Early Pliocene Fish Remains from Arctic Canada Support a Pre-Pleistocene Dispersal of Percids (Teleostei: Perciformes)

557 Early Pliocene fish remains from Arctic Canada support a pre-Pleistocene dispersal of percids (Teleostei: Perciformes)

Alison M. Murray, Stephen L. Cumbaa, C. Richard Harington, Gerald R. Smith, and Natalia Rybczynski

Abstract: Percid remains from Pliocene deposits on Ellesmere Island, Arctic Canada, are identified as a species of Sander, similar to the walleye and sauger of North America and the pike–perch of Europe and western Asia. They are named as a new species, Sander teneri. These remains are the most northerly percid elements found to date and suggest the palaeoen- vironment was significantly warmer in the Pliocene than it is currently. The fossil remains show the presence in North America of the family Percidae as well as the genus Sander prior to the Pleistocene, indicating a previously proposed Pleistocene immigration from Europe or Asia can be discounted. These fossils contradict an earlier hypothesis that percids, in particular Sander, crossed from Eurasia to North America in the Pleistocene; instead, the fossils show percids were al- ready in the area by the Pliocene. Re´sume´ : Des restes de percide´s provenant de de´poˆts plioce`nes de l’ıˆle Ellesmere, dans l’Arctique canadien, sont attribue´s a` une espe`ce de Sander semblable au dore´ jaune et au dore´ noir d’Ame´rique du Nord, et au sandre d’Europe et d’Asie oc- cidentale. Le nouveau nom d’espe`ce Sander teneri leur est attribue´. Ces restes constituent les e´le´ments de percide´s les plus nordiques trouve´sa` ce jour et portent a` croire que le pale´omilieu e´tait sensiblement plus chaud au Plioce`ne qu’il ne l’est actuellement. Les restes fossiles de´montrent que la famille des Percide´sdemeˆme que le genre Sander e´taient pre´sents en Ame´rique du Nord avant le Ple´istoce`ne, ce qui invalide la the`se ante´rieurement propose´e d’une immigration ple´istoce`ne provenant d’Europe ou d’Asie. L’existence de ces fossiles vient contredire une hypothe`se ante´rieure voulant que les per- cide´s, et plus particulie`rement Sander, soient arrive´s en Ame´rique du Nord en provenance de l’Eurasie au Ple´istoce`ne. Les fossiles montrent plutoˆt que les percide´se´taient de´ja` pre´sents en Ame´rique du Nord durant le Plioce`ne. [Traduit par la Re´daction]

Introduction 2001; Tedford and Harington 2003). However, the beaver remains belong to the extinct Dipoides, and further evidence A unique locality in the Canadian Arctic has produced a is lacking to indicate this genus was a dam-builder (Rybc- number of fossil remains of a small perciform fish. The lo- zynski 2008). cality, south of the head of Strathcona Fiord on Ellesmere Island in Nunavut (Fig. 1), preserves as peat the organic Although it is possible that the small beaver periodically sediment that accumulated in a pond in the early Pliocene dammed the stream, allowing a small lake or pond to form, (Harington 2001; Tedford and Harington 2003). Named the the water apparently was occasionally drained by a break in Beaver Pond site, the sediments are possibly the result of the dam (C.R. Harington, personal observation, 2006). Many beaver activity, as an accumulation of intertwined beaver- specimens are relatively complete and in a fine state of pres- cut sticks, mainly aligned northwest–southeast in a matrix ervation. The perciform fish probably would not have lived in of silty sand and cobbles, might represent part of a beaver bog conditions but could have been in the body of freshwater dam and beaver remains including a partial skeleton have that occupied the area for relatively long periods, which are been recovered from the site (C.R. Harington, field notes, marked by silty stringers between peat layers in the stratigra- phy (C.R. Harington, personal observation, 2006). The Beaver Pond site, currently thought to date between 5 Received 23 March 2009. Accepted 31 July 2009. Published on and 4 Ma (Tedford and Harington 2003), preserves remains the NRC Research Press Web site at cjes.nrc.ca on of mammals that indicate an interchange of fauna between 16 September 2009. Asia and North America that probably occurred prior to Paper handled by Associate Editor H.-D. Sues. *5 million years ago (Tedford and Harington 2003; Daw- son and Harington 2007). This locality, above 788N latitude, A.M. Murray.1 Department of Biological Sciences, University is the only known North American Pliocene vertebrate site of Alberta, Edmonton, AB T6G 2E9, Canada. north of 558N latitude (Dawson and Harington 2007). S.L. Cumbaa, C.R. Harington, and N. Rybczynski. Canadian Museum of Nature, P.O. Box 3443, Station D, Ottawa, ON K1P 6P4, Canada. Geological setting G.R. Smith. Fish Division, Museum of Zoology, University of The Beaver Pond site at 78831’N, 82822’W, at an elevation Michigan, Ann Arbor, MI 48109-1079, USA. of 378 m, is located south of the head of Strathcona Fiord. 1Corresponding author (e-mail: [email protected]). This is one of many late Tertiary fossiliferous sites on Elles-

Can. J. Earth Sci. 46: 557–570 (2009) doi:10.1139/E09-037 Published by NRC Research Press 558 Can. J. Earth Sci. Vol. 46, 2009

Fig. 1. Location of the Beaver Pond site marked with a star. (A) Map of the lands around the Arctic Circle with Ellesmere Island labelled; (B) Ellesmere Island showing the location of Bay Fiord; (C) location of the Beaver Pond site at the southeast end of Bay Fiord.

mere Island referred to as the ‘‘high terrace sequence.’’ Typi- associated with one another or the 1992 specimen, and at cally found high on valley sides, these unconsolidated depos- least one (a cleithrum) may be a distinct species from the its unconformably overlie the Late Cretaceous to early 1992 individual. However, all the bones recovered belong Tertiary Eureka Sound Group. They are capped with glacial to perciform fishes. deposits and were presumably deposited prior to significant landscape downcutting (Mathews and Fyles 2000). The Bea- Comparisons with other species ver Pond site occurs within a high terrace deposit that is an The most diagnostic fossil elements from Strathcona *30–40 m thick unit of mostly sands and gravels. The site Fiord, particularly the cleithra and anguloarticular itself appears to be a slump block (see Mathews and Ovenden bones, were compared with a diversity of perciform species, 1990) derived from a peat layer that occurs *5 m below the including those that currently inhabit freshwaters of Canada capping glacial till. The elevation of the hill crest above the and the United States, as well as perciforms found in eastern site is *402 m. The fossil beds and associated beds above Asia (Siberia) and Europe, and additional marine or fresh and below the locality are all freshwater, as indicated by sili- water taxa (e.g., Lates) that are morphologically similar. ceous microfossils (Harington 2001), with no indication of in- Most of the comparative material examined (Appendix A) terbedded marine sequences in the vicinity. consists of prepared, dried skeletons with a few exceptions Discovered in 1961, the site was not intensively excavated being cleared and stained material or fossil material as noted until 1992. For the following 10 years, most collection ef- in Appendix A. forts were focussed on the northeast end of the site, where the mossy peat unit was originally *2.4 m thick. The Bea- Estimation of body size ver Pond site as currently excavated (in 2006) is 7 m across To estimate the body size of the fossil specimen, linear re- section facing towards the southeast. The stratigraphy, as de- gression was performed using the approach of Casteel (1976). termined in 2006 (Fig. 2A), is variable across section, indi- We used log-base 10 anterior centrum width regressed against cating a complex depositional history. The fish remains log-base 10 standard length on the second anterior vertebra in reported here were recovered from the peat unit at the north- 10 specimens of Sander canadensis of known size in the col- east end of the site (Fig. 2B). lections of the Canadian Museum of Nature.

Material and methods Institutional abbreviations CMN, Canadian Museum of Nature, Ottawa, Ontario; Fossil material FMNH, Field Museum of Natural History, Chicago, Illinois; The perciform elements were collected by C.R. Harington ROM, Royal Ontario Museum, Toronto, Ontario; UAMZ, and colleagues in 1992 and 1995. The elements collected in University of Alberta Museum of Zoology, Edmonton, Al- 1992 (CMN 53270), discovered by John Tener, were all as- berta; UMMZ, University of Michigan Museum of Zoology, sociated in a single, small bolus of fibrous peat located Ann Arbor, Michigan. within 20 cm of the base of the a peat layer, some 60 cm from the northeastern end of the site, and seemed to repre- sent a small individual fish identified in the field as ‘‘Per- Systematic palaeontology cidae?’’ (C.R. Harington, field notes, 1992). Other elements Superorder Acanthopterygii (sensu Johnson and Patterson, collected in 1995 (CMN 52226, 52230, and 53271) are not 1993)

Published by NRC Research Press Murray et al. 559

Fig. 2. (A) Stratigraphy of the Beaver Pond site excavation as it appeared at the end of the 2006 field season. The right side of the section is toward the northeast. The star shows the estimated location of the type specimen (Canadian Museum of Nature (CMN) 53270) of the new taxon collected in 1992. (B) Photograph of the site in 1993. The fossil fish remains were collected from the area indicated by the arrow.

Series Percomorpha Rosen, 1973 vitreus has 0–6 spines, usually none, which are weaker than Order Perciformes Bleeker, 1859 in S. teneri and S. canadensis; and S. volgensis has 6–7 Family Percidae Cuvier, 1817 spines); the dorsal process of the cleithrum being inclined at Genus Sander Oken, 1817 a shallower angle rather than more vertically in the other spe- Sander teneri n. sp. cies; the flanges on the posterior edge of the dorsal plate of the cleithrum being placed more dorsally and being more ro- HOLOTYPE: CMN 53270, associated but disarticulated cranial bust than in the other species; the mandibular sensory canal of and postcranial elements. the anguloarticular long, wide, and of uniform width along its length, not tapered, extending under posterior 0.6 of condyle LOCALITY AND HORIZON: The Beaver Pond site, Strathcona at a moderately steep angle, surface under canal pore pitted, Fiord, Ellesmere Island, Canada; 78833’N, 82822’W; 5– surface adjoining retroarticular vertical and angular, not gen- 4 Ma (Pliocene). tly curved; parasphenoid with a broad mid-ventral keel. DIAGNOSIS: Sander teneri is diagnosed by the presence of four or fewer moderately strong spines on the posterodorsal mar- ETYMOLOGY: the specific epithet is in honour of Dr. John S. gin of the cleithrum as in most S. canadensis and S. lucius (S. Tener, pioneering Arctic biologist who discovered the speci-

Published by NRC Research Press 560 Can. J. Earth Sci. Vol. 46, 2009 men and spent eight field seasons collecting at the Beaver Fig. 3. Elements of the holotype of Sander teneri n. sp. (A) Head Pond site. of left maxilla; (B) possible partial left palatine. Scale bars equal 2 mm. Description

Jaws, suspensorium, and hyoid bones Several partial elements can be identified as jaw or palate elements although they do not seem to be distinctively dif- ferent from those of other perciforms. This includes the ar- ticular head of a left maxilla and a partial left palatine (Fig. 3). About two thirds of the mid-portion of a left cera- tohyal and the articular facet of the left quadrate are also in the collection. Elements of the lower jaw appear to be more distinctive, based on the posterior portion of a left anguloarticular bone, without the retroarticular (Fig. 4). This element measures just under 6 mm long, with the facet being *3.9 mm long anteroposteriorly. The articular facet is broad and shallow. The angle formed by the dorsal and ventral edges anterior to the facet is *688. There is a foramen in the angle be- tween dorsal and ventral edges. In lateral view, there is a deep, open groove for the mandibular sensory canal that ex- tends from almost the posterior end or the bone anteriorly to a point under the anterior third of the articular facet. Such a groove is found in a number of perciforms (e.g., Perca, Sander, Ambloplites, Micropterus, Lepomis, Pomoxis, and Morone), but it is most similar to that of Sander. The bone in the area where the retroarticular would fit bears a number of deep pits. In medial view, there is an open trough along the anteroventral edge of the specimen. Parasphenoid There are three portions of a parasphenoid that do not ar- ticulate with each other; the total preserved length is *17.5 mm. The parasphenoid is 2 mm wide along most of the preserved length but widens to 3 mm at the posterior that is positioned centrally on the base but shifts to a slight end. The segment anterior to the area for attachment for the left lateral position higher on the shaft. On the posterior sur- branchial apparatus has a broad mid-ventral keel (Fig. 5), face near the base, there are two low, rounded projections somewhat similar to that of S. volgensis, which is not medially (the locking processess of Gayet 1987). Broken present in other Sander species (Fig. 6). parts of several other spines are present but lack diagnostic characters. Pectoral girdle A partial right posttemporal is preserved (Fig. 7) and sim- Vertebrae ilar to those of Perca and Sander. The preserved dorsal part of a right supracleithrum is also similar to other perciforms Six vertebral centra were recovered from the Beaver Pond and does not seem distinctive. However, the left and right site, two anterior and four posterior. One is from the anterior partial cleithra present in the associated material are distinc- part of the vertebral column (Fig. 10), the other two are tive to this fish. The dorsal portion of a left cleithrum and from a more posterior position. The vertebrae are 3.5–4 mm about half of the dorsal plate of a right cleithrum are present in diameter. All centra are formed of lacy- or open-strut- (Fig. 8). The right cleithrum clearly preserves three flattened type bone, as found in many perciforms. The amphicoelus spines and the base of a fourth, extending from the posterior centra preserve the ‘‘notochordal tube’’ in the dorsal portion edge of the dorsal plate. There are two convex flanges along of the hollow. The centra are subcircular, with the anterior the posterior edge. The dorsal process of the cleithrum is one being more irregular in shape than the other two. long and robust. Scales Fin spine Approximately 30 scales were recovered, four of them A single, nearly complete fin spine (Fig. 9) is included in complete. The scales are square in shape, being 4 mm ante- the elements of the holotype. It is 10.5 mm long and missing roposteriorly and dorsoventrally. There are six strong radii only the distalmost tip. The distal-half of the spine is curved fanning from a posteriorly placed focus to the anterior edge slightly to the left and also posteriorly. The proximal base of of the scale (Fig. 11). The anterior edge thus bears six deep the spine at 2 mm wide is about twice as broad as the rest of notches at the points of the radii creating a strongly scal- the element. In anterior view, it has a low, rounded ridge loped border. The scale focus is positioned just anterior to

Published by NRC Research Press Murray et al. 561

Fig. 4. Photograph of the left partial anguloarticular bone of the Size of individual holotype of Sander teneri n. sp. in (A) dorsal, (B) lateral, and The body shape of the fossil fish cannot be reconstructed (C) medial views. Scale bar equals 1 mm. from the elements that are preserved. Similarly, the total num- ber of vertebral centra in the living individual, which would effect its overall length, cannot be determined. However, to estimate the body size of the fossil specimen, we performed a linear regression (see Casteel 1976) on the fossil vertebral el- ements and compared them with 10 specimens of S. canaden- sis of known size. The regression equation is log-base 10 standard length = 0.7854 log-base 10 centrum width + 1.8981 (r2 = 0.872, p = 0.000). Using the second anterior ver- tebral centrum width value of 2.6 mm of the Beaver Pond fos- sil in this equation resulted in a standard length value of 167.5 mm. As standard length averages *83.5% of the total length in S. canadensis (personal observation), we estimate a total length of *200 mm for the fossil specimen, assuming it had roughly the same proportions as S. canadensis.

Unassociated elements Several additional fish bones were recovered from the Beaver Pond site in 1995 (Fig. 12A–12F). These elements are not associated with each other or with the original 1992 material. They cannot be convincingly assigned to this new species, and some of the elements may not be con-specific. They consist of the following elements: (1) A partial left opercular bone (CMN 53271) preserves only the horizontal spine with the attached facet for ar- ticulation with the hyomandibula. It is likely perciform but has not been identified further. (2) The posterior portion of a right frontal (Figs. 12A, 12B; CMN 53271), *8 mm wide at the posterior end, pre- serves a posterior medial excavation similar to that found in Morone saxatilis (A.M. Murray, personal obser- vation, 2008) The posterior edge of the frontal is straight from the midline, with the lateral edge strongly angled laterally. The lateral edge curves medially in the anterior part of the preserved portion of the frontal. A strong, rounded ridge extends anteriorly across the bone, which is visible in both dorsal and ventral views. (3) Specimen CMN 52230 is a partial cleithrum (Fig. 12C) that preserves convex flanges similar to the cleithra of the associated 1992 material (S. teneri holotype, CMN 53270), but no spines are preserved. This cleithrum also differs in that the dorsal edge of the dorsal plate appears straight in the holotype of S. teneri whereas it is slightly convex in CMN 52230. A second unassociated partial cleithrum (CMN 53271, Fig. 12D) seems more similar to that of the holotype of S. teneri, in that there appear the field of ctenii, separated from it by three circuli. The to be the remains of spines and the dorsal edge of the posterior edge of the scales bears a dorsoventrally elongate dorsal plate is straight. Both cleithra (CMN 52230 and patch of ctenii, which covers approximately one quarter of 53271) are believed to be perciform and may well be- the scale (1 mm at deepest point). The posterior edge of long to S. teneri, as the posterodorsal margin of the clei- this ctenial field is roughly perpendicular to the longitudinal thrum is variable in S. vitreus (Fig. 13) and may have axis of the scale. The ctenii under a dissecting microscope been in the fossil species also. are the transforming ctenii form of Roberts (1993). They (4) Specimen CMN 52226 is a first anal pterygiophore are clearly aligned in rows, not randomly distributed, and (Figs. 12E, 12F). It is missing the proximal tip, part of with 11–12 rows in the middle (deepest) portion of the field. the left flange and part of the anterior flange. There is a There are *42 ctenii along the posterior edge of the scale. deep groove, oriented anterodorsally, between the ante- The ctenii do not bear alae (see Coburn and Gaglione 1992). rior flange and the head of the pterygiophore for the ar- Circuli are also visible and number *90. The scales appear ticulation of the anal fin spine. This element is also to be from a fish at least 7 years old and slow growing. identified only as perciform.

Published by NRC Research Press 562 Can. J. Earth Sci. Vol. 46, 2009

Fig. 5. Photograph and drawing of parasphenoid of the holotype of Sander teneri n. sp. in dorsal (top) and ventral (bottom) views. Scale bar equals 2 mm. dk, thin keel on dorsal surface; vf, facet for articulation of vomer; vk, broad keel on ventral surface.

Fig. 6. Parasphenoid of (A) Sander canadensis, CMN 78-0181; and (B) S. vitreus, CMN Z4184. Upper of each pair in dorsal view, lower of each pair in ventral view.

Fig. 7. Partial right posttemporal bone of the holotype of Sander teneri n. sp. in external (left) and internal (right) views. Scale bar equals 2 mm.

Relationships of Sander teneri n. sp Percidae based on its similarity to both Perca and Sander in the general morphology of the scales, cleithrum, and angu- Familial identification loarticular. In particular, the cleithra in these genera bear The Beaver Pond fish was first identified as belonging to spines of similar shape and size in the same position on the

Published by NRC Research Press Murray et al. 563

Fig. 8. Photograph of the cleithra of the holotype of Sander teneri Fig. 10. Photograph of an anterior vertebra of the holotype of San- n. sp. Arrows indicate three spines and one spine base. (A) Left der teneri n. sp. in (A) anterior, (B) ventral, (C) left lateral, and cleithrum; (B) right cleithrum. Scale bar equals 2 mm. (D) right lateral views. Scale bar equals 2 mm.

Fig. 9. Photograph of the fin spine of the holotype of Sander teneri n. sp. in (A) anterior (B) left lateral, and (C) posterior views. Scale bar equals 2 mm.

quite different shape, having a dorsal plate that is narrow an- teroposteriorly (e.g., Embiotocidae). The anguloarticular bone of the fossil is also similar to that of Perca and Sander in the shallow facet for the quad- rate and the open groove underneath the facet, unlike the deeper facet with no groove found in Embiotocidae and Centrarchidae.

Generic identification Within the Percidae, there are 10 genera with 201 living species, 187 of which are found in North America and the other 14 found in Europe and Asia (Nelson 2006). Relation- ships among the genera are still a matter of debate. The per- cids have been grouped into three subfamilies; one of which, the Etheostomatinae (Ammocrypta, Crystallaria, Etheos- toma, and Percina), contains predominantly very small indi- viduals (Nelson 2006) and seems to form a monophyletic group (e.g., Sloss et al. 2004). The other two subfamilies are the Percinae and Luciopercinae (e.g., Song et al. 1998). The latter is a group of three genera, Sander, Zingel, and Romanichthys; and the former includes Perca and Gymnoce- phalus. Percarina has also been placed in Percinae (e.g., Nelson 2006) but has been excluded from most analyses. Most genera of the family Percidae, excluding introductions, are restricted to either North America or Eurasia with only two genera, Perca and Sander, being shared between the posterior part of the dorsal plate. Most of the other material two land masses. However, Gymnocephalus is found in Si- examined either have no spines on the cleithra (e.g., Cen- beria. Based on distribution alone, these three extant genera, trarchidae, Gerreidae) or the spines are very fine and numer- Perca, Sander, and Gymnocephalus, are the best candidates ous (e.g., Moronidae). The cleithra of other families are of for the Beaver Pond site fish.

Published by NRC Research Press 564 Can. J. Earth Sci. Vol. 46, 2009

Fig. 11. Photograph of the two scales of the holotype of Sander te- rior to posterior edge of the bone. Mioplosus, an Eocene neri n. sp. Scale bar equals 2 mm. percomorph from Wyoming, is similar to all of the species just mentioned but lacks posterior spines or ridges. Zingel has one or two very strong spines. Most species of Sander have four or fewer spines as in the fossil. The fossil cleithra (Figs. 8, 12C, 12D) show variable pos- terior spines as is found in Sander vitreus (Fig. 13). The pos- terior fragment of the right cleithrum of S. teneri shows four well-defined posterior spines. CMN 52230 (Fig. 12D) possi- bly represents a different taxon but bears two strong spines. Most recent specimens of S. vitreus (N = 51) show 0–3 weak posterior spines, but this is variable — several specimens have six or more weak denticulations. Most S. canadensis (N = 11) and S. lucioperca (N = 4) show 2–4 posterior spines whereas the single specimen of S. volgensis we examined has six (right side) or eight (left side) spines. The presence of spines at the posterior edge of the cleithrum is a wide- spread character among percoids; however, the form of the spine, with confluent weak lateral ridges, appears to be unique to species of Perca and Sander, including S. teneri. Anguloarticular: Potentially informative characters for this element include the angle of the coronoid (dorsal) process, morphology of the posterior sensory canal pore, and orna- mentation and angle of the posterodorsal process. The fossil has a high coronoid angle, similar to that of P. flavescens, P. fluviatilis, and S. vitreus. Morone, S. canadensis, S. vol- gensis, and S. lucioperca have a lower coranoid angle; and Gymnocephalus and Zingel have a curved coranoid with low angle. The posterior sensory canal pore is long in S. teneri, ex- tending under the posterior 0.6 of the articular facet for the quadrate, and is longer, relatively wider, and not tapered as in most percids. The canal lies at a moderately steep angle in the fossil. Gymnocephalus and Zingel differ from the fos- The identification of the fossil based on general morphol- sil, Perca spp. and living Sander spp. in having extremely ogy of the bones (as Perca or Sander) and supported by dis- enlarged canals with the adjacent structures reduced to ac- tribution (as Perca, Sander,orGymnocephalus) is reinforced commodate the canals and large pores (as is the case for the by the morphology of the scales. Coburn and Gaglione darters (Etheostominae), which also bear enlarged foramina (1992) determined a number of characters for the scales of for the sensory pores in the bones (Wiley 1992; A.M. Mur- percids and related families. The ctenii aligned in rows (as ray, personal observation, 2008)). In Perca flavescens, the in Percidae and Moronidae) and absence of alae on the cte- posterior canal pore is attenuate posteriorly. P. fluviatilis is nial bases (which are present in Percina, Etheostoma, Am- similar, but the anterior-half of the canal pore is narrow and mocrypta, Crystallaria, Romanichthys, and Zingel, as well slit-like. Morone has a short posterior canal pore, extending as nonpercids; Coburn and Gaglione 1992) indicate the fos- only under the last 0.1 of the articulating surface; it is wid- sil scales belong to Perca, Sander,orGymnocephalus. Iden- est in its center and attenuate posterodorsally. S. volgensis tification as one of these three genera is also supported by has a wide canal pore, but the pore is as long or longer than the strongly scalloped anterior border of the scale and poste- that of the fossil. In S. vitreus, the posterior canal is nar- rior position of the focus. rower anteriorly, and the ridge between articulation suface We further assessed various characters available in the and the canal pore is less rounded than the fossil. S. cana- fossil material to confirm the correct genus for this new spe- densis and S. lucioperca have a wide canal pore, but it is cies. Characters were examined in all Percinae and Lucio- shorter and more posterior and of variable width. However, percinae taxa available to us, as well as nonpercid taxa the canal pore form in some S. canadensis is identical to the (Appendix A). fossil. Cleithrum: Perca flavescens has 12 or more posterior In the fossil, the surface beneath the sensory canal pore is spines on the cleithrum, and P. fluviatilis has about six strongly pitted and the surface adjoining the retroarticular is spines with distinct confluent ridges extending anteroven- vertical below, with a sharp angle, a horizontal edge, and a trally along the surface of the cleithrum. Gymnocephalus short posterodorsal vertical portion. The posterodorsal proc- has a cluster of two to four long, strong ridges, bound to- ess behind the jaw articulation is low and robust but prob- gether and terminating in a weak spine-like process. Morone ably abraded. chrysops bears about six spines, and M. saxitilis about eight In P. flavescens, the lateral surface below the canal is pit- spines, at the ends of prominent surface ridges on the ante- ted as in the fossil, and the surface adjoining the retroarticu-

Published by NRC Research Press Murray et al. 565

Fig. 12. Photograph of the unassociated 1995 elements from the Beaver Pond site. (A, B) partial right frontal bone in (A) dorsal and (B) ventral views, anterior towards bottom of page; (C) cleithrum CMN 52230 in lateral view; (D) cleithrum CMN 53271 in lateral view; (E, F) first anal pterygiophore CMN 52226 in (E) anterior and (F) right lateral views. Scale bars equal 2 mm.

lar is angled and gently concave. In P. fluviatilis, the lateral gle projects similarly to the fossil. The angles in this contact surface of the bone bears fewer pits. The lateral surface of are reduced in S. lucioperca, more similar to Perca. the anguloarticular in Morone is not pitted, and the contacts The fossil is most like Sander species in its bone shape, adjoining the retroarticular angle forward 458 and are con- canal pore shape, and surfaces for adjoining the retroarticu- cave down. lar. In general, the anguloarticular of S. teneri shows a mo- Only a few pits are found on the bone surface below the saic of characters shared with each of the other species of pore in Sander volgensis.InS. vitreus, the surface is not pit- Sander but is most similar to S. canadensis and S. lucioperca. ted but striated. The surface below the canal pore is strongly Parasphenoid: The parasphenoid of S. teneri is long and striated with some pits in both S. canadensis and S. lucio- slender with a short, wider gill-arch attachment portion; the perca. The external surface adjoining the retroarticular of palatal surface bears a distinctive broad midventral keel as S. canadensis is angular as in S. vitreus, but the medial an- well as the sharp dorsal keel found in other percids. The

Published by NRC Research Press 566 Can. J. Earth Sci. Vol. 46, 2009

Fig. 13. Photographs of left and right cleithra in lateral view of several specimens of Sander vitreus and S. canadensis showing variation in posterior spines. (A) S. canadensis, CMN 78-0178; (B) S. canadensis, CMN 78-0182; (C) S. vitreus, CMN Z4184; (D) S. vitreus, CMN 81-0847; (E) S. vitreus, CMN 81-0828. parasphenoid in Perca and Zingel is shorter with no keel ventrally except for the posterior part near the gill-arch at- tachment area. Gymnocephalus has a short, slender para- sphenoid with a rounded double keel midventrally for attachment of eye muscles. The parasphenoid of Morone is also shorter with a keel developed posteriorly. Extant spe- cies of Sander have proportions similar to the fossil, but the ventral keel is not as developed as in S. teneri. The fossil bears the proportions of Sander spp. but is unique among species of Sander in the form of the mid-ventral keel. The features of the cleithrum, anguloarticular, and para- sphenoid bones that we can assess indicate to us that the placement of the new species is within the genus Sander.

Relationships within Sander Several molecular analyses have included more than one species of Sander. These studies (Billington et al. 1991; Faber and Stepien 1998; Sloss et al. 2004) indicate that the two North American species (S. vitreus and S. canadensis) form a monophyletic group that is the sister to the group formed by the two Eurasian species (S. volgensis and S. lu- cioperca — none of the analyses included S. marinus). If these relationships are correct, S. teneri might be expected to be most similar to the two North American species of Sander unless multiple lineages of the genus are present in North America. A more robust test of the relationships of S. teneri is not possible at present, as there are few or no phylogenetic analyses of percids using osteological charac- ters that are relevent to the fossil elements. However, we here suggest that S. teneri is indeed closer to the two North American species of Sander based on the features described previously.

Discussion Palaeoenvironment The Beaver Pond site preserves the fossil-bearing peat de- posits intercalated with sands and gravels. The peat contains fossils that include mostly extant species of mosses and mainly wetland species of vascular plants (Dawson and Har- ington 2007). The preserved plants (larch (dominant), spruce, pine, alder, and birch) indicate an open woodland habitat (Tedford and Harington 2003). Molluscs have also been recovered (Dawson and Harington 2007). Aquatic ver- tebrate fossils (frogs, scaup duck, the small beaver Dipoides, and the fish reported here) have been collected, as well as other mammals (Dawson and Harington 2007). The mam- mals from the site (beaver, rabbit, a small canid, shrew, a cricetid rodent, mustelids, horse, deer, and bear) indicate a mixed North American – Asian fauna (Tedford and Haring- ton 2003) and represent genera that are known from Mio- cene and Pliocene localities further south in North America and Asia (Dawson and Harington 2007). The mammalian fossils for the most part represent extinct species with their

Published by NRC Research Press Murray et al. 567 nearest modern relatives now found considerably farther west by crossing the northern Pacific. An alternate hypothe- south (Ballantyne et al. 2006). However, Dawson and Har- sis suggests the North American fauna at least in part was ington (2007) noted that some fossil mammals at the Beaver derived from the east — invaders crossed the northern At- Pond site also indicate endemism in a region of the High lantic prior to the Eocene to reach North America (e.g., Col- Arctic. lette and Ba˘na˘rescu 1977; Gaudant 1988). Palaeobotanical evidence shows that the climate was Three hypotheses of the origin of Perca and Sander in much warmer in the Pliocene than it is today (Mathews and North America — one Pacific and two Atlantic — have Fyles 2000), and climatic range analysis of the fossil beetle been well summarized by Carney and Dick (2000) although assemblage suggests temperatures 10–15 8C warmer than they were specifically referring to the genus Perca. In es- present with winter lows averaging *–27 8C in the coldest sence, these fish may have arrived via the Bering Isthmus month and summer highs averaging *12 8C (Elias and Mat- in the Pleistocene (Collette and Ba˘na˘rescu 1977), which is thews 2002). Ballantyne et al. (2006) obtained similar tem- the Pacific route; they may have arrived from Europe via a perature estimates from fossil wood. Stable oxygen isotopes North Atlantic land connection during the Oligocene or ear- in combination with annual ring-width of larch from the lier or have been present in both Europe and North America Beaver Pond site yielded a mean annual temperature at the prior to the loss of connection between the two in the early site of between –7.4 and –3.6 8C, *14.2 8C warmer than to- Tertiary; or they may have traversed the North Atlantic dur- day (Ballantyne et al. 2006). A similar temperature range in ing the late Pleistocene via brackish water at the edge of re- North America today is found in mid-latitude areas of Cen- treating glaciers (Cˇ iharˇ 1975; Faber and Stepien 1998). tral Canada and mid to northern areas of the Prairies, as well These alternate hypotheses for the dispersal of the percid as in southern Yukon Territory. genera shared between North America and Europe (Perca The pond itself was likely no more than 3 m deep based and Sander) are supported by different evidence. on analysis of siliceous microfossils (Harington 2003). This A barrier to freshwater dispersal via Beringia is assumed shallow, quiet environment would likely increase the water to have been present prior to *1.5 Ma because of the as- temperature within the pond compared to flowing streams sumed timing of glaciations (during glacial periods, the ice in the surrounding areas. The confined nature and shallow barred these fish from using the Bering Isthmus, whereas depth would have constrained the diversity of fishes in the during interglacials, marine waters barred the route). How- pond particularly if winter or summer hypoxia and restricted ever, Carney and Dick (2000) noted that the Pleistocene fos- movement were factors. There is some evidence, based on sils in North America are much farther south than the dropstones, that ice formed on the pond surface in winter. southern edge of the ice sheets, indicating Perca at least Although there is as yet no further evidence as to whether was already in North America before the Pleistocene. the pond resulted from the construction of a dam by beavers, McPhail and Lindsey (1970) also suggested this and consid- the following is worth noting. Keast and Fox (1990) found a ered both Perca and Sander to have survived glaciation in beaver pond in Ontario to be occupied only by smaller the Mississippi refugium. fishes, including three perciforms: smaller individuals of An early Tertiary Laurasian distribution of Perca (and P. flavescens (yellow perch), Lepomis gibbosus (pumpkin- similarly, Sander) or a dispersal from Europe to North seed), and a percid of small adult size Etheostoma exile America across the North Atlantic indicates that the genus (Iowa darter). Large-bodied perciforms such as walleye must be at least early Oligocene in age, prior to the loss of (S. vitreus) were not reported from the beaver pond. How- the freshwater connection between the two continents ever, the Strathcona Fiord fossil represents a fairly small in- (Carney and Dick 2000). There is an otolith record of Perca dividual. (P. hassiacia) from the early Oligocene of Germany (Weiler 1961), but the fossil skeletal record of Perca, with the oldest Palaeobiogeography known skeletal remains in western Europe of Miocene age Although freshwater perciform fishes have been found in and the oldest North America Perca of Pleistocene age, in- Eocene deposits of North America (e.g., Priscacaridae, Mio- dicated to Carney and Dick (2000) that the Percidae origi- plosidae; Cavender 1998), the earliest members of the fam- nated in Europe (which was separated from Siberia by the ily Percidae in the continent were previously only known Obik Sea from the early mid-Cretaceous through to the Oli- from the Pleistocene with remains identified as Perca, gocene *30 Ma (Carney and Dick 2000)) and from there Sander, Etheostoma, and Percina (Cavender 1998); the old- dispersed into Siberia and then to North America across the est percids elsewhere are probably those skeletal remains North Atlantic. However, this early dispersal of percids is from the Miocene of Siberia (Sytchevskaya 1989; Cavender contradicted by molecular clock evidence, which puts the di- 1998) or otoliths from the Oligocene of Europe (Weiler vergence between Eurasian and North American species of 1961). The diversification of North American Percidae, in- Sander much later than this, at *7–10 Ma (Billington et al. cluding the strong differentiation of P. flavescens and P. flu- 1990, 1991). This age similarly contraindicates a late Pleis- viatilis, suggests that percids have been in North America tocene northern Atlantic dispersal route proposed by Cˇ iharˇ for much of the Cenozoic. (1975) but supports a Miocene dispersal (Billington et al. Cavender (1998) noted that the distribution of fossil and 1991). living percids suggested a Tertiary circumpolar distribution Similarly, Faber and Stepien (1998) estimated divergence of the family with subsequent displacement southwards times from mtDNA genetic distances that supported a dis- caused by the glaciations of the Plio-Pleistocene. He sup- persal of Sander (= Stizostedion in their paper) from Eurasia ported the idea that much of the North American freshwater to North America across the North Pacific (Beringian land fauna was the result of Siberian forms arriving from the bridge) *4 million years ago in the early Pliocene. The ge-

Published by NRC Research Press 568 Can. J. Earth Sci. Vol. 46, 2009 netic differences found in their study indicated that the Eur- ton and Marissa Gilbert, CMN who assisted with Fig. 2. We asian and North American species of Sander diverged are also grateful to John Tener for finding the bones of the þ *4.05 0:50 million years BP, the two North American Pliocene percid. CRH is grateful to the late Dr. John þ species diverged *2.75 0:40 million years BP, and two of G. Fyles for discovering the Beaver Pond site and introduc- the Eurasian species (S. lucioperca and S. volgensis) di- ing him to its potential as a fossil locality in 1992. We thank þ verged *1.8 0:30 million years BP; and populations of R. Mooi, T. Cavender, and H. Sues for helpful reviews and S. vitreus in the Great Lakes and the Ohio River diverged editing that improved the manuscript. This research was sup- þ 0.029 0:005 million years BP. Even with alternate calcula- ported by the Canadian Museum of Nature, the Polar Conti- tions using additional data that supported older divergence nental Shelf Project, and Natural Science and Engineering times, Faber and Stepien (1998) found the divergence be- Research Council of Canada Discovery grant 327448 to tween Eurasian and North American Sander was still placed AMM. in the early Pliocene. They noted that these divergence times support a Pliocene Beringian (North Pacific) dispersal of References Sander from Europe to North America and that this route Ballantyne, A.P., Rybczynski, N., Baker, P.A., Harington, C.R., should have been available during low sea levels, which are and White, D. 2006. Pliocene Arctic temperature constraints believed to have occurred in the middle or late Miocene and from the growth rings and isotopic composition of fossil larch. also in the early Pliocene. Palaeogeography, Palaeoclimatology, Palaeoecology, 242(3–4): Land connections between Siberia and North America are 188–200. doi:10.1016/j.palaeo.2006.05.016. believed to have been present during the Pliocene at 4.8, Billington, N., Hebert, P.D.N., and Ward, R.D. 1990. Allozyme and 3.7, 2.5, and 2 million years ago; and marine waters are be- mitochondrial DNA variation among three species of Stizoste- lieved to have connected the Arctic and Pacific oceans and dion (Percidae): phylogenetic and zoogeographical implications. therefore barred migration of freshwater organisms from Canadian Journal of Fisheries and Aquatic Sciences, 47(6): 4.2–3.0, 2.5, 2.2 Ma (Ogasawara 1998). The proposed con- 1093–1102. doi:10.1139/f90-126. nection at 4.8 Ma is the youngest potential time when the Billington, N., Danzmann, R.G., Hebert, P.D.N., and Ward, R.D. Ellesmere Island percid could have invaded North America. 1991. Phylogenetic relationships among four memebers of Sti- However, Dawson and Harington (2007) indicated that inter- zostedion (Percidae) determined by mitochondrial DNA and change between the two areas and subsequent differentiation allozyme analyses. Journal of Fish Biology, 39(Suppl. A): 251– of organisms occurred prior to this time, likely before the 258. doi:10.1111/j.1095-8649.1991.tb05088.x. 5.5–4.8 Ma opening (now revised to 5.4–5.5 Ma by Gladen- Bleeker, P. 1859. Enumeratio specierum piscium hucusque in Ar- kov et al. 2002) of the Bering Strait. chipelago Indico. Bataviae typis Lagii et Soc. Dr. W. Junk B.V. The most pertinent argument against the Bering Isthmus Publishers, The Hague. Collected Fish Papers of Pieter Bleeker, route for percids has been the modern distribution of these Vol. IX, Paper 10, pp. 1–276. Reprinted 1975. fishes in Eurasia and North America. Both Perca and Carney, J.P., and Dick, T.A. 2000. The historical ecology of yellow perch (Perca flavescens [Mitchill]) and their parasites. Journal Sander are distributed only in the western part of Siberia of Biogeography, 27(6): 1337–1347. doi:10.1046/j.1365-2699. and east of the Rockies in North America, with no modern 2000.00511.x. populations occurring naturally in the Pacific regions ˇ Casteel, R.W. 1976. Fish remains in archaeology and paleo- (McPhail and Lindsey 1970; Ciharˇ 1975). However, the dis- environmental studies. Academic Press, London, UK. covery of the Ellesmere Island fossils reported here at the Cavender, T.M. 1998. Development of the North American Ter- very least demonstrates that some of these fishes had a tiary freshwater fish fauna with a look at parallel trends found much expanded range in the past. in the European record. The Italian Journal of Zoology, The find of a fossil Sander from Ellesmere Island deposits 65(Suppl.): 149–161. doi:10.1080/11250009809386807. *4–5 Ma indicates that migration from Eurasia to North Cˇ iharˇ, J. 1975. Geographical and ecological variability of perch America may have occurred in the Late Tertiary via the Be- [Perca fluviatilis (Linnaeus)] and history of its distribution from ring Isthmus. This small fish clearly indicates that percids Eurasia to North America. Acta Musei Nationalis Pragae, 31: were present in North America by *4 Ma and living in the 57–89. High Arctic. Coburn, M.M., and Gaglione, J.I. 1992. A comparative study of percid scales (Teleostei: Perciformes). Copeia, 1992(4): 986– Acknowledgements 1001. doi:10.2307/1446628. Collette, B.B., and Ba˘na˘rescu, P. 1977. Systematics and zoogeogra- We thank Mary Anne Rogers, Field Museum of Natural phy of the fishes of the family Percidae. Journal of the Fisheries History, and Kevin Seymour, Royal Ontario Museum, for Research Board of Canada, 34: 1450–1463. providing access to specimens in their care. We are particu- Cuvier, G. 1817. Le Re`gne Animal distribue´ d’apre`s son organisa- larly grateful to John Bruner, University of Alberta, for hav- tion pour servir de base a` l’histoire naturelle des animaux et ing accumulated and prepared a wonderful diversity of d’introduction a` l’anatomie compare´e. Les reptiles, les poissons, percid material that made this work possible and generously les mollusques et les anne´lides. Vol. 2. A. Belin, Paris, France. sharing his knowledge of percids with us. Thanks also to Dawson, M.R., and Harington, C.R. 2007. Boreameryx, an unusual Michael Newbrey, University of Alberta, for informative new artiodactyl (Mammalia) from the Pliocene of Arctic Canada discussions. Thanks to Jon Chicoine, Ontario Ministry of and endemism in Arctic fossil mammals. Canadian Journal of Natural Resources, who supplied Eurasian ruffe (Gymnoce- Earth Sciences, 44(5): 585–592. doi:10.1139/E06-111. phalus) specimens, Noel Alfonso, Canadian Museum of Nat- Elias, S.A., and Matthews, J.V., Jr. 2002. Arctic North American ural (CMN) for help with the statistical analysis to seasonal temperatures from the latest Miocene to the early Pleis- determine potential size of the specimen, and Donna Naugh- tocene, based on mutual climatic range analysis of fossil beetle

Published by NRC Research Press Murray et al. 569

assemblages. Canadian Journal of Earth Sciences, 39(6): 911– among Percid fishes as inferred from mitochondrial cytochrome 920. doi:10.1139/e01-096. b DNA sequence data. Molecular Phylogenetics and Evolution, Faber, J.E., and Stepien, C.A. 1998. Tandemly repeated sequences 10(3): 343–353. doi:10.1006/mpev.1998.0542. in the mitochondrial DNA control region and phylogeography of Sytchevskaya, E.K. 1989. Neogene freshwater fish fauna of Mon- the pike–perches Stizostedion. Molecular Phylogenetics and golia. Transactions of the Joint Soviet–Mongolian Paleontologi- Evolution, 10(3): 310–322. doi:10.1006/mpev.1998.0530. cal Expedition, 39: 1–140. [In Russian.] Gaudant, J. 1988. L’ichthyofaune E´ oce´ne de Messel et du Geiseltal Tedford, R.H., and Harington, C.R. 2003. An Arctic mammal fauna (Allemagne): essai d’approche pale´obioge´ographique. Courier from the early Pliocene of North America. Nature, 425(6956): Forschung Institut Senckenberg, 107: 355–367. 388–390. doi:10.1038/nature01892. Gayet, M. 1987. Lower vertebrates from the early–middle Eocene Weiler, W. 1961. Die Fischfauna des unteroligoza¨nen Melanientons Kuldana Formation of Kohat (Pakistan): Holostei and Teleostei. und des Rupeltons in der Hessischen Senke. Notizblatt des Hes- Contributions from the Museum of Paleontology, University of sischen Landesamtes fu¨r Bodenforschung, 89: 44–65. Michigan, 27: 151–168. Wiley, E.O. 1992. Phylogenetic relationships of the Percidae (Tele- Gladenkov, A.Yu., Oleinik, A.E., Marincovich, L., Jr., and Bari- ostei: Perciformes): a preliminary hypothesis. In Systematics, nov, K.B. 2002. A refined age for the earliest opening of Bering historical ecology, and North American freshwater fishes. Edited Strait. Palaeogeography, Palaeoclimatology, Palaeoecology, by R.L. Mayden. Stanford University Press, Stanford, Calif., 183(3–4): 321–328. doi:10.1016/S0031-0182(02)00249-3. pp. 247–267. Harington, C.R. 2001. Life at a 3.5 million-year-old beaver pond in the Canadian Arctic Islands and the modern scene. Meridian, Fall/Winter: 11–13. Appendix A Harington, C.R. 2003. Life at an early Pliocene beaver pond in the Canadian high Arctic. Journal of Vertebrate Paleontology, List of Comparative Material Examined 23(Suppl. to 3): 59A. Centrarchidae Johnson, G.D., and Patterson, C. 1993. Percomorph phylogeny: a Ambloplites rupestris CMN 78-0143 survey of acanthomorphs and a new proposal. Bulletin of Mar- Archoplites interruptus UMMZ 179981 ine Science, 52: 554–626. Lepomis cyanellus FMNH 60009 Keast, A., and Fox, M.G. 1990. Fish community structure, spatial L. gulosus FMNH 10108 distribution and feeding ecology in a beaver pond. Environmen- L. macrochirus tal Biology of Fishes, 27(3): 201–214. doi:10.1007/BF00001673. CMN 78-0194 Mathews, J.V., Jr., and Fyles, J.G. 2000. Late Tertiary plant and Micropterus dolomieui CMN 81-0825 arthropod fossils from the High Terrace Sediments on the Pomoxis nigromaculatus CMN Z4176 Fosheim Peninsula of Ellesmere Island (Northwest Territories, P. sparoides (= P. nigramaculatus) FMNH 83733 District of Franklin). Geological Survey of Canada Bulletin, Centropomidae 529: 295–317. Centropomus undecimalis ROM 1599 Mathews, J.V., Jr., and Ovenden, L.E. 1990. Later Tertiary plant macrofossils from localities in Arctic/Subarctic North America: Embiotocidae a review of the data. Arctic, 43(4): 364–392. Cymatogaster aggregata CMN Z4210 McPhail, J.D., and Lindsey, C.C. 1970. Freshwater fishes of north- Embiotoca lateralis CMN 77-0272 western Canada and Alaska. Bulletin of the Fisheries Research Board of Canada, 173, pp. 1–381. Gerreidae Nelson, J.S. 2006. Fishes of the World. 3rd ed. John Wiley & Sons Gerres nigri CMN 83-0271 Inc., Toronto, Ont. Diapterus auratus CMN 87-0348 Ogasawara, K. 1998. Review and comments on late Neogene cli- matic fluctuations and the intermittence of the Bering Land Gobiidae Bridge. Journal of Asian Earth Sciences, 16(1): 45–48. doi:10. Rhinogobiops nicholsii CMN 69-2435 1016/S0743-9547(97)00042-1. Oken, L. 1817. Cuvier’s und Oken’s zoologien naben einander ges- Latidae tellt. Isis, 8(148): 1179–1185. Lates microlepis UMMZ 199757 Roberts, C.D. 1993. Comparative morphology of spined scales and Mioplosidae their phylogenetic significance in the Teleostei. Bulletin of Mar- Mioplosis labracoides (fossil material) UALVP 24234 ine Science, 52(1): 60–113. Rosen, D.E. 1973. Interrelationships of higher euteleostean fishes. Moronidae In Interrelationships of fishes. Edited by P.H. Greenwood, R.S. Morone americana FMNH 50206; CMN 76-0054 Miles, and C. Patterson. Published for Zoological Journal of the M. chrysops FMNH 63703 Linnean Society. Academic Press, London, UK, pp. 397–513. M. saxatilis FMNH 10120 (as Morone lineatus); UAMZ un- Rybczynski, N. 2008. Woodcutting behavior in beavers (Castor- catalogued idae, Rodentia): estimating ecological performance in a modern and a fossil taxon. Paleobiology, 34(3): 389–402. doi:10.1666/ Mugillidae 06085.1. Mugil cephalus UAMZ uncatalogued; CMN Z-617 Sloss, B.L., Billington, N., and Burr, B.M. 2004. A molecular phy- logeny of the Percidae (Teleostei, Perciformes) based on mito- Percidae chondrial DNA sequence. Molecular Phylogenetics and Etheostoma caeruleum UAMZ uncataloged Evolution, 32(2): 545–562. doi:10.1016/j.ympev.2004.01.011. Gymnocephalus cernuus UAMZ uncatalogued, two speci- Song, C.B., Near, T.J., and Page, L.M. 1998. Phylogenetic relations mens

Published by NRC Research Press 570 Can. J. Earth Sci. Vol. 46, 2009

Perca flavescens FMNH 62324; UAMZ 4821; CMN 87- Sander canadensis UAMZ 6992 and 6993, Alberta, Canada, 0018; various UMMZ assessed for cleithrum spine counts both cleared and stained; CMN 77-0338, 78-0178, 78- Perca fluviatilis UAMZ 6962 cleared and stained; various 0181, 78-0182, all from Ontario, Canada; various UMMZ UMMZ assessed for cleithrum spine counts assessed for cleithrum spine counts Percina caprodes CMN 87-0364 Sander lucioperca UAMZ 6921 and 6922, Tulcea County, Romanichthys valsanicola UAMZ 6785 and 6865, Romania, Sulina, Romania, both cleared and stained; various both cleared and stained UMMZ assessed for cleithrum spine counts Sander vitreus FMNH 72205, John G. Shedd Aquarium; Sander volgensis UAMZ CS 164, Belaton, Tihany CMN 73-0231b, Ontario, Canada; 81-0828, Ontario, Ca- Zingel zingel UAMZ 58.18.1 nada; 81-0847, Manitoba, Canada; CMN Z4184; UAMZ uncatalogued, Alberta, Canada; various UMMZ assessed Sciaenidae Aplodinotus grunniens for cleithrum spine counts CMN 76-198 and 60-462a

Published by NRC Research Press