INTRODUCTION

1 A recently discovered vertebrate fossil belonging to a porolepiform fish suggests that the Kennebecasis Formation of southern New Brunswick (sensu Bailey and Matthew 1872) is Late Devonian in age. We provide a description of this new specimen identified as Holoptychius sp.indet. and review the existing fossil flora and fauna from the Kennebecasis Formation to reconcile the formation’s current assignment to the Early Carboniferous (Williams et al. 1985; Barr and White 2004). Formation information can be found on-line in the Lexicon Database, Bedrock Nomenclature of New Brunswick. We consider the plant fossils used to date the Kennebecasis Formation as Carboniferous (Bell 1927; 1960) are not from this formation. The rocks yielding plant fossils originate from an overlying formation that may be an unmapped westward extension of the Carboniferous Albert Formation to the Saint John area. The occurrence of Holoptychius represents the only fossil known from the Kennebecasis Formation redbeds. Given an otherwise absence of fossils from the Kennebecasis Formation, age determinations are extremely difficult. Consequently, this discovery carries significant interest for stratigraphers and vertebrate palaeontologists.

2 The presence of Holoptychius in southern New Brunswick suggests the cryptic presence of a classic Late Devonian fish assemblage, in particular suggesting the possibility of finding arthrodire and antiarch placoderms and tristichopterid sarcopterygians (Schultze and Cloutier 1996), though none have been found to date. The vast majority of Devonian tetrapod localities are known to also yield remains of Holoptychius (Table 1 in Clack 2005).

GEOLOGY AND STRATIGRAPHY

3 The Kennebecasis Formation is a poorly dated clastic sedimentary sequence from southern New Brunswick that unconformably overlies Neoproterozoic and Early Paleozoic rocks (Fig. 1). The Kennebecasis fault separates the Kennebecasis Formation from Early Silurian rocks of the Kingston Group to the northwest and a fault separates Neoproterozoic and Cambrian rocks to the southeast (Barr and White 1998, 2004). The Kennebecasis Formation refers to a sequence of conglomerates and coarse-grained sandstones on the shores of the confluence of the Saint John and Kennebecasis Rivers in southern New Brunswick, Canada (Williams et al. 1985). The formation crops out in a northeasterly trend for a distance of about 40 km. It has been dated as either Devonian or Early Carboniferous by various authors since Bailey and Matthew (1872) first used the name Kennebecasis conglomerates to describe the red conglomerate, red sandstone and red shale units found along the shore of the Kennebecasis River. Gesner (1839) provided an early description of these rocks, but the name Kennebecasis Formation was first used by Hayes (1927) who described the unit as exposed from the confluence of the Kennebecasis and Saint John Rivers, east to Hampton, and that it overlaid the Lower Carboniferous Albert Formation. Hayes and Howell (1937) later described it as being partly a lateral equivalent to the Albert Formation and reported that "a gradation from the gray sandstones, grits and shales of the Albert to the brown conglomerates, sandstones and shales of the Kennebecasis formation" could be seen in the vicinity of Moosehorn Creek east of Saint John (Hayes and Howell 1937, p. 106). They suggested the two formations were partly contemporaneous, but deposited in very different environments. However, it seems more likely that Hayes and Howell (1937) were describing the contact of the Albert Formation with the overlying Lower Carboniferous Weldon Formation (McLeod and Johnson 1999) which crops out in this region. The Albert Formation has been mapped as straddling the Devonian/Lower Carboniferous boundary in southeastern New Brunswick and is now considered Tournaisian in age (St. Peter 2003), and part of the Middle Devonian-Lower Carboniferous Horton Group (Williams et al. 1985).

Display large image of Figure 1

4 The relationship of the Kennebecasis Formation to the Memramcook Formation to the east is uncertain. The Memramcook Formation is dated as Fammenian to Tournaisian and may be in part laterally equivalent to the Late Devonian to Early Carboniferous Albert Formation. Recent maps show the Kennebecasis Formation as younger than the Memramcook Formation (Barr and White 2004); however, it may be equivalent to the lowest part of the Memramcook Formation.

5 The specimen described here was found on Kennebecasis Island in the red sandstones of the Kennebecasis Formation (Fig. 1, locality 1). The age of the Kennebecasis Formation (Williams et al. 1985) as described by Hayes (1927) and Hayes and Howell (1937), has been based on plants found in grey to brown siltstone and sandstone (Fig. 1, localities 2–4). The Early Carboniferous age was determined by Bell (1927, 1960) who used the plants to compare it to the Horton Group in Nova Scotia. The most recent geology maps place the Kennebecasis Formation as middle Lower Carboniferous (McLeod et al. 1994; Currie 1997; Barr and White 2004) or as part of a Late Devonian-Carboniferous package (Barr and White 1998; McLeod and Johnson 1999), and it is clear that the age is still uncertain. The discovery of a fossil vertebrate, here identified as Holoptychius sp., suggests a Late Devonian age, approximately equivalent to the Escuminac Formation of Gaspé (Miguasha) and the Perry Formation of southwestern New Brunswick. This age determination disagrees with the Early Carboniferous age based on plants (Bell 1960).

6 We argue below that the plant fossils are not from the Kennebecasis Formation as originally described by Bailey and Matthew (1872). A review of plant and arthropod fossils, our preliminary field studies, and early descriptions of the history of the mapping of the formation suggest that the Kennebecasis Formation, as defined (Bailey and Matthew 1872; Hayes 1927; Alcock 1938; Williams et al. 1985), does not include the plant and arthropod-bearing strata previously described from Kennebecasis Island (Copeland 1957; Bell 1960) and the Milkish Head area (Matthew 19). The fossiliferous outcrops on the north shore of Kennebecasis Island (Fig. 1, locality 2) (Wilson 1914a, b; Bell 1960) and at Polly Sams Point (Fig.1, locality 3) on Milkish Head (NBM collection) that contain a "Lepidodendron" flora (Matthew 19; Wilson 1914a; Bell 1960) and conchostracan remains (Copeland 1957) may be part of the Albert Formation, not currently recognized in the Saint John area. Further mapping will be required to determine the position of the plant-bearing beds. However, we tentatively retain them as Early Carboniferous and predict that the plant-bearing beds are at least equivalent in age to the Albert Formation and overlie red conglomerates and sandstones of the Kennebecasis Formation. Early suggestions by Ells (1906) that the Kennebecasis rocks (the red conglomerate -sandstone typical of the Kennebecasis Formation) are overlain by grey plant-bearing beds of Lower Carboniferous age, would seem to agree with this and will require observations from a direct geologic contact.

7 Almost all of the early work aimed at determining the age of the Kennebecasis Formation was published in annual reports of the Geological Survey of Canada with little detail concerning fossil locations. The most recent Survey map (Currie 1997) does not include the most notable fossil locations represented by GSC site numbers. Preliminary field assessment to discover plant localities described from the early 1900’s suggest these localities are silty grey beds exposed on the north shore of the island (Fig. 1, locality 2) and brown sandstone beds at Polly Sams Point and Sea Dog Cove on Milkish Head (Fig. 1, locality 3). Plant-bearing beds on the north shore of Kennebecasis Island contain abundant plant fragments in a fine-grained rippled unit, typical of the Albert Formation further west near Norton (Falcon-Lang 2004). Similarly, Lepidodendropsis cf. L. corrugata recovered at Polly Sams Point and the flora identified by G.F. Matthew from ‘Milkish’ (New Brunswick Museum collection) are also typical of the Albert Formation. ‘Milkish’ samples likely originated from outcrops on the mainland at Milkish Head or west of Kennebecasis Island (Fig. 1, locality 4). The New Brunswick Museum collections also include a single, partial fish fossil from "Milkish, N.B.". The specimen (NBMG 3085), from a brown sandy sediment, belongs to an actinopterygian cf. Elonichthys, also typical of the Albert Formation fauna at Norton (Miller and McGovern 1997) and Albert Mines (Lambe 1910a). However, the material consists of only a patch of articulated scales and a partial median fin, and so is not very diagnostic. Nevertheless, all these specimens come from a greenish grey siltstone that is consistent with the lithology of Albert Formation beds, and are therefore of dubious provenance in the Kennebecasis Formation.

8 Sedimentary rocks containing Holoptychius comprise red sandstone interbedded with coarse, angular conglomerate typical of the Kennebecasis Formation. So far plants have not been recovered in these sandstone beds, with Holoptychius being the only fossil remains recovered.

FOSSIL FISH FROM THE KENNEBECASIS FORMATION

9 Examination of the red sandstones of the Kennebecasis Formation over the past 20 years by one of us (RFM) has failed to produce any fossil material. Generally considered unfossiliferous, the sandstone and conglomerate units of the Kennebecasis Formation have been largely unexplored for fossils. Field work in 2005 (RFM) resulted in a single locality producing vertebrate remains in a red sandstone interbedded with conglomerate. The specimen is reposited in the New Brunswick Museum palaeontology collection (NBMG).

SYSTEMATIC PALAEONTOLOGY

AGE OF THE KENNEBECASIS FORMATION

16 If we assign the Kennebecasis Formation to the Devonian based on the presence of Holoptychius, what of the flora and fauna previously included in this formation? The age of the rocks has been considered Lower Carboniferous or Devonian based on interpretations of the plants. The Devonian age was a result of work by Matthew (19) who wrongly applied Devonian ages to Carboniferous plants in southern New Brunswick (Falcon-Lang and Miller 20). Plants were first collected during the summer of 1906 by R.W. Ells who reported that concerning the "rocks on Kennebecasis bay, they were found to be overlaid conformably by grey sandstones, grits and shales, which contain flora of Devonian types. A large collection of these plant fossils was made and examined by Dr. G.F. Matthew of Saint John, and their Devonian facies clearly determined. As these grey beds are unquestionably above the Perry dark red conglomerates the Devonian age of these is still more clearly defined." (Ells 1906, p. 132). Matthew (19) noted the plants in his review of new species of Silurian and Devonian plants. Interest in the age of the Kennebecasis rocks had to do with the Devonian versus Late Carboniferous age controversy concerning the nearby outcrops at ‘Fern Ledges’. At the time there was considerable and heated discussion about the age of plant-bearing beds in Saint John. There was significant disagreement about the age of the Late Carboniferous ‘Fern Ledges’ (Falcon-Lang and Miller 20), which was considered Devonian by Matthew and others.

17 After making comparisons to recent work on the Perry Formation by David White, Ells (1906, p. 131) stated "One of the most important of the changes made in the geological work about St. John is the transference of the dark red conglomerates and sandstones of Kennebecasis bay, several miles north of the city, from the Lower Carboniferous to the Devonian" from their previous position published in 1879 (Bailey et al. 1880). Ells (1908) noted the Devonian age of the Kennebecasis rocks again in his report for 19 as he laid out the results of continued mapping in southern New Brunswick.

18 Interest in the Kennebecasis fossils continued for a few years as Summary Reports for the years 1906, 1908 (Wilson 1909) and 1909 (Lambe 1910b; Wilson 1910) list several collections made by Geological Survey of Canada personnel. Henri Ami added 50 fossils from Kennebecasis Island to the Survey collections in 19 (Whiteaves 1908); Robert Ells, Lawrence Lambe and William Wilson added Kennebecasis Island and Milkish Creek area fossils (part of a lot of 2,000 specimens) in 1909. Wilson (1914a) described the plant assemblage found earlier on Kennebecasis Island and identified by Matthew to include Lepidodendron corrugatum, L. gaspianum, L. chemungense, L. sternbergii, L. aculeatum, L. rimosum, and Psilophyton princeps. Although Matthew described the plant assemblage as Devonian, White (p. 408, in Wilson 1914a) who received a small collection to examine, believed that "The plant fragments you sent represent, almost without a doubt, a basal Lower Carboniferous flora...... Dr. Matthew, on the other hand, refers these rocks to the Upper Devonian ". White thought only one species of Lepidodendron could be identified. Wilson (1909, p. 184) described his collecting for 1908 at Kennebecasis Island in "rocks forming the western part of this island are grey sandstones, grits and shales: dipping N. 20°, W. at an angle of 58°. The shale is mostly fine-grained, greenish grey, breaking in places into irregular concoidal pieces. These shaly beds contain many fern stems, cordaites, lepidodendra, fronds of ferns and separate pinules in good state of preservation.". Wilson then spent two weeks at Moosehorn Creek collecting in the Albert Formation. He compared the Kennebecasis Island specimens to those found further east at Moosehorn Creek, a site within the Albert Formation (Falcon-Lang 2004).

19 Our examination of the plant fossil locality on the north side of Kennebecasis Island agrees with the determination by White (in Wilson 1914a) and Bell (1960), that specimens represent an Early Carboniferous assemblage and that the rocks containing plants might be part of the Albert Formation which crops out less than 40 km east of Saint John along the Kennebecasis River valley near Bloomfield (McLeod and Johnson 1999). Bell (1960) listed six plants from Kennebecasis Island in his review of the Horton Group flora; Lepidodendropsis corrugata (Dawson), Nematophyllum sp., Aneimites acadica Dawson, Sphenopteris strigosa Bell, Diplotmema patentissimum (Ettingshausen), and Carpolithus tenellus (Dawson).

20 Hayes (1914) in his report on the geology of Saint John noted collections of the arthropod Estheria cf. E. dawsoni made by Kindle at Milkish Head (Fig.1) and that its range is confined to the Lower Carboniferous. Wilson (1915) in his Palaeobotany Report for 1914 recorded about 100 fossil plants collected by A.O. Hayes from Kennebecasis Island, mostly Lepidodendron corrugatum. The Summary Report for 1915 (Kindle 1916, p. 200) listed as an addition to the Survey invertebrate collections "A crustacean fragment from the Carboniferous formation at Kennebecasis island, N.B." collected by Bell.

21 Hayes and Howell (1937, p. 1) reported Lepidodendron corrugatum and other plant remains from the northwestern shore of Kennebecasis Island which Bell (1927) used to suggest an Early Carboniferous age for the Kennebecasis Formation. In the same report Hayes and Howell (1937) noted Hayes had discovered abundant examples of the conchostracan Estheria dawsoni in a grey shale on the south shore of Milkish Head and on the most southerly point west of Sea Dog Cove (Fig. 1).

22 Copeland (1957) listed two species of concostrachans, "Asmussia" alta (sic) and A. tenella from GSC localities 1684, 3520 and 1684 respectively. GSC 1684 is described as the Lancaster Formation near Milkish Head, east of Saint John and collected by A.O. Hayes, 1913. GSC 3520 is reported to be a cove to the south of Milkish Head, Kennebecasis Bay near Saint John, collected by I.R. Jones, 1914. Copeland (1957) reported both of these as Upper Carboniferous, Cumberland Group, although both localities are in the Kennebecasis Formation as presently mapped. There may have been some error in the field notes for these sites as no Cumberland Group rocks have ever been identified from the Milkish Head area. Copeland (1957) did not list Euestheria (Estheria) dawsoni from either of these localities. Asmusia ranges from the Devonian to the Cretaceous. Asmusia membranacea is known from Miguasha (Martens 1996), occurring on clay bedding planes. Specimens labelled ‘Estheria’ (NBMG 10115-10117) from Milkish are found on thin clay slabs although the source beds of these fossils has not been located. The specimens have not been identified; however, they are not A. membranacea and resemble A. tenella figured by Copeland (1957).

23 This summary of the history of plant identification and dating of the Kennebecasis Formation leads us to believe the plant assemblages initially collected by Ells in 1906 and described by Matthew (19), along with subsequent collections of plants and conchostracans, originate from Lower Carboniferous rocks overlying the Devonian red sandstone and conglomerate of the Kennebecasis Formation (Hayes 1927).

DISCUSSION

24 Holoptychius has a global distribution, known from the eastern United States (Catskill Formation, uppermost Famennian; Thomson 1976; Schultze and Cloutier 1996), Quebec (Escuminac Formation, lower Frasnian; Cloutier and Schultze 1996; Jarvik 1972), the Canadian Arctic (see Kiaer 1915 in Ørvig 1957), Greenland (Jarvik 1972), numerous localities in Europe, particularly the Baltic states, Russia, and Britain, and Australia (Young 1993; Schultze and Cloutier 1996). Since Schultze and Cloutier (1996) this genus has been described from the Frasnian of Colombia (Janvier and Villarroel 2000). The occurrence of definitive remains of Holoptychius in the Kennebecasis Formation lends strong support for the assignment of a Late Devonian age for this formation.

25 Conversely, this occurrence could be interpreted as the first known Carboniferous occurrence of Holoptychius. This conclusion is dubious. The reason an Early Carboniferous age assigned to the Kennebecasis Formation may be incorrect was discussed above. Although Bell (1960) described the lycopsids from Kennebecasis Island as an Early Carboniferous assemblage, it is possible they represent a Late Devonian flora such as that from Red Hill (Cressler and Pfefferkorn 2005) where Holoptychius also occurs. Further work on the plants is required. Holoptychius tends to be a rather abundant and common component of fish faunas, frequently represented by numerous scales. Such scales or any other evidence of Holoptychius is lacking from any of the fossil fish beds in the Albert Formation, which would be in close stratigraphic apposition with the Kennebecasis Formation.

26 Work by Greiner (1977, 1978) seemed to suggest that basal beds of the Albert Formation may have in fact been latest Devonian. This conclusion was initially based on an osteolepi-form that was referred to the genus Latvius (Greiner 1977) known from the Late Devonian of the Baltic and Germany. However, Greiner cites no characters supporting this conclusion and a recent reappraisal of this and newly collected articulated material suggests a megalichthyid affinity, quite typical of many Carboniferous vertebrate faunas. Recent work by the authors, involving intensive collecting in the Albert Formation, shows that this formation is more typical of Carboniferous vertebrate assemblages, with Holoptychius or any porolepiforms entirely lacking. Spore analyses by St. Peter (2003) indicate a Tournaisian (Carboniferous) age for the Albert Formation beds containing the osteolepiform remains. Therefore, we do not see Greiner’s conclusions as related to the problem we have addressed here.

27 The taxonomy of Holoptychius spp. is a subject requiring considerable revision as few apomorphies can be used to diagnose the many species (Cloutier and Schultze 1996). Consequently, we do not attempt to assign this material to any existing species nor do we attempt to establish it as a new taxon within the genus. The material is too incomplete to warrant a comparison with other known species of Holoptychius.

CONCLUSION

28 The Kennebecasis Formation is considered to be Late Devonian, not Early Carboniferous, based on the identification of the porolepiform fish Holoptychius sp. in red sandstone beds typical of the formation. Plant-bearing beds are lithologically distinct from the red sandstone and conglomerate of the Kennebecasis Formation as originally defined. Plant fossils, including Lepidodendron corrugatum and other Lower Carboniferous Horton Group flora, were likely collected above an unmapped contact from rocks equivalent to the Albert Formation. These facts help resolve the confusion surrounding the Middle Palaeozoic stratigraphy in this region. It further implies that the Kennebecasis Formation may be in part equivalent to the well-established Upper Devonian Memramcook Formation; however, we retain the distinction until proper stratigraphic work is accomplished. The presence of Holoptychius opens the possibility for the Kennebecasis Formation to yield additional Late Devonian fishes and possibly even tetrapods.