Leech specimens were collected from hand-captured individuals of A. marmorata along a 40-mile stretch of the Lower Rogue River in Josephine and Curry Counties, southwest Oregon, USA (Table 1) during 24–27 May, 15–18 June, and 13–16 July of 2021 as described in Reilly et al. (2023); see Table 2. Preservation methods for the leeches were provided in Reilly et al. (2023). In addition to sequences of P. parasitica from Richardson et al. (2020), a total of 3 samples were obtained from Maryland, New Hampshire, and Vermont (Table 3). Leeches were examined with a Wild M5 stereomicroscope. Representative voucher specimens of P. parasitica from Oregon and localities within the native regions were deposited in the Department of Invertebrate Zoology Collections at the National Museum of Natural History (NMNH), Smithsonian Institution and the North Carolina Museum of Natural Sciences (Tables 2, 3). Information about the voucher specimens of the sequences used in Richardson et al. (2020) are available in that publication.

Novel sequences in this study were generated at the Laboratories of Analytical Biology (L.A.B.) at the NMNH, Smithsonian Institution. Caudal sucker tissue was sampled in order to avoid contamination of host DNA from the gastric region. Total genomic DNA was extracted from tissue of the caudal sucker of 62 individuals of P. parasitica from the lower Rogue River, Oregon. DNeasy® blood and tissue kit (Qiagen, Inc., Valencia, California) was used for tissue lysis and DNA purification. For the proteinase K treatment step, tissue samples were lysed overnight at 56. DNA was eluted from the spin columns with 100 µL of buffer twice for a total volume of 200 µL. Partial sequences of the mitochondrial COI gene fragment were obtained with the use of the primers as specified by Folmer et al. (1994) in order to confirm the species identification based on morphological examination. Polymerase chain reaction (PCR) was performed in 10 µL reactions with 1 µL of genomic DNA and final concentrations of 3 µmol of each primer, 1x GoTaq® Hot Start Master Mix (Promega, Madison, WI), 0.2 ug/uL BSA (0.1 uL of 20 mg/ml BSA per 10 uL of reaction mixture), 1% DMSO (0.1 uL DMSO per 10 uL of reaction mixture). PCRs were performed in an Applied Biosystems 2720 Thermal Cycler (Applied Biosystems, Foster City, California) under the following thermal profile for COI: 95 (5 min), followed by 5 cycles of 94 (30 sec), 50 (45 sec), 72 (1 min), then 35 cycles of 94 (30 sec), 45 (45 sec), 72 (1 min), and final extension at 72 (8 min). ExoSAP-IT® Express (Affymetrix/Thermofisher Scientific) was used to purify the PCR products, and cycle sequencing was performed with BigDye® Terminator v3.1 (Applied Biosystems, Foster City, California). Cycle-sequenced products were purified using Sephadex™ G-50 Fine (GE Healthcare, Chicago, Illinois), and DNA sequencing was performed on an ABI 3730 at the L.A.B.

Contig assembly and sequence editing was performed using Geneious Prime, Version 2023.1.2 (http://www.geneious.com, Biomatters 2023). Novel molecular sequences were deposited in GenBank (Tables 2, 3; Sayers et al. 2022). Sequences were compared to the GenBank non-redundant (nr) database using BLASTn to affirm species affinity and to detect any potential contamination.

Additional DNA sequence data of COI for representatives of P. parasitica from across North America and the outgroups analyzed in Richardson et al. (2020) were downloaded from GenBank (see Richardson et al. 2020 for specimen data). Outgroup selection was based on sister group relationships from the most recent published phylogeny of the genus Placobdella (de Carle et al. 2017). Sequence alignment was accomplished using the MAFFT online platform (Katoh and Standley 2013) applying default settings. The alignment was checked by eye for gaps and the sequences were translated to amino acids as an independent assessment of sequence quality, however the dataset was subsequently checked for gaps and stop codons prior to phylogenetic analyses using Mesquite v.3.70 (Maddison and Maddison 2021).

Pairwise (uncorrected p) sequence distances were calculated in MEGA 11: Molecular Evolutionary Genetics Analysis version 11 (Tamura et al. 2021). The genetic distances were calculated using the uncorrected p-distance model, with uniform rates among sites, and a complete deletion of missing data. Since COI sequences of all specimens of P. parasitica from the Rogue River were effectively identical based on p-distance data and amino acid translation, a single sequence (139TS) was included as a representative in the phylogenetic analysis and species delimitation analyses.

An alignment of only P. parasitica sequences was used to generate a haplotype network and to calculate haplotype number, nucleotide diversity, and Tajima’s D (Tajima 1989) with the TCS algorithm (Clement et al. 2002) in PopArt (Leigh and Bryant 2015). For the Tajima’s D statistic, a negative value indicates a recent population expansion, a value of zero indicates no selection, and a positive value indicates either a balancing selection or population contraction.

The COI data was partitioned by codon position for a total of three partitions. Estimation of substitution models by codon position was performed using ModelFinder within IQTREE (Kalyaanamoorthy et al. 2017), resulting in the following models as best fit by partition by the Bayesian information criterion (BIC): first codon position = TNe+G4, second codon position = F81+F+I, and third codon position = HKY+F+G4. Maximum likelihood (ML) analyses were performed with IQTREE multicore version 2.1.3 (Nguyen et al. 2015) using the models suggested for each unlinked partition, the -spp option that allowed each partition to have its own evolutionary rate, and 1,000 standard bootstrap replicates (SBS; Felsenstein 1985) on “Hydra”, the Smithsonian Institution High Performance Cluster (https://doi.org/10.25572/SIHPC). Helobdella modesta (Verrill, 1872), Helobdella bowermani Moser, Fend, Richardson, Hammond, Lazo-Wasem, Govedich & Gullo, 2013, and Helobdella octatestisaca Lai & Chang, 2009 served as outgroups. Trees were visualized in FigTree, version 1.4.4 (Rambaut 2018) and edited with Adobe^© Illustrator Creative Cloud (https://www.adobe.com).

We used unique COI sequence data to assess if P. parasitica samples represented a single species or if separate, independently evolving entities exist. Three methods were used to assess species delineation, including the Automatic Barcode Gap Discovery (ABGD; Puillandre et al. 2012)), the Poisson tree process (PTP; Zhang et al. 2013) including multi-rate (mPTP; Kapli et al. 2017) and Bayesian implementation (bPTP), and the generalized mixed Yule-coalescent (GMYC; Fujisawa and Barraglough 2013). Outgroups were removed prior to implementation of these methods. Identical sequences were identified using the genetic distance data and subsequently removed to avoid the inclusion of redundant information.

The methods of mPTP, bPTP, and GMYC utilized ultrametric trees generated using Bayesian Inference in BEAST v1.10.4 as implemented on the CIPRES Science Gateway (Miller et al. 2010). BEAUTi (Bayesian Evolutionary Analysis Utility) v1.10.4 generated .xml files for all BEAST runs. Tree priors for all analyses were selected under a Coalescent Process with constant population size. Nucleotide substitution models were estimated by the corrected Bayesian information criteria (BICc) using jModelTest (Posada 2008): COI first position = F81+F+I, COI second position = TPM2+F+G4, COI third position = TNe+G4. Since none of these models could be utilized in the Bayesian framework, each partition used GTR + G4. The dataset was analyzed with MCMC analyses with 10⁸ generations and trees were sampled every 10,000 generations. TRACER v1.7.2 was used to verify convergence of the MCMC. A maximum clade credibility (MCC) consensus tree was obtained for the BEAST dataset in Tree Annotator v1.10.4 after annotating the remaining 9001 trees after burn-in.

To test the strength of our mPTP and bPTP delineations, these analyses were also run with trees generated with ML analyses in IQTREE. Analyses using GMYC were performed in R v.3.5.1 (R Core Team 2021) using the package SPLITS v.1.0-19 (Ezard et al. 2017) on the phylogeny. PTP analyses were carried out on the mPTP online server (http://mptp.h-its.org) and the bPTP online server (http://species.h-its.org). No changeable settings are present on the mPTP online server, however, bPTP analyses were run using 10⁴ MCMC generations with a burn-in of 0.1. The COI dataset was uploaded on the ABGD online platform and was analyzed using preset parameters (https://bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html).

Specimens of P. parasitica from the Rogue River, Oregon (n = 62) were virtually identical to each other (99.8–100% similar) based on the genetic distances and included two haplotypes. Uncorrected p-distances between COI sequences are given in Suppl. material 1. The less dominant haplotype was only one base pair different in the COI sequence of 16 out of the 62 leeches from Oregon and this difference was determined to be a synonymous substitution. These 16 leeches came from collection sites throughout the sampled area of the Lower Rogue River. All of the P. parasitica sequences (native and non-native regions) were 91.2–100% similar to one another. The largest genetic distance of 9.1% was found between a sequence from Algonquin Provincial Park, Ontario, Canada and the samples from Wayne County, Tennessee. The largest genetic distance of 8.8% was found between samples from Oregon and the samples from Wayne County, Tennessee when comparing the samples from Oregon to the P. parasitica samples from other localities.

The haplotype network contained 37 unique haplotypes from 45 localities of P. parasitica from introduced and native regions (Fig. 2). Sequences used to generate the haplotype network (n = 117) had 122 segregating sites, high nucleotide diversity (π = 0.041), and evidence indicating balancing selection maintaining variation or a recent population contraction (Tajima’s D = 0.07). A total of 35 haplotypes were present from native regions. The most common haplotype included in our analysis from outside of Oregon was found in seven samples from Illinois, Ohio, Nebraska and Algonquin Provincial Park, Ontario, Canada, and the second most common haplotype was found in five samples from Connecticut, Massachusetts, New Hampshire, and Rhode Island.

Figure 2. 

A TCS haplotype network for COI of Placobdella parasitica collected from 45 unique localities across North America.

We found very little genetic diversity in P. parasitica sequences from Oregon. All of the leeches from Oregon belonged to one of two unique haplotypes. Neither of these haplotypes were exhibited by any of the native samples, although there were only 1–2 base pairs difference between the samples from Oregon and the sample from Missouri (Fig. 2). Samples from the Midwestern part of North America (oranges, pinks, and reds) were most similar to one another, while samples from localities east of the Appalachian Mountains (purples, blues) had more changes between nodes (Fig. 3; Suppl. material 2). Samples from Alabama and Mississippi (yellows) had a large number of changes between them and samples from other regions. Samples from the region of the Appalachian Mountains (Braxton County and Monroe County, West Virginia and Wayne County, Tennessee) were more similar to samples from eastern localities than those from localities west of the Appalachian Mountains. The sample from Chittenden County, Vermont was more similar to samples from the Midwest than to samples from New England or other eastern localities.

Figure 3. 

Map of North America showing the collection sites of Placobdella parasitica as circles. Colors and patterns of the circles match those in the haplotype network (Fig. 2). The historical distribution of P. parasitica based on Klemm (1985) and Richardson et al. (2020) is shaded in gray. Solid lines represent groupings from Fig. 3. The dashed line represents the invasion of P. parasitica into the Rogue River, Oregon.

The complete dataset analyzed included 61 representatives, 58 members of Placobdella parasitica as well as three members of Helobdella that served as outgroups, and a total of 658 aligned base pairs for COI. The topology was similar to that recovered in Richardson et al. (2020) and the numbered clades follow the numbering convention established in that study (Fig. 4). All analyses recovered P. parasitica as monophyletic (SBS = 100) and the same nine groups defined in Richardson et al. (2020) were recovered, albeit with many unsupported nodes. The representative from the Rogue River, Oregon placed closest to representatives from Missouri and Lincoln County, Arkansas and within a larger clade of closely related members from the central USA and Ontario, Canada (Groups 1a, b, c). Group 1a included representatives from the type locality of P. parasitica in Lily Lake, Minnesota. Sister to Group 1 is the sample from Braxton County, West Virginia (Group 2). Samples from Alabama and Mississippi formed two reciprocally monophyletic clades (Groups 3a, b) that were sister to Groups 1 and 2. Samples from the northeastern USA formed two groups, the first including those from Massachusetts, New York, and Vermont (Group 4; SBS = 81) and the second that included samples from Connecticut, Massachusetts, New Hampshire, and Rhode Island (Group 5a; SBS = 92). Sister to Group 4 was a clade made up of samples from North Carolina and Monroe County, West Virginia (Group 6; SBS = 97). Sister to Groups 4, 5a, and 6 was the sample from Maryland (Group 5b; SBS = 92). The larger clade formed by Groups 4, 5, and 6 was sister to the larger clade formed by Groups 1, 2, and 3, although this node lacked support. Samples from Florida (Group 9; SBS = 100) were monophyletic and were recovered as sister to the sample from South Carolina (Group 7). Samples from Tennessee formed a monophyletic clade (Group 8; SBS = 100) that was sister to all other ingroup taxa. Samples new to this study included two additional representatives from Florida that were identical to the other samples from Florida (SBS = 100), and the representatives from Maryland, New Hampshire, and Chittenden County, Vermont. The representative from New Hampshire placed within the clade of other representatives from New England (Group 5). The representative from Maryland placed outside of the subclades that includes representatives from New England, Monroe County, West Virginia, New York, and Bennington County, Vermont (Groups 4, 5a, and 6). The representative from Chittenden County, Vermont placed within the large subclade of closely related members from the central USA and Ontario, Canada (Group 1a).

Figure 4. 

A maximum likelihood phylogenetic tree of the COI dataset for Placobdella parasitica. Support values are standard bootstrap values. Values of <70 are not shown. The full tree with outgroups is available as Suppl. material 2.

Delimitation results were based on comparisons of COI sequence data from 46 specimens originating from across central and eastern North America, southwestern Oregon, and Ontario, Canada (Fig. 3). Our findings were based on phylogenetic entities using the most conservative delimitation estimates in common among the employed methods. Results of delimitation analyses for identifying phylogenetic entities among the COI dataset showed that our sampling minimally consisted of 13 separate phylogenetic entities (Table 4). The ABGD analysis recovered 13 distinct phylogenetic entities. The GMYC analysis also indicated 13 entities, although the result was not statistically significant (p = 0.0712). Analyses using mPTP with BEAST trees estimated 13 phylogenetic entities, while bPTP analyses estimated 13–22 phylogenetic entities. The largest entity was consistent with the large subclade that has a broad geographic distribution across the central USA and Ontario, Canada. This entity also included the sample from the Rogue River, Oregon. The second largest entity included samples from New England (Connecticut, Massachusetts, New Hampshire, and Rhode Island). Other defined entities included specimens from more restricted geographic regions or were made up of a single specimen (e.g. samples from Braxton County, West Virginia, Maryland, and South Carolina).

All of the samples of P. parasitica from the Rogue River, Oregon were nearly identical and included two haplotypes that differed by one base pair that was a synonymous substitution. The less common haplotype was found at collection sites (Battle Bar, Clay Hill Rapids, Doe/Fawn/Tyee, and Haas Island) located throughout the sampled area of the Rogue River that also had individuals with the dominant haplotype. We were unable to determine if the initial introduction occurred within, upstream, or downstream of the sampled area. Low genetic diversity within the introduced population compared to the high haplotype diversity in the native region can be attributed to the Founder Effect. Similar observations of extremely low levels of genetic variation within introduced populations of Helobdella europaea Kutschera, 1987 have been reported in Spain and Ukraine (Reyes-Prieto et al. 2014; Morhun et al. 2020). Low genetic diversity suggests the leeches from the Rogue River all came from the same source population, likely by a single introduction event of a single or few related individuals. There are a couple of possibilities for how this could have happened. First, more than one adult leech with the two haplotypes represented was introduced. Second, one or more leeches with the same haplotype were introduced and a mutation in the COI sequence developed since the introduction (i.e. genetic drift). It is unlikely that only eggs without the parent leech were the source of the introduction since leech eggs do not survive after being separated from the leech parent. It is possible that a single individual leech was introduced, although only a small number of leech species have been shown experimentally to be capable of self-fertilization, such as B. weberi (Govedich et al. 2003; Sawyer 2020). Self-fertilization has not been reported in P. parasitica, although it has been reported for other glossiphoniid species (Iyer et al. 2019). Despite this, a single individual brooding eggs, carrying hatchlings, or that was gravid could have been the founder of this introduced population.

None of the haplotypes from Oregon were 100% identical to any of the haplotypes in our analysis from the native range suggesting that the exact source population was not represented in our analysis. The representative from Oregon placed within the large clade (Clade 1) from the central USA and Ontario, Canada (Fig. 3). Within Clade 1, the representative from Oregon placed in a subclade with samples from Missouri and northern Arkansas and was closest to the sample from Missouri in the haplotype network (Fig. 2). While our analyses could not pinpoint the precise source population of the introduced P. parasitica in the Rogue River, it is most likely near the region of southern Missouri and northern Arkansas. More fine-scale sampling across the geographic distribution of P. parasitica, within the Missouri and Arkansas region in particular, could result in collecting specimens with a haplotype matching the samples from the introduced population in the Rogue River.

Our samples included specimens of all size classes including juveniles, adults, and adults with hatchlings suggesting that this is an established breeding population of P. parasitica in the Rogue River. The short branch lengths of the representative from Oregon in the phylogeny suggests that this is a recent introduction, most likely human-mediated, and there has been limited opportunity to accumulate new haplotypes through genetic drift and subsequent admixture. The high similarity of haplotypes across the samples from the Rogue River adds further support for the assumption made by Reilly et al. (2023) that the introduction of P. parasitica in this section of the Rogue River occurred within the past 20 years, which was based on the absence of any leech observations over the course of four years of sampling more than 200 turtles in the early 2000s (Galea Wildlife Consulting, unpublished reports). The low number and high similarity of haplotypes in the Rogue River suggests this likely is not an ongoing process of introduction. Broader sampling of the river system and watersheds in the Pacific Northwest is needed to determine the extent of the introduced range of P. parasitica, to search for additional haplotypes, and to determine if the leech invasion is expanding.

The introduction of leeches to regions outside of their native geographic distributions is not an uncommon occurrence and usually is unintentionally human-mediated. In most cases, the exact mode of introduction and timing is unknown, as is the case of P. parasitica in the Rogue River. In this study, high diversity between sampled areas within the native region suggests that members of P. parasitica have naturally low dispersal capability. The primary explanation for introductions of Placobdella species has been presumed to be via escaped or abandoned pet turtles. Placobdella ornata (Verrill, 1872), a North American species, was found in Belgium and was thought to have been introduced to the region while attached to a turtle host (Soors et al. 2015). Kvist et al. (2018) reported a specimen of P. parasitica found in a waterway near the Panama Canal and suggested the most likely scenario was that the leech was introduced by humans transporting pet turtles. Chelydra serpentina (Linnaeus, 1758), the Common Snapping Turtle, and T. s. elegans are both very common in the pet trade, are known to host P. parasitica, and are thought to have been introduced to Oregon via exotic pets that were abandoned by their owners. While T. s. elegans was not detected in the Rogue River study area of Reilly et al. (2023) despite multiple surveys, https://iNaturalist.org (accessed 12 September 2024) reports research-grade observations of T. s. elegans within the Rogue River watershed within 50 km (30 miles) upstream from the sampled area in connected waterways and adjacent ponds, including in Grants Pass (2022), Denman Wildlife Area (2016), Bear Creek (2019, 2020, 2023), and Ashland Creek (2018, 2022). If introduced T. s. elegans also introduced P. parasitica, the leeches drifting downstream to the section of the Lower Rogue River sampled in Reilly et al. (2023) is plausible.

The second possible mode of leech introductions has been suspected to be by aquatic plants or snails transported as part of the aquarium and garden pond industry. This has contributed to the now widespread distribution of the Asian leech species Barbronia weberi (Blanchard, 1897), documented on all continents except Antarctica (e.g. Moore 1946; Mason 1976; Pamplin and Rocha 2000; Rutter and Klemm 2001; Govedich et al. 2003; Ludányi et al. 2019). Helobdella europaea with South American origins is now widespread in Europe and recorded in New Zealand, South Africa, Australia, Hawaii, and Fiji and has been suspected to have been introduced via commercially-sold aquatic plants or snails (Boisen Bennike 1943; Pfeiffer et al. 2004; Siddall and Budinoff 2005; Reyes-Prieto et al. 2014; Morhun et al. 2020; Rashni et al. 2023). While small glossiphoniid species are frequently attached to leaves and stems of aquatic plants in the horticulture industry, P. parasitica is more frequently found on larger submerged debris, stones, and logs and unlikely to be by-catch in aquatic plant harvests. If P. parasitica was being transported and introduced by commercially-sold aquatic plants or snails, we would expect more haplotype mixing in the native regions. For example, the human-mediated movement of leeches sold as fishing bait in North America has likely contributed to the homogenization of genetic diversity of Macrobdella decora (Say, 1824) and Erpobdella obscura (Verrill, 1872) across the species distributions (Anderson et al. 2020; Kennedy et al. 2024).

Another possible dispersal mechanism could be by watercraft, equipment, or trailers that were improperly cleaned or drained after use in the native regions and then transported to the Pacific Northwest. Prolific commercial and recreational use of jet boats and whitewater rafting occurs on the Lower Rogue River, upstream, downstream, and within the study area of Reilly et al. (2023). Within the Rogue River study area, daily shuttle of rafts for commercial whitewater adventures provides ample opportunities for redistribution of leeches throughout the study area. Although P. parasitica can tolerate a body weight loss of 70.4% from desiccation if refrigerated (Hall 1922), it seems unlikely that leeches could survive a road trip from the central USA over the Rocky Mountains exposed in plant, plant fragments, mud, or other debris attached to a watercraft or trailer. There is potential for the translocation of leeches to the Rogue River through commercial or recreational watercraft use, especially in jet boat engine bilge or ballast water or in damp crevices of a rolled-up raft in a truck or trailer. The abundant lakes, reservoirs, and rivers of southern Missouri and northern Arkansas are popular for jet boat enthusiasts and river rafters, including annual speed boat races along the Missouri-Arkansas border. In particular, the potential for jet boats to be the vector of introduction seems plausible given the propensity for jetboat use in both the source region and the introduced region, and jetboats in particular provide a reservoir of water that would insulate leeches from desiccation because the mechanics of jetboat engines are designed to hold water. State and federal agencies are already monitoring the movement of watercraft to prevent the spread of other invasive aquatic species and pathogens; leeches being transported via these activities may be overlooked at inspection stations, although there are no definitive reports yet.

Another possibility is that the population in the Rogue River could have started with leeches that were released, possibly after being used as educational tools or as fishing bait. Educators aren’t restricted to buying invertebrates as teaching materials from biological or educational companies and usually look for economical vendors because of limited funding. Companies with their own colonies may supply particular species (e.g. “medicinal leeches” that are Hirudo verbana Carena, 1820 or M. decora), but small suppliers or individuals sometimes provide a “grab-bag” of leech species that are seasonally and locally abundant. Educators, students, or their families that are unaware of the consequences of introduced and invasive species to ecosystems may release leeches into their local waterways with the best intentions of preserving life. Whether leeches are being dumped after use in classrooms or discarded as fishing bait, dumped leeches could be the source of non-native introductions.

The introduction pathway of P. parasitica to the Rogue River remains unknown, but of the potential mechanisms described above, non-native turtles or transported watercraft are the most plausible considering the setting and activities associated with the study area. Any turtle species in North America is considered a potential host for P. parasitica meaning that leeches introduced by any pathway have an increased chance of surviving and establishing a breeding population (Moser et al. 2005). Physical removal strategies, such as trapping, will not be discriminatory to just P. parasitica and will pose a threat to native leech species, such as Placobdella kwetlumye Oceguera-Figueroa, Kvist, Watson, Sankar, Overstreet & Siddall, 2010, Placobdella sophieae Oceguera-Figueroa, Kvist, Watson, Sankar, Overstreet & Siddall, 2010, Placobdella burresonae Siddall & Bowerman, 2006, Placobdella montifera Moore, 1906, and other aquatic organisms. Having a better understanding of the dispersal capabilities of P. parasitica and other glossiphoniid leeches will facilitate the detection and monitoring of new introductions, manage persistent ones, and more effectively minimize movement of these leeches between geographic regions.

Richardson et al. (2020) assessed the genetic variation of P. parasitica with phylogenetic analysis of COI sequence data from samples collected from across the geographic distribution. Their phylogeny recovered eastern and western clades divided along the Appalachian Mountains and identified 9 subclades or groups that corresponded to geographic regions in central and eastern North America. They concluded that P. parasitica is a widely distributed and molecularly diverse species, yet morphologically conservative and lacking distinguishing characters that correspond to the defined groups. As expected, the phylogeny and genetic distance data based on our dataset recovered a very similar tree topology to the results of Richardson et al. (2020). This geographic structure mirrored the relationships in the haplotype network. Adding DNA sequence data of another mitochondrial gene and possibly nuclear genes would strengthen confidence in the relationships between the groups.

Differences between the groupings in Richardson et al. (2020) and the 13 entities identified by the species delimitation analyses were in Groups 1 and 3. Group 1 was divided into 3 subgroups: Group 1a included members from the central USA, Chittenden County, Vermont, and Ontario, Canada, Group 1b consisted of members from Louisiana, Oklahoma, and Texas, and Group 1c solely contained samples from Lincoln County, Arkansas. Group 3 was divided into two groups: Group 3a included samples from Mississippi and Mobile County, Alabama, and Group 3b included only samples from Baldwin County, Alabama. Otherwise, the groupings of Richardson et al. (2020) matched the entities defined by the species delimitation analyses. These newly recognized subdivisions are located in the Mississippi River delta and the Mobile Basin, suggesting this is a geographic region with complex diversity. More fine scale collecting of specimens from this area is needed to assess if additional diversity exists and to refine the geographic boundaries of these groups.

There is high genetic diversity between our sample localities for P. parasitica in the native regions suggesting that P. parasitica does not naturally disperse very far. Since essentially any North American turtle species can potentially host P. parasitica, it appears that wild turtles are not a major dispersal mechanism of P. parasitica. Mack et al. (2019) in their study of P. rugosa found low genetic diversity among their samples and proposed host dispersal as a potential driver. They suggested that host dispersal was most likely by repeated instances over short distances since the leeches do not remain on the host very long after feeding. By comparison, P. parasitica has been reported to be seasonally associated with a host and remains attached after feeding (Sawyer 1972). In the haplotype network (Fig. 2), the two most common haplotypes in the central USA and northeastern USA indicate genetic connectivity among P. parasitica individuals in those regions compared to the rest of the sample localities, either by natural dispersal, host dispersal, or human activity. If the movement of turtles through the pet trade is a major factor for the dispersal of P. parasitica, we would expect to see more translocation and homogenization of haplotypes within the native regions. Sampling from additional localities across the native range would fill in sampling gaps and provide a more complete picture of the haplotype diversity and genetic connectivity across the native range of P. parasitica. This more complete mapping of the haplotype diversity across the native range would also allow for a more precise mapping of the source population of the leeches introduced into the Rogue River.