 164) from the impoundmentthat will be dewatered following the Fort Halifax damremoval in the Sebasticook River near Winslow, Maine.In November 2004 (24–35 d after capture), wepartitioned the mussels into a control (n ¼ 40) and 3tag-type treatment groups: 23-mm tags (n ¼ 40), 12-mm tags (n¼ 44), and 12-mm tags with an antimigra-tion cap (a plastic sleeve encasing one end of the 12-mm tag to encourage tissue adherence; Biomark, Boise,Idaho; n ¼ 40). Each group consisted of mussels of allsizes (control: length 55–99 mm, 23-mm tags: length58–101 mm, 12-mm tags: length 58–99 mm, 12-mmtags with cap: length 58–96 mm).

We maintained mussels in the Aquaculture ResearchCenter (ARC), University of Maine, Orono, Maine, inthree 2.44 3 0.61 3 0.30-m fiberglass tanks filled withsand (13 cm deep) and recirculating water. We dividedthe mussels in each group among 3 replicates (13–15mussels/replicate) and distributed 1 replicate fromeach group in each tank.

We fed the mussels an algal diet (Phaeodactylumtricornutum, Chaetocerus-B., and Nannochloropsis oculata;Algae Spat Formula [Innovative Aquaculture Solu-tions, Inc., Vancouver, British Columbia]) 3 times/wk.During each feeding, we stopped water recirculationand applied 40 to 50 3 109 algal cells/tank (R. Mair,Virginia Polytechnic Institute and State University,personal communication). To simulate changes inseasonal water temperature, we gradually reducedwater temperature from 188C (October) to 108C(December) and maintained 108C until the followingApril, then gradually increased the temperature to188C by June. We monitored the mussels for mortality3 times/wk and examined them for tag retention inNovember 2004 and in February, April, and June 2005.

Internal PIT tagging: mantle incision

We developed a 2nd internal PIT-tagging method(mantle incision) with techniques from the culturedpearl industry (H. Dan, Virginia Polytechnic Instituteand State University, personal communication). Weimplanted PIT tags by inserting a micropipette tipbetween the mussel valves to separate them by ;5 mm,making an incision with a scalpel in the midventralmantle tissue, inserting the tag between the mantle andthe shell through the incision, and then removing the

254 [Volume 26J. KURTH ET AL.

micropipette tip. We also marked all mussels externallywith bee tags on the posterior end of the left valve.Inserting the tags took little time (20 mussels/h). Mostof our time was spent waiting for mussels to gape sowe could insert the micropipette tip.

In June 2005, we collected 112 eastern lampmussels(43–101 mm length) from the Sebasticook Riverimpoundment and randomly assigned the musselsinto 3 groups consisting of a control (n¼ 27) and 2 tag-type treatment groups (23-mm tags: n ¼ 43, 12-mmtags with cap: n ¼ 42) with 3 replicates/group (9–15mussels/replicate), being careful to include mussels ofall sizes in each group. We did not test the 12-mm tagswithout caps because of poor retention in the mantle-separation experiment.

We maintained tagged mussels in the ARC for 21 d toensure tag retention and then placed 1 replicate fromeach group in sand in each of 3 enclosures (1 3 2-mpolyvinyl chloride [PVC] pipe and rebar frames coveredin hardware cloth) in Unity Pond, Maine. Unity Pond isa 1039-ha lake connected to the Sebasticook Riverupstream of the Winslow mussel collection site. UnityPond contains a natural population of eastern lamp-mussels and thus is suitable habitat for the species.Before placing the mussels in the enclosures, wereinserted rejected tags (n ¼ 9). We examined themussels to assess tag retention and survival 60 d(August 2005) and 371 d (June 2006) after tagging.

External PIT tagging

We tested the reliability of external PIT-tag attach-ment and determined the probability of recapturingtranslocated PIT-tagged mussels that were not confinedto enclosures (as in the previous experiment). Weplaced external PIT tags on 238 eastern lampmussels(41–88 mm length) collected during September andOctober 2004 from various sites in Unity Pond (n¼ 90),Sandy Stream (a 1st-order, spring-fed stream that drainsinto Unity Pond; n ¼ 88), and the Sebasticook Riverimpoundment near Winslow (n ¼ 60). We chose thesewater bodies because they had naturally occurringpopulations of eastern lampmussels and the 2 listedspecies, and because, based on neutral markers,Sebasticook River and Sandy Stream populations ofthese mussels were genetically similar (Kelly 2004).

We tagged mussels by cementing a PIT tag to theposterior end of the right valve and a numbered beetag to the posterior end of the left valve. After the first30 tags (at Unity Pond), we completely encapsulatedthe PIT tag in dental cement to increase tag retention.We placed tagged mussels in water before the cementwas fully cured (;5 min after application) to avoidoverdrying and cracking of the cement. We tagged

;30 mussels/h with this method. Most of our timewas spent waiting for the bee-tag sealant to dry. Weused 23-mm tags at all sites. We also used some 12-mmtags at Sandy Stream and Unity Pond because of alimited supply of cement.

We compared survival of translocated musselsamong within-water body, between-water body, andwithin-site (control) translocation treatments. We mea-sured, tagged, and moved mussels to 1 3 2-m plots orreplaced them where they had been found (Table 1). Wemarked the corners of the plots with stakes withflagging, and recorded Global Positioning System(GPS) locations for each plot and for each of the taggedmussels that were returned to their original location.

We recaptured externally PIT-tagged mussels with amobile PIT detection unit (PIT pack). The PIT packused Destron Fearing FS1001A DC-powered, fullduplex transceivers and custom-designed portableantennas. When a PIT tag was within range of anantenna (;0.5 m), the tag emitted a 134.2-kHz (ISOstandard frequency) radio frequency, which wastransmitted back to the receiver for decoding. Theantennas, enclosed in an airtight PVC wand andattached to the transceiver, consisted of several wrapsof 12- to 18-gauge wire, with inductance valuesranging from 325 to 375 lH and a set of capacitors(Hill et al. 2006). The capacitors were attached to anantenna lead cable from the transceiver, fixing thecapacitance between 33 and 44 nF. The fixed capaci-tance was used within the transceiver in conjunctionwith the adjustable capacitance to tune the resonancefrequency of the system to 134.2 kHz (Hill et al. 2006).We tuned the adjustable capacitor while antennas weresubmerged. We conducted all field experiments withthe PIT pack tuned to phase 0 to 2%, signal 1 to 20%,and current 2.5 to 5.0 amps.

We searched the release sites for externally PIT-tagged mussels ;30 d after tagging (October 2004) andvisually confirmed recaptures with snorkeling. If thePIT-tag reader registered a tag but no mussel wasobserved, we assumed the mussel had burrowed intothe substrate. To minimize substrate disturbance, wedid not excavate burrowed mussels preparing to

TABLE 1. Numbers of mussels tagged with passiveintegrated transponder tags in each translocation treatmentduring September and October 2004.

Site

Tagged andreplaced

(site control)

Movedwithin

water body

Translocatedfrom

SebasticookRiver

Sandy Stream 30 26 32Unity Pond 30 30 29Sebasticook River 30 30 –

2007] 255PIT TAGS AND FRESHWATER MUSSEL TRANSLOCATIONS

overwinter. These data were not used in the calcula-tions of recapture success because the signals mayhave been from detached tags.

During June and July 2005 (271–355 d after tagging)and July and August 2006 (670–750 d after tagging),we searched again for PIT-tagged mussels at therelease sites, beginning at the last location recordedwith GPS during October 2004. In 2005, we conductedinitial searches without the PIT pack to providerecapture percentages with visual searches only. Wevisually searched each site for 2 d. Approximately 1wk later, we searched the sites using PIT-pack searcheswith visual confirmation and excavation to confirmrecaptures (3–4 d/site). In 2006, we repeated the PIT-pack searches with visual confirmation (3 d/site).Water clarity was too poor to conduct visual searchesin 2006. If the PIT pack detected a tagged mussel, butwe did not see the mussel, we excavated the areawithin 0.5 m of the signal to 15 to 45 cm deep todetermine if the signal was coming from a burrowedmussel or an unattached tag. If we found no taggedmussel after excavation, we assumed the tag hadbecome detached. We searched (with snorkeling andthe PIT pack) the sites at Unity Pond and theSebasticook River 4 times each to at least 3 m beyondthe perimeter of the original study area to detectmussels that may have moved. We also searched theshorelines for valves from dead mussels. Extensive icescouring and spring flooding substantially reconfig-ured the substrate at the Sandy Stream site, so inaddition to searching the study area plus 3 m beyondthe perimeter, we also swept the antenna bank to bankdownstream of the site for 200 m over a total of 3 d. Wecalculated recapture rates by dividing the number ofmussels recaptured at each site by the number tagged.

Data analysis

We used adjusted v2 for small sample sizes (Gotelliand Ellison 2004) for all analyses.

We compared long-term tag retention among tagtypes and mussel mortality among treatments andcontrols for both mantle separation and mantleincision methods. We compared the percentages ofrecaptures using visual searches alone with thenumber of recaptures using PIT-pack searches withvisual confirmation.

Results

Mussel retention of internal PIT tags in the laboratory(mantle separation)

Five percent of the PIT tags were rejected within 2wk of internal placement via mantle separation. By 100

d after tagging, rejection had increased to 10% for 12-mm tags with caps, 12.5% for 23-mm tags, and 30% for12-mm tags without caps. High mortality with thismethod was more troubling than the rejection rates. By100 d after tagging, mortality rates were 3% for thecontrol group (no tags), 10% for the group with 12-mmtags with caps, 25% for the group with 23-mm tags,and 27% for the group with 12-mm tags without caps.This mortality may have been caused by inexperiencewith the tagging procedures and mussel aquaculturehusbandry (mortality in control mussels was 3% 100 dafter tagging and 73% 244 d after tagging), so wediscontinued using the 12-mm tags without caps,switched to the mantle-incision method, and retainedthe tagged mussels in field enclosures.

Long-term tag retention did not differ among tagtypes (adjusted v2 ¼ 5.61, p ¼ 0.691, df ¼ 8), andmortality did not differ among the tag-type and controlgroups (adjusted v2¼7.97, p¼0.716, df¼11) 100 d aftertagging. We examined the condition of the PIT tags inall mussels that died over winter. By 90 d after tagging,all 12-mm PIT tags with caps were coated with nacreand attached to a valve. By 120 d after tagging, 23-mmand 12-mm PIT tags without caps that had not beenrejected were similarly attached.

Mussel retention of internal PIT tags in field enclosures(mantle incision)

All mussels in the control and tag-type groups(mantle incision) were still alive 60 d after tagging (40d after transport from the ARC to the Unity Pondenclosures) (Table 2). One 23-mm tag was rejected afterthe mussels were placed in the enclosures; this rejectedtag was not one of the tags that had been rejected andreinserted within the 2-wk posttagging observationperiod. By June 2006 (371 d after tagging), 2 mussels inthe enclosures had died (1 control, 1 with a 23-mmtag), and one 12-mm tag with cap was rejected. Long-term tag retention did not differ among tag types(adjusted v2¼ 4.26, p¼ 0.833, df¼ 8), and mortality didnot differ among control and tag-type groups (adjust-ed v2 ¼ 3.72, p ¼ 0.882, df ¼ 11) 371 d after tagging.

Retention of external PIT tags and recapture of mussels inthe field

Overall, ;93% of the recaptured tagged musselsretained the PIT tag (Table 3). Recapture rates withPIT-pack searches with visual confirmation exceededrecaptures from visual searches alone at all study sitesduring June and July 2005 (adjusted v2 ¼ 10.198, p ¼0.0014, df ¼ 1; Fig. 1). During June and July 2005 andJuly and August 2006, we used a combination of visualsearches alone and PIT-pack searches with visual

256 [Volume 26J. KURTH ET AL.

confirmations to recapture 77% of externally taggedmussels at Unity Pond and 80% of externally taggedmussels in the Sebasticook River (combined resultsfrom 2005 and 2006 recaptures). In Sandy Stream,where ice scouring and spring flooding reconfiguredthe substrate, we recovered only 25% of the taggedmussels. Ninety-five percent of the mussels we didrecapture were found using PIT-pack searches withvisual confirmation, and only 1 mussel was foundusing visual searches alone. In Sandy Stream, wefound 71% of recaptured mussels .100 m from theirOctober 2004 locations, whereas we found recoveredmussels in Unity Pond and the Sebasticook River ,2m from their September–October 2004 locations.Seventeen percent (Unity Pond), 17% (SebasticookRiver), and 3.5% (Sandy Stream) of the recapturedmussels found with the PIT pack were completelyburrowed into the substrate (Fig. 1). We found mostburrowed mussels within 6 cm of the sedimentsurface. However, the PIT pack detected 1 tagged(23-mm tag) living mussel burrowed 45 cm into thesubstrate and 3 tagged dead mussels 20 to 30 cmbelow the substrate surface in Sandy Stream. We alsofound 1 dead mussel with a PIT tag during shoresweeps at the Sebasticook River site.

Discussion

Tagging methods

Low mortality (,2%), high tag retention (;97%),and evidence that tags had fused to the shell 3 to 4 moafter tagging suggest that internal PIT tagging usingthe mantle-incision method may be a viable method oftagging thick-shelled freshwater mussel species thatcan be pried open for tag insertion without damagingthe shell. Long-term survival of captive freshwatermussels is low (Patterson et al. 1997, 1999, Nichols andGarling 2002), and high mortality of captive mussels inour study (73–93% 255 d after tagging) might beattributed to inadequate nutrition, winter watertemperatures in the ARC that exceeded temperaturesat the mussel collection sites, and physiologicalstresses experienced by captive mussels that weregravid when captured. The low mortality of musselstagged with the mantle-incision method and placed inthe enclosures at Unity Pond supports this assertion.We strongly recommend field trials rather thanaquaculture experiments for testing methods intendedfor use in the field to remove uncertainty of the effectsof captivity on mussel survival.

External PIT-tag retention also was high (;93%)

TABLE 2. Percent mortality and % tag retention (60 d and 371 d after tagging using the mantle-incision method) of easternlampmussels with internal passive integrated transponder tags in field enclosures in Unity Pond, Maine.

60 d after tagging 371 d after tagging

Treatment % mortality % tag retention % mortality % tag retentiona

23-mm tag (n ¼ 43) 0 98 2.5 97.512-mm tag with cap (n ¼ 41) 0 100 0 97.4Control (no tag) (n ¼ 27) 0 – 4.3 –

a Includes mussels that died with retained tags

TABLE 3. Percent recapture, % mortality, and % tag retention of externally passive integrated transponder–tagged easternlampmussels in translocation experiments within and among sites (;21 mo after tagging) in Maine.

Sitea Treatment Number tagged % recapture % mortalityb % tag retentionc

Unity PondTranslocated from Sebasticook River

impoundment29 93.1 0 100

Translocated within Unity Pond 32 74.2 0 78.3Site control (not moved) 30 63.3 0 89.5

Sebasticook RiverTranslocated within Sebasticook River

impoundment30 93.3 0 96.4

Site control (not moved) 30 66.7 6.7 100Total 151 78.0 1.3 93.2

a Sandy Stream data omitted because of winter ice scouring and spring floodingb Percent mortality calculated only for recaptured musselsc Retention calculated as % recaptured mussels retaining tags

2007] 257PIT TAGS AND FRESHWATER MUSSEL TRANSLOCATIONS

when the PIT tag was completely encapsulated incement and the mussel was placed in water within 5min of cementing. However, retention was morevariable with external tagging than with internaltagging methods, and ranged from ;78 to 100% atthe Unity Pond site 9 mo after tagging. We attributelow retention to incomplete coverage with cement.Retention of tags completely encapsulated withcement ranged from 89.5 to 100%. We observedevidence of some cement loss from recapturedmussels; occasional reapplication of cement willensure long-term retention of external PIT tags.Internal tag placement via mantle incision is a viablealternative to external attachment in environmentswhere tag loss from abrasion is likely.

Previous studies assessed external freshwater mus-sels tagging methods with visual searches to relocatemussels marked with numbered tags (Lemarie et al.2000) or coded wire tags inserted into mussels held insuspended pocket-nets (Layzer and Heinricher 2004).Both of these tagging methods resulted in higher tagretention than in our study, but mussels tagged usingthese methods can be detected only with visualsearches. PIT tags provide an alternative tool forfinding mussels, and this method is especially usefulfor long-term monitoring or where visual searches areimpractical or time consuming.

Mussel recapture efficiency

The proportion of mussels visible at the substratesurface may vary by locality, time of year, species, andgender. Smith et al. (2001) detected only 31% ofclubshells (Pleurobema clava) at the substrate surface,whereas 52% of northern riffleshells (Epioblasmatorulosa rangiana; 80% females, 45% males) werevisible. Wick (2006) observed that .90% of eastern

lampmussels had burrowed to 10 to 15 cm at SandyStream by August, but only 26% had burrowed in theSebasticook River impoundment at that time.

Because the water was turbid, we found burrowedmussels and mussels that would have been overlookedhad the sites been searched only visually. For example,water clarity in Unity Pond was routinely poor, andonly 47% of tagged mussels were recaptured visually,whereas 72% of tagged mussels were recaptured withthe PIT pack and visual confirmation. In the Sebasti-cook River, where the visibility was compromised bysilt covering the mussels, the recaptures with the PITpack and visual confirmation (80%) were .23 those ofthe visual searches alone (29%). Initially, PIT tags alsoprovided a visual cue of tagged mussels in clear water,but after several months in the water, the cement wasstained or covered with algae and indistinguishablefrom the shell. When first applied, the white cementmight provide a visual cue to predators, but only 1shell was found in a shoreline midden in our study.Tinting the cement a dark color might eliminate thispossible problem.

Low recaptures in Sandy Stream probably werecaused by extensive downstream displacement ofmussels in late winter and early spring when ice scourand high water flows during snowmelt reconfiguredthe stream bottom. The low recapture rates of PIT-tagged mussels at this site were attributed to tag lossfrom severe abrasion, burial in sediment beyond thedetection limit, or transport beyond the regionssearched.

Limitations of PIT tags in field applications

Debris on the substrate and signal interferencecaused by nearby iron objects (Hill et al. 2006) canaffect reliability of the PIT pack. The antenna config-uration we used also is limited to sites with waterdepth ,2 m. Maximum effective depth and antennarange are not necessarily uniform among sites; theselimitations should be identified at each field site so thatmussel absence can be distinguished from nondetec-tion caused by equipment limitations. Reducing theantenna size for use while snorkeling, waterproofingthe PIT pack for diver use, and lengthening theantenna handle are modifications that will broadenfield use of this tool. At present, PIT-tag use is limitedto larger mussels (length .20 mm). However, smallertags with greater detection ranges are in development,and eventually it should be possible to tag smallermussels, at least externally. Although internal tagswere retained, the ;3-wk captive period to ensure tagretention could limit the usefulness of internal tags.Internally tagged mussels should be held in field

FIG. 1. Percentages of mussels externally tagged withpassive integrated transponder (PIT) tags recaptured usingdifferent methods during June and July 2005.

258 [Volume 26J. KURTH ET AL.

enclosures during the initial posttagging period whentag rejection may occur. Retaining a subset ofinternally tagged mussels may be a viable alternativefor estimating tag retention proportions when largenumbers of mussels are translocated.

The initial cost of the PIT tags and reader mayexceed start-up costs for other mussel-tagging meth-ods. The PIT pack (transceivers, batteries, antenna) weused cost ;$10,000 to construct and was designed forresearch on a variety of organisms such as fish,mussels, and amphibians. Smaller units can bedeveloped for ;$2500. The PIT tags we used cost$3.50 each, but the tags work indefinitely. On the otherhand, the percentage of tagged mussels recapturedusing PIT tags far exceeded the percentage recapturedduring visual searches. Visual searches can be timeconsuming and labor intensive. For long-term moni-toring of individuals and populations, the addedinitial costs may be recouped over time, and it maybe possible to share the costs with other investigatorsusing PIT tags.

In conclusion, PIT tags permit repeated, nondestruc-tive sampling of individuals with little disturbance,last indefinitely, and appear to have negligible effectson short-term survival of freshwater mussels. PIT tagswere retained using both internal and externalattachment methods. Thus, the choice of taggingmethod will depend on shell thickness, habitatcharacteristics, and ease of implementation in the field.

The need for freshwater mussel translocations toprotect and conserve threatened and endangeredmussel species will increase as aquatic habitat alter-ation continues. Superior recapture rates with PIT tagssuggest that this tool is valuable for use in musseltranslocations and monitoring and may improveaccuracy of survival estimates for assessing transloca-tion success. Because PIT tags have indefinite longev-ity, they can be used in monitoring both translocatedmussels and populations at sites of concern, especiallypopulations of endangered or threatened species.Moreover, because PIT tags provide reliable individualidentification, they may be a useful tool for monitoringthe growth and survival of individual mussels.

Acknowledgements

The authors thank C. Currier and R. Clark for fieldassistance, N. Greenberg (University of Maine Aqua-culture Research Center), B. Swartz, and K. Kemper(Maine Department of Inland Fisheries and Wildlife),and R. Mair, H. Dan, and R. Neves (VirginiaPolytechnic Institute and State University) for adviceand assistance. We thank G. Zydlewski, M. Kinnison,B. Swartz, 2 anonymous referees, and D. Strayer for

reviewing early drafts of the manuscript. Our researchwas funded by the Maine Department of InlandFisheries and Wildlife through the State WildlifeGrants Program and the Endangered and NongameWildlife Fund, and by the US Geological Survey–StatePartnership Program, the US Geological Survey MaineCooperative Fish and Wildlife Research Unit, theMaine Outdoor Heritage Fund, and the University ofMaine, Department of Wildlife Ecology. Reference totrade names and commercial products does not implyendorsement or recommendation by the US Govern-ment. This is article no. 2903 of the Maine Agriculturaland Forest Experiment Station.

Literature Cited

BOLDEN, S. R., AND K. M. BROWN. 2002. Role of stream, habitat,and density in predicting translocation success in thethreatened Louisiana pearlshell, Margaritifera hembeli(Conrad). Journal of the North American BenthologicalSociety 21:89–96.

COPE, W. G., M. C. HOVE, D. L. WALLER, D. J. HORNBACH, M. R.BARTSCH, L. A. CUNNINGHAM, H. L. DUNN, AND A. R.KAPUSCINSKI. 2003. Evaluation of translocation of unionidmussels to in situ refugia. Journal of Molluscan Studies69:27–34.

COPE, W. G., AND D. L. WALLER. 1995. Evaluation of freshwatermussel translocation as a conservation and managementstrategy. Regulated Rivers: Research and Management11:147–155.

DUNN, H. L., AND B. E. SIETMAN. 1997. Guidelines used in fourgeographically diverse unionid translocations. Pages176–183 in K. S. Cummings, A. C. Buchanan, C. A.Mayer, and T. J. Naimo (editors). Conservation andmanagement of freshwater mussels II: initiatives for thefuture. Upper Mississippi River Research ConservationCommittee, Rock Island, Illinois.

DUNN, H. L., B. E. SIETMAN, AND D. E. KELNER. 2000. Evaluationof recent unionid (Bivalvia) translocations and sugges-tions for future translocations and reintroductions. Pages169–183 in P. D. Johnson and R. S. Butler (editors).Freshwater Mollusk Symposia Proceedings: Part II.Proceedings of the First Freshwater Mollusk Conserva-tion Society Symposium, March 1999, Chattanooga,Tennessee. Ohio Biological Survey, Columbus, Ohio.

GIBBONS, J. W., AND K. M. ANDREWS. 2004. PIT tagging: simpletechnology at its best. BioScience 54:447–454.

GOTELLI, N. J., AND A. M. ELLISON. 2004. A primer of ecologicalstatistics. Sinauer Associates, Sunderland, Massachu-setts.

HAMILTON, H., J. B. BOX, AND R. M. DORAZIO. 1997. Effects ofhabitat suitability on the survival of translocatedfreshwater mussels. Regulated Rivers: Research andManagement 13:537–541.

HAVLIK, M. E. 1995. Are unionid mollusk translocations aviable mitigation technique? The Wolf River, Wisconsin,experience, 1992–1995. Pages 184–195 in K. S. Cum-mings, A. C. Buchanan, C. A. Mayer, and T. J. Naimo

2007] 259PIT TAGS AND FRESHWATER MUSSEL TRANSLOCATIONS

(editors). Conservation and management of freshwatermussels II: initiatives for the future. Upper MississippiRiver Research Conservation Committee, Rock Island,Illinois.

HILL, M. S., G. B. ZYDLEWSKI, J. D. ZYDLEWSKI, AND J. D.GASVODA. 2006. Development and evaluation of portablePIT tag detection units: PITpacks. Fisheries Research 77:102–109.

KELLY, M. W. 2004. Conservation genetics of two rarefreshwater mussels: the tidewater mucket (Leptodeaochracea) and the yellow lampmussel (Lampsilis cariosa).MS Thesis, University of Maine, Orono, Maine.

LAYZER, J. B., AND M. E. GORDON. 1993. Reintroduction ofmussels into the Upper Duck River, Tennessee. Pages 89–92 in K. S. Cummings, A. C. Buchanan, and L. M. Koch(editors). Conservation and management of freshwatermussels. Upper Mississippi River Conservation Com-mittee, Rock Island, Illinois.

LAYZER, J. B., AND J. R. HEINRICHER. 2004. Coded wire tagretention in ebonyshell mussels Fusconaia ebena. NorthAmerican Journal of Fisheries Management 24:228–230.

LEMARIE, D. P., D. R. SMITH, R. F. VILLELLA, AND D. A. WELLER.2000. Evaluation of tag types and adhesives for markingfreshwater mussels (Mollusca: Unionidae). Journal ofShellfish Research 19:247–250.

NATIONAL NATIVE MUSSEL CONSERVATION COMMITTEE. 1997.National strategy for the conservation of native fresh-water mussels. Journal of Shellfish Research 17:1419–1428.

NICHOLS, S. J., AND D. GARLING. 2002. Evaluation of substitutediets for live algae in the captive maintenance of adult

and subadult Unionidae. Journal of Shellfish Research21:875–881.

PATTERSON, M. A., B. C. PARKER, AND R. J. NEVES. 1997. Effectsof quarantine times on glycogen levels of nativefreshwater mussels (Bivalvia: Unionidae) previouslyinfested with zebra mussels. American MalacologicalBulletin 14:75–79.

PATTERSON, M. A., B. C. PARKER, AND R. J. NEVES. 1999.Glycogen concentration in the mantle tissue of freshwa-ter mussels (Bivalvia: Unionidae) during starvation andcontrolled feeding. American Malacological Bulletin 15:47–50.

ROUSSEL, J.-M., A. HARO, AND R. A. CUNJAK. 2000. Field-test ofa new method for tracking small fishes in shallow riversusing passive integrated transponder (PIT) technology.Canadian Journal of Fisheries and Aquatic Sciences 57:1326–1329.

SMITH, D. R., R. F. VILLELLA, AND D. P. LEMARIE. 2001. Surveyprotocol for assessment of endangered freshwatermussels in the Allegheny River, Pennsylvania. Journalof the North American Benthological Society 20:118–132.

WICK, P. C. 2006. Fish host and demographics of Lampsiliscariosa and Leptodea ochracea, two threatened freshwatermussels in Maine. MS Thesis, University of Maine,Orono, Maine.

ZYDLEWSKI, G. B., A. HARO, G. WHALEN, AND S. D. MCCORMICK.2001. Performance of stationary and portable passivetransponder detection systems for monitoring of fishmovements. Journal of Fish Biology 58:1471–1475.

Received: 4 April 2006Accepted: 18 December 2006

260 [Volume 26J. KURTH ET AL.