Xiphactinus audax Leidy 1870 from the Puskwaskau Formation (Santonian to Campanian) of northwestern Alberta, Canada and the distribution of Xiphactinus in North America

Xiphactinus is one of the largest teleost fish known from the Late Cretaceous of North America, and has been found across much of the Western Interior Basin. Despite extensive Late Cretaceous marine deposits occurring in Alberta, there has previously been only two possible records of Xiphactinus from the province, neither of which has been diagnosable to the species level. We describe here a portion of the lower jaws, including teeth, of Xiphactinus audax from northeast of Grande Prairie, Alberta. The fossil has large, thecodont teeth that are circular in cross section and lack any carinae, and are highly variable in their overall size. This fossil is the first diagnostic material of X . a udax from Alberta, and extends the range of the species by over a thousand kilometres. During the Late Cretaceous, the area the fossil was found in was near the Arctic Circle, and represents an important datapoint within the poorly known, northern portion of the Western Interior Basin.


INTRODUCTION
The ichthyodectid Xiphactinus is the largest teleost fish known from the Late Cretaceous of North America (Bardack 1965;Schwimmer et al. 1997;Shimada et al. 2006). Specimens of Xiphactinus have been found from Cenomanian (Cumbaa et al. 2006) through Campanian (DeMar andBreithaupt 2006) deposits throughout much of the southern Western Interior Basin, from central Saskatchewan (Cumbaa and Tokaryk 1999;Cumbaa et al. 2006) south to Texas (Bardack 1965), as well as Campanian to Maastrichtian beds along the Atlantic coast in New Jersey (Grandstaff et al. 1992), North Carolina, Georgia and Alabama (Schwimmer et al. 1997). Although known from a variety of formations, the most spectacular specimens of Xiphactinus come from the Niobrara Formation of Kansas, where well-preserved, virtually complete skeletons were being excavated as early as the 1870s (Cope 1872;Bardack 1965). Some of these nearly complete specimens are close to 5 m in total length, although isolated elements from even larger individuals are also known (Bardack 1965; Shimada and Everhart 2004). Despite extensive marine deposits in Alberta, to date only two other questionable records of Xiphactinus have been reported from the province. The first record is from "east of Lesser Slave Lake" noted by Bardack (1965). However, he listed no referred material and gave no description of any specimen, with the record consisting only of a personal communication from Wann Langston. A search of likely museum collections where any such fossils would have been accessioned has yielded nothing. The general lack of appropriate-aged fossil bearing deposits from that area, as well as the lack of referable or figured specimens, makes this record dubious. A second Albertan record consists of two isolated scales from the Kaskapau Formation near Watino, Alberta (Wilson and Chalifa 1989), although those authors note that the scales do not conform in every aspect with those of Xiphactinus and so referred the specimens to "cf. Xiphactinus". The specimen we describe here, TMP 1973.011.3081, was first mentioned in an unpublished PhD thesis by Christopher Collom (2001), and later mentioned again by Bell et al. (2014). However, the specimen itself was never 90 described and was only assigned to the genus Xiphactinus. Further study of the specimen has revealed a number of characteristics that allow the assignment of the specimen more precisely to Xiphactinus audax, expanding the range of this species considerably. The specimen itself is notable for its relatively northern location within the Western Interior Seaway, and also confirms and fully documents the presence of Xiphactinus in Alberta.

GEOLOGICAL SETTING
The Puskwaskau Formation is a marine, mudstone-dominated wedge up to 350 m in thickness, that was deposited in relatively shallow waters of the Western Interior Seaway (Hu and Plint 2009). The Puskwaskau was deposited during the Niobrara Cycle (Kauffman 1977), and is Santonian to early Campanian in age (Hu and Plint 2009;Fig. 1). The Puskwaskau is divided into five members, which are, in ascending order, Dowling, Thistle, Hanson, Chungo and Nomad (Stott 1963(Stott , 1967Hu andPlint 2009). Collom (2001) suggested that the Santonian/Campanian boundary is within the upper part of the Hanson Member, placing the Chungo and Nomad members within the lower Campanian.
Although the exact stratigraphic position of the fossil is unknown, the matrix still attached to the fossil is similar in nature to a number of other marine vertebrate specimens found in the area, particularly those discussed by Bell et al. (2014). The majority of the specimens discussed in that paper were inferred to have come from the upper portions of the Puskwaskau Formation, likely the Chungo or Nomad members (Bell et al. 2014). This is further corroborated by the location where TMP 1973.011.3081 was found, on a relatively small, shallow creek that drains a limited area. Geologic maps of the region indicate that the creek drains an area that is mainly part of the upper portions of the Puskwaskau Formation, concordant with the idea that the fossil comes from one of the upper members of the formation. Assuming TMP 1973.011.3081 is from one of these upper members, it would be early Campanian in age.

91
Locality and Horizon: The fossil was likely collected sometime in the 1950s or 1960s, based on the personal collection records of George Robinson, the original collector. The notes associated with the fossil in the TMP collections state that the specimen was "from the Bad Heart River near bridge." However, in Robinson's original collections ledger, he noted that he found the fossil along "Kakut C k near bridge on Wanham Rd" and that it came from the "Bad Heart Sst." Another note he made, beside the entry for several invertebrate specimens collected from the same location, states that the bridge had been replaced by a culvert. His description, and comparison with historical maps of the area and modern maps, shows that the fossil came from exposures near Kakut Creek where the present day Secondary Highway 733 crosses it, at approximately 55.61°N 118.39°W. The exact horizon the fossil came from is unknown, and it was likely found ex situ, as the unprepared portions of matrix still attached appear to have been water worn, likely from sitting in the creek itself for some time. Posterior view. For all line drawings matrix is shaded white, bone is shaded light grey, teeth are shaded dark grey, and cross-hatching indicates areas of exposed tooth root.

92
complete, missing only a small portion at the labial side of the tip, and reaches 5.5 cm above the dentary bone. The dentary in this area had been prepared away so that the tooth root is visible (Fig. 2B). The tooth is clearly thecodont as the root reaches all the way to the ventral margin of the jaw, making the complete tooth just short of 13 cm long. The fourth tooth, visible on both sides, is the smallest present, and quite slender. The almost complete left fourth tooth is 1.2 cm high, and would have been only slightly taller when complete. Only the bases of the fifth and sixth teeth on the left dentary are preserved. On the right dentary, the tip of the fifth tooth is present, and this tooth is 1.8 cm above the margin of the bone. The sixth tooth is almost as large as the anterior teeth, at about 2.8 cm tall. The root of the largest tooth on the left side, where the jaw bone has been removed, can be seen clearly, and it is round in cross section. The tooth bases of the fifth and sixth positions on the left dentary are also round in cross section. None of the preserved teeth possess carinae. Because the jaw is broken at the posterior end, the cross-sectional shape of the two dentaries can be seen. The alveolar surface is much wider than the ventral part of the dentary. The cross section is comma shaped, with the ventral tail of the comma curving medially (Fig. 2C).
Remarks: Ichthyodectes and Gillicus, the two other large Cretaceous ichthyodectid fishes known from North America, both have teeth that are much smaller and of more uniform size than those of Xiphactinus (e.g., Bardack, 1965). Species of Xiphactinus have much larger teeth and display heterodonty in their size. Therefore, this specimen clearly belongs to Xiphactinus. Additionally, Schwimmer et al. (1997) noted the characteristic thecodont dentition of Xiphactinus -this is well exemplified in the Alberta specimen by the tooth exposed by preparation so that the whole root, reaching to the ventral edge of the jaw, is visible. This identification is also supported by the undulating alveolar margin of the jaw (Bardack 1965) and the slight sigmoidal curve of the symphysis; both the alveolar margin and symphysis are relatively straight in the other two genera. Two North American species of Xiphactinus are currently considered valid (see discussion below). The teeth of X. audax lack carinae, while those of X. vetus are somewhat laterally compressed and recurved, and bear an anterior or anterolingual carina and sometimes a posterior carina (Schwimmer et al., 1997). The teeth of the Alberta specimen closely resemble those of X. audax, and therefore, this specimen can be confidently assigned to this species. The relative size of the Albertan fish can be estimated based on the size of the jaw fragment. Bardack (1965) noted that the mean length of the symphysis in X. audax is 8.7 cm, with a range of 6.0-13.85 based on measurements of 81 specimens. The symphysis of the Alberta specimen was about 9.2 cm, indicating the fish that this dentary came from was of an intermediate size, possibly around 3 m in total length. Additional Material: A second specimen (Fig. 3), in the collections of the Canadian Museum of Nature (specimen number CMN 51273), likely came from the same area as TMP 1973.011.3081. The label states it is from Bad Heart, Alberta, collected in 1961 by Robert Cochrane, and identified as cf. Xiphactinus by 'CMS' (which would be C.M. Sternberg). CMN 51273 preserves the symphysis and partial left and right dentaries, along with other elements that could benefit from more preparation. The heterodont, thecodont dentition, sigmoidal curve of the symphysis and the undulating dorsal edge of the dentary confirm the identification of Xiphactinus. The teeth have no carinae and circular cross section (pers. obs.), and so we consider this specimen to also belong to X. audax. The CMN specimen is from a slightly smaller fish than the TMP specimen.

DISCUSSION
The most comprehensive review of Xiphactinus was by Bardack (1965), as part of a larger treatise on the taxonomy of chirocentrid fishes. The understanding of the relationship of chirocentrids and ichthyodectids has changed considerably since that publication [see Nelson (1973) and Patterson and Rosen (1977) for a more thorough review of Ichthyodectiformes taxonomy]; however, Bardack's revisions to the generic taxonomy of Xiphactinus have largely stood unchanged. Bardack (1965) retained four species in the genus Xiphactinus, but did so based on their distribution in time and space rather than for morphological

93
reasons. Xiphactinus gaultinus and Xiphactinus mantelli are known from Europe, while Xiphactinus australis is from Australia. Specimens of these three species are quite fragmentary (Bardack 1965) making them difficult to distinguish from X. audax. All the North American species known at that time were considered synonyms of X. audax by Bardack (1965). New records of Xiphactinus sp. have since been recovered from South America (Carrillo-Briceño et al. 2012), and a separate species of Xiphactinus from North America, X. vetus, was re-erected by Schwimmer et al. (1997). This second species of North American Xiphactinus can be relatively easily distinguished from X. audax by the morphology of the teeth (see Remarks above). The two North American species show a large amount of overlap in their distributions ( Fig. 4; Tab. 1). Xiphactinus audax has been found from northern Alberta through the Western Interior Basin and along the Atlantic Coastal plain. Remains of X. vetus are most commonly found along the Atlantic Coastal plain (Schwimmer et al. 1997), although recent finds in Wyoming have significantly extended its range (DeMar and Breithaupt 2006). The relative rarity of X. vetus in the Western Interior or X. audax along the eastern seaboard may have more to do with the lack of appropriate aged marine deposits, rather than any  Kauffman (1984). In cases where multiple localities are in very close proximity, some localities have been omitted for clarity. For a more comprehensive list of localities, please consult the supplementary materials, as well as Bardack (1965) and Schwimmer et al. (1997). true range boundary (Schwimmer et al. 1997). Although the two species are found within geographically overlapping strata, they do not appear to have co-existed. Definitive remains of X. audax have been recovered from Cenomanian through to early Campanian deposits, with a possible occurrence from the upper Albian of Kansas (Williston 1894), while X. vetus is only known from younger, middle Campanian through to lower Maastrichtian deposits (Schwimmer et al. 1997;DeMar and Breithaupt 2006). Although presently far south of the Arctic Circle, the locality where TMP 1973.011.3081 was found would have been close to 65°N during the early Campanian (Irving et al. 1993;Smith et al. 1994;Bell et al. 2014;Vavrek et al. 2014a).
Excluding the two records of indeterminate Xiphactinus and Xiphactinus-like taxa from Arctic Canada, this represents the northernmost record of X. audax. During an expedition in 1965 to the early Campanian (Hills et al. 1999) Smoking Hills Formation along the Anderson River, Northwest Territories, Russell (1967) found and uncovered the tail and posterior abdominal portion of a very large fish he referred to as ?Xiphactinus. However, due to a number of factors the specimen was not collected, and so its identity cannot be confirmed. A second specimen of a possible Xiphactinus from the Arctic was reported by Hills et al. (1999), from the Kanguk Table 1. Partial list of occurrences of Xiphactinus from North America. Note that several of the Kansas occurrences of X. audax from Table 3 in Bardack (1965) have been omitted due to their close proximity to one another both stratigraphically and geographically. Formation on Ellesmere Island. The specimen was collected but not described, and at this time cannot be relocated, so it is presently unknown even what element(s) is/are represented. This locality was dated as early to middle Campanian based on palynomorphs recovered from the same section. Additional material (in the CMN collections) identified by S.L. Cumbaa as belonging to cf. Xiphactinus and Xiphactinus audax was collected from Cretaceous deposits of Devon Island by the field parties of J. Eberle andK. Chin in 1998 and2003. The geographic location of some of these finds is interesting, as during the Late Cretaceous, these Arctic localities may have actually been closer to Europe than many of the locations within the Western Interior Basin (e.g., see map in Vavrek et al. 2014a). Regardless of the exact taxonomic affiliation of these Arctic finds, Xiphactinus audax still possesses an incredibly large geographic range, stretching for thousands of kilometres within the Western Interior Basin. Although this additional record of Xiphactinus audax adds to our growing knowledge of marine vertebrates from the northern part of the Western Interior Basin (e.g., Cumbaa and Tokaryk 2009;Cumbaa et al. 2006;Cook et al. 2008Cook et al. , 2010Cook et al. , 2012Bell et al. 2014;Vavrek et al. 2014b) the region still remains understudied compared to more southerly regions (Cook et al. 2008, Vavrek et al. 2014b. Further work in northern areas is required so as to better understand the true ranges and biogeography of marine vertebrates during the Late Cretaceous.