The freshwater mussel (Bivalvia: Unionidae) fauna of North America is the richest of any continent, yet over 70% of these molluscs are listed as ‘endangered’, ‘threatened’ or of ‘special concern’ (Williams et al., 1993). Declines in richness and abundance at local and regional scales have continued largely due to anthropogenic activities (Nedeau, McCollough & Swartz, 2000). With few exceptions, the larvae of freshwater mussels, called glochidia, are obligate parasites of specific fish species, and mussel dispersal occurs primarily while glochidia are attached to the gills, fins or scales of a host fish. Their dependence on suitable host fish to metamorphose into juveniles and complete their life cycle (Watters, 1992) increases their susceptibility to environmental perturbation, as declines in host fish abundance, introduction of exotic fishes and alterations of fish communities can have drastic consequences on the ability of mussels to reproduce successfully. Host identification has been the focus of considerable research but, for many mussel species, the list of suitable host fish is incomplete or restricted to species in a limited portion of the mussel’s range. Identification of host fish is generally done under laboratory conditions; however, comparatively little is known about host use in the wild, and the relative importance of some host species to recruitment and dispersal of freshwater mussels (Martel & Lauzon-Guay, 2005) cannot be determined in the laboratory. Laboratory designation of host fish may not take into consideration the behavioural characteristics of the host fish, such as habitat use (Martel & Lauzon-Guay, 2005) and attraction to mantle lures or conglutinates, and host fish deemed suitable in the laboratory may serve a limited role in natural populations. On a larger scale, relationships between mussels and their host fish are further complicated as fish communities and host abundance differ among localities and watersheds (Vaughn & Taylor, 2000; Martel & Lauzon-Guay, 2005), and the ability of mussels to adapt to local host species (Rogers, Watson & Neves, 2001) demonstrates the need for a more complete understanding of host use in the wild. This information is critical for conservation and effective planning for natural recovery of many declining mussel species. The study of natural parasitism is greatly inhibited by the difficulty in obtaining reliable identification of encysted glochidia (Hoggarth, 1992). Identification to species using morphological characters is not always possible, and in some mussel communities it is not possible to identify glochidia below the level of subfamily (Weiss & Layzer, 1995). In an effort to provide a more reliable method for identification of glochidia, White, McPheron & Stauffer (1996) developed a dichotomous molecular identification key for the unionids of French Creek, PA, USA, that utilized restriction-fragment length polymorphisms (RFLPs) visualized on agarose gels for species identification. Using a similar approach that added a nested polymerase chain reaction (PCR), a dichotomous RFLP-based molecular key was developed for the unionids of Europe (Gerke & Tiedemann, 2001). However, these keys have not been rigorously applied to the study of natural parasitism in unionids, and there has been difficulty in effectively extracting DNA from a single glochidium using the published methods. The purpose of creating a molecular identification key to the ten species of freshwater mussels that occur in Maine was to determine host fish use in the wild for two species that are state-listed as ‘threatened’, the yellow lampmussel [Lampsilis cariosa (Say, 1817)] and tidewater mucket [Leptodea ochracea (Say, 1817)]. The goals were to examine host use in multiple localities, to determine if host fish identified in the laboratory are used as hosts in the wild and to determine if additional species are potential hosts by capturing naturally parasitized fish and identifying glochidia with the key. In addition, three other mussel species are currently state-listed as ‘special concern’, so identification of fish hosts for wild populations of mussels in Maine is essential information for regional conservation planning. To develop the key, all ten species were sampled at or within close proximity to localities in the three river drainages where Lampsilis cariosa and Leptodea ochracea occur (Fig. 1). Approximately 15 mg of mantle tissue was sampled from morphologically identified adult mussels using methods that are minimally invasive (Berg et al., 1995), and DNA was extracted from tissue samples using a QIAamp DNA Mini Kit (Qiagen Inc., Valencia, CA, USA) following the manufacturer’s instructions for the Tissue Protocol. The NADH dehydrogenase subunit 1 (ND1) gene of the mitochondrial (mtDNA) genome was amplified from each species using the unionid-specific primers Leu-uurF and LoGlyR (Serb, Buhay & Lydeard, 2003). Two specimens of each species from the widest possible distribution within the study area were selected for sequencing to examine intraspecific variation within the ND1 gene. PCR reactions consisted of 200 ng genomic DNA, 1 PCR buffer (20 mM Tris-HCl, pH 8.4, 50 mM KCl), 2 mM MgCl2, 0.2 mM dNTPs, 0.5 mM forward and reverse primers, 1.25 U Taq Polymerase (Invitrogen, Carlsbad, CA, USA), and sterile water to a total reaction volume of 50 ml. Reactions were amplified with a PTC programmable thermocycler with a heated lid (MJC Research, Inc., Watertown, MA, USA). Reaction conditions for doublestranded amplification consisted of an initial denaturation at 948C for 5 min, followed by 30 cycles of 948C for 45 s, 548C for 60 s, and 728C for 60 s; and a final extension of 728C for 5 min. PCR products were purified with Nanosep 30 k Centrifugal Devices (Pall Corp., Ann Arbor, MI, USA) following the manufacturer’s instructions. Products were sequenced in both directions using Big Dye (Perkin Elmer) terminator cycle sequencing (v. 3.1) and read with an ABI 3730 Capillary Sequencer. Glochidia were obtained from gravid female mussels by flushing the marsupium with water using a syringe. On naturally parasitized fish, individual glochidia were removed from ethanolpreserved gills with dissecting probes under 10 magnification using a dissecting microscope, and effort was made to minimize the amount of gill tissue attached to each glochidium. DNA was extracted from a single glochidium using a QIAamp DNA Mini Kit with a slightly modified Tissue Protocol. Detailed instructions of the Tissue Protocol included in the QIAamp DNA Correspondence: S.C. Kneeland; e-mail: stephen.kneeland@umit. maine.edu