Morphological Comparison of the Chesapeake Logperch Percina bimaculata with the Logperch Percina c. caprodes and Percina c. semifasciata in Pennsylvania

: The Chesapeake logperch, Percina bimaculata (Halderman) has a disjunct distribution when compared to other species in the subgenus Percina . Members of this subgenus in Pennsylvania include Percina caprodes caprodes (Raﬁnesque), Percina caprodes semifasciata (DeKay), and P. bimaculata . Historically the Chesapeake logperch was known only from the Susquehanna River and Potomac River basins. Its range is now restricted to the Susquehanna River below Holtwood Dam and upper Chesapeake Bay. It has been extirpated from the Potomac River and the type locality near Columbia, PA. Attempts are being made to reintroduce it into tributaries of the Susquehanna River near Columbia, PA. We postulate that P. bimaculata diverged from a population of Percina caprodes semifasciata . A naked nape and the fact that both of these species do not occur above the fall line in Pennsylvania support such a relationship.


Introduction
The Chesapeake logperch, Percina bimaculata (Percidae) was described by Haldeman [1] from the Susquehanna River, near Columbia, Lancaster County, PA as Perca (Percina) nebulosa. In 1844, Haldeman [2] described P. bimaculata from the same area of the river. The two were subsequently deemed to be the same species, and were eventually synonymized with the Logperch, Percina caprodes (Rafinesque), which ranges through the Great Lakes-St. Lawrence, Hudson Bay, and Mississippi River systems [3]. Percina c. caprodes was originally described from the Ohio River. Subsequently, DeKaye [4] described Percina c. semifasciata from Lake Champlain, NY and Morris and Page [5] described Percina c. fulvitaenia from the Big Piney River (Gasconade drainage) from Texas County, Missouri. Currently, the Logperch is comprised of three subspecies: P. c. caprodes (Rafinesque), P. c. semifasciata [4], and P. c. fulvitaenia [5]. Near [6], using morphological characters and mitochondrial and nuclear gene sequence data, concluded that the Chesapeake logperch is in fact a distinct species and that it was closely related to the Mobile logperch, Percina kathae, and the

Materials and Methods
We collected all of the fish, except those examined from the Illinois Natural History Survey, using seines, back-pack electrofishing units, boat electrofishing, or electrified benthic trawls [11]. All of the fish were anesthetized with clove oil, euthanized in 1% formalin, preserved in 10% formalin, and placed in permanent storage in 70% ethanol in the Pennsylvania State University Fish Museum. All collections followed the methods approved by the Animal Use and Care Committee at Pennsylvania State University (PROTP201800659).
Eighteen measurements and seven counts were taken for each individual following the procedure of Konings and Stauffer [12]. The morphometric data were taken with digital calipers and measured to the nearest 0.1 mm. All rays of the pectoral fin were counted, including the small splinter on the upper edge of the fin. Lateral-line scales were counted from the anterior to the hypural plate. Pored lateral-line scales posterior to the hypural plate were counted separately. All counts and measurements were taken from the left side of the body except for gill-raker counts, which were taken on the right side.
Percina bimaculata were collected from tributaries of the  (Figure 1).
Analysis of morphometrics and meristics was conducted using sheared principal component analysis (SPCA) and principal component analysis (PCA), respectively, as described by Humphries et al. [13] and Stauffer et al. [14]. Principal component analysis was used to analyze meristic data with the correlation matrix factored. Body shape differences were analyzed using SPCA with the covariance matrix factored. To illustrate differences in the counts and measurements among the species, the sheared second principal components of the morphometric data were plotted against the first principal components of the meristic data. The first sheared principal component of the morphometric data accounted for variation in individual size. Similarly, the sheared second principal components explained the additional variation in shape. An ANOVA, in conjunction with Duncan's multiple range test (p < 0.05) was used to determine if the populations differed along the axis of the sheared second principal component (morphometric data) or the axis of the first principal component of the meristic data. If the clusters were not significantly different along one axis independent of the other, then a MANOVA, in conjunction with a Hotelling-Lawley trace, was used to determine whether the mean multivariate scores of the clusters formed by the minimum polygons of the PCA scores were significantly different (p < 0.05).
principal components of the morphometric data were plotted against the first principal components of the meristic data. The first sheared principal component of the morphometric data accounted for variation in individual size. Similarly, the sheared second principal components explained the additional variation in shape. An ANOVA, in conjunction with Duncan's multiple range test (p < 0.05) was used to determine if the populations differed along the axis of the sheared second principal component (morphometric data) or the axis of the first principal component of the meristic data. If the clusters were not significantly different along one axis independent of the other, then a MANOVA, in conjunction with a Hotelling-Lawley trace, was used to determine whether the mean multivariate scores of the clusters formed by the minimum polygons of the PCA scores were significantly different (p < 0.05).

Results
The morphometric and meristic data for P. bimaculata captured in Octoraro Creek and the Susquehanna River are summarized in Table 1. A plot of the SHRD PC2 of the morphometric data against the PC1 of the meristic data illustrates that the minimum polygon clusters are coincident ( Figure 2). A MANOVA showed that these clusters were not significantly different (p > 0.05).
The morphometric and meristic data for P. c. semifasciata captured in Lake Erie drainage in Pennsylvania and the Mississippi River drainage in Illinois and Minnesota are summarized in Table 2, and P. c. caprodes from the Allegheny River and Genesee River are summarized in Table 3.

Results
The morphometric and meristic data for P. bimaculata captured in Octoraro Creek and the Susquehanna River are summarized in Table 1. A plot of the SHRD PC2 of the morphometric data against the PC1 of the meristic data illustrates that the minimum polygon clusters are coincident ( Figure 2). A MANOVA showed that these clusters were not significantly different (p > 0.05).     The morphometric and meristic data for P. c. semifasciata captured in Lake Erie drainage in Pennsylvania and the Mississippi River drainage in Illinois and Minnesota are summarized in Table 2, and P. c. caprodes from the Allegheny River and Genesee River are summarized in Table 3.   A plot of the SHRD PC2 of the morphometric data against the PC1 of the meristic data illustrates that the minimum polygon clusters for P. bimaculata do not overlap with the minimum polygon clusters of either P. c. caprodes or P. c. semifasciata (Figure 3). Size accounted for 84.3% and the second principal component accounted for 11.0% of the observed variance in the morphometric data. Variables with the highest loadings on the SPC2 components were the distance between the posterior insertion of the second dorsal fin and the insertion of the ventral caudal fin (0.76), the distance between the posterior insertion of the second dorsal fin and the insertion of the pelvic fin (−0.38), and the distance between the posterior insertion of the anal fin and the dorsal insertion of the caudal fin (−0.30). The first principal component of the meristic data accounted for 48% of the total variance. Variables with the highest loadings on the first principal component were pored scales posterior to the hypural plate (0.47), lateral line scales (0.47), and the number of anal fin rays (−0.46).
A plot of the SHRD PC2 of the morphometric data against the PC1 of the meristic data for P. c. semifasciata from Lake Erie, P. c. semifasciata from the Mississippi River, P. c. caprodes from the Allegheny River, and P. c. caprodes from the Genesee River is illustrated in Figure 3. The minimum polygon clusters were all significantly different (p < 0.05) from each other along the SHRD PC2 axis. The populations from the Genesee River and the Allegheny River were not significantly (p > 0.05) different from each other along the PC1 axis of the meristic data ( Figure 4). Moreover, the populations from Lake Erie and the Mississippi River were not significantly different along the PC1 axis. The populations from the Genesee River and Allegheny River were significantly different (p < 0.05) from the populations from Lake Erie and the Mississippi River along PC1 (meristic data) ( of P. c. caprodes is fully scaled, while the napes of P. c. semifasciata and P. bimaculata are naked ( Figure 5). A plot of the SHRD PC2 of the morphometric data against the PC1 of the meristic data for P. c. semifasciata from Lake Erie, P. c. semifasciata from the Mississippi River, P. c. caprodes from the Allegheny River, and P. c. caprodes from the Genesee River is illustrated in Figure 3. The minimum polygon clusters were all significantly different (p < 0.05) from each other along the SHRD PC2 axis. The populations from the Genesee River and the Allegheny River were not significantly (p > 0.05) different from each other along the PC1 axis of the meristic data ( Figure 4). Moreover, the populations from Lake Erie and the Mississippi River were not significantly different along the PC1 axis. The populations from the Genesee River and Allegheny River were significantly different (p < 0.05) from the populations from Lake Erie and the Mississippi River along PC1 (meristic data) ( Figure 4). Size accounted for 85.9% and the second principal component accounted for 3.5% of the observed variance in the morphometric data. Variables with the highest loadings on the SPC2 components were the dorsal fin base length of the second dorsal fin (0.69), the dorsal fin base length of the first dorsal fin (0.33), and the horizontal eye diameter (−0.28). The first principal component of the meristic data accounted for 33% of the total variance. Variables with the highest loadings on the first principal component were lateral line scales (0.65), the number of dorsal fin spines (0.54), and the number of pectoral fin rays (0.47). The nape of P. c. caprodes is fully scaled, while the napes of P. c. semifasciata and P. bimaculata are naked ( Figure 5).

Discussion
Initially, it was thought that the morphology of populations of P. bimaculata from the main channel Susquehanna River might differ from the populations inhabiting Octoraro Creek. No differences in the morphology of these populations were observed. Furthermore, the populations of P. c. caprodes from the Allegheny River and P. bimaculata from the Susquehanna River are clearly different.

Discussion
Initially, it was thought that the morphology of populations of P. bimaculata from the main channel Susquehanna River might differ from the populations inhabiting Octoraro Creek. No differences in the morphology of these populations were observed. Furthermore, the populations of P. c. caprodes from the Allegheny River and P. bimaculata from the Susquehanna River are clearly different.
The Genesee River flows into Lake Ontario; thus, one might expect that they would be more similar to populations from Lake Erie. The Genessee River, however, has outlets to the Allegheny River via the Cuban glacial outlet [10,14], Oswayo Creek, and Honeoye Creek [15]. Connections between the Genesee and Susquehanna rivers included Cowanesque Creek and Pine Creek [16]. The fact that populations of P. caprodes from the upper Genesee River are more closely aligned morphologically with populations in the upper Allegheny River supports this connection. The Potomac River, where P. bimaculata, was native, heads up against the Youghiogheny River (Monongahela River tributary). The Savage River (Potomac River tributary) captured a portion of the Casselman River (Youghiogheny River tributary) [15][16][17][18][19]. Similarly, Gandy Creek (Monongahela tributary via the Cheat River) was captured by the North Fork of the South Branch of the Potomac River [20,21]. It should be noted that our specimens from the Genesee River were collected near Wellsville, New York, which is upriver of a series of waterfalls. These waterfalls have effectively isolated the upper river from fish invading from Lake Ontario.
Historically, Percina nebulosa and P. bimaculata were considered synonyms of P. caprodes. Near [6] elevated P. bimaculata to species status and because P. nebulosa was already occupied determined P. bimaculata to be the available and appropriate specific epithet. Thompson [22] showed the distribution of the populations of P. caprodes (P. bimaculata) in the Susquehanna and Potomac rivers as being completely disjunct from other populations of P. caprodes in Pennsylvania. Based on mtDNA and the S7 intron, Near [6] concluded that P. bimaculata was a sister species to a clade containing Percina austroperca and Percina kathae. Percina kathae is endemic to the Mobile Basin in Mississippi, Alabama, and Georgia [22], while P. austroperca is restricted to the Choctawhatchee and Escambia drainages in Alabama and Florida [23]. In 2011, however, Near [24] reported a phylogeny that aligned P. bimaculata more closely to P. caprodes than to either P. austroperca or P. kathae. It could not be determined if the P. caprodes reported by Near [25] were P. c. caprodes or P. c. semifasciata. Certainly, the native fish of the Susquehanna River system include those from the northern Mississippi River via the Allegheny River, along shorelines of glacial lakes in the developing Laurentian Basin [25], and ancient connections with the Finger Lakes that include P. c. semifasciata [16]. Percina c. semifasciata is known from the upper Mississippi River, the Great Lakes, and the Hudson Bay drainages [26]. There are also the Teays/Monongahela/Potomac connections (see [16]), which would have provided a route to the lower Greater Susquehanna channel during glacial periods. Furthermore, fish (e.g., Enneacanthus gloriosus) [27] may have invaded from the northeast and southeast. Although now probably extinct, Etheostoma sellare and P. bimaculata are the only recent (within the last 100 years) endemic fish on the Atlantic Slope that occur in and north of the Potomac and south of St. Lawrence. Although P. bimaculata and P. c. semifasciata are quite different morphologically (Figure 2), we postulate that P. bimaculata diverged from a population of P. c. semifasciata that originated in the Great Lakes and occupied the Potomac and Susquehanna rivers. Jenkins et al. [28] noted that both Percopsis omiscomaycus (Walbaum) and P. c. semifasciata (now P. bimaculata) reached the Potomac from the north since they are absent in the upper Monongahela system. They further state that a route using the Greater Susquehanna is favored by their occupation of large rivers. The close relationship of P. c. semifasciata and P. bimaculata is supported by the naked nape found in both of these species, the fact that both species do not occur above the fall line in Pennsylvania and may explain why Collette and Knapp [29] regarded P. bimaculata as a synonym of P. c. semifasciata. Knapp [30] considered scalation of the nape to be an important characteristic to separate the races of Etheostoma caeruleum, and Esmond and Stauffer [31] used morphological evidence to support their hypothesis that E. caeruleum was native to the Potomac River. Furthermore, most haplochromine cichlids are described on the basis of morphology alone because of the current lack of detectable fixed genetic differentiation [32,33].

Conclusions
A detailed morphological comparison of P. bimaculata with other species in the subgenus Percina is essential, given that there exists present or threatened destruction, modification, or curtailment of its habitat or range; the inadequacy of existing regulatory mechanisms; and other natural or manmade factors affecting its continued existence as listed in the Federal Register [34]. Efforts led by the Pennsylvania Fish and Boat Commission to reintroduce the Chesapeake logperch into its original habitat require that the populations that are introduced are representative of the races that were native. Certainly, a more detailed study, including molecular data on the systematics of the subgenus Percina is needed. It may be that P. c. semifasciata from the Great Lakes should be elevated to species status. The morphological and habitat (e.g., occurring below the fall line) data presented herein suggest that P. bimaculata is closely aligned with P. c. semifasciata from the Great Lakes.

Institutional Review Board Statement:
The study was approved by the Animal Use and Care Committee at Pennsylvania State University (protocol code PROTP201800659).