Strata of the Economy Member (defined in Appendix; Fig. 4G) consist largely of fluvial deposits interbedded with a few significant intervals of aeolian sandstone, which are generally absent in the rest of the Wolfville Formation. The type area at Lower Economy in Colchester County is in a structurally isolated block on the northern side of the Minas Basin. A very similar sequence, including aeolian strata, on the southern shore of the Minas Basin (Leleu et al. 2010) that we place in the Economy Member, is overlain by the middle Wolfville Formation (Evangeline Member) (see Appendix).

Starting in 1966, Donald Baird and his teams prospected coastal exposures of the Economy Member in the Lower Economy area (Fig. 8), recovering dissociated bones and teeth representing a considerable diversity of tetrapods. More recently PEO, Tim Fedak (Fundy Geological Museum) and others have collected additional vertebrate fossils from this section. Unfortunately, most specimens are too fragmentary to permit lower-level taxonomic identification and often even basic anatomical assessment. However, enough material has now been identified to establish that this tetrapod assemblage is distinct from other, younger assemblages in the Newark Supergroup and elsewhere.

Bones occur mostly as isolated clasts in fluvial, calcite- cemented intraformational conglomerate and lithic- clast, pebbly intraformational conglomerate, along with occasional unionoid bivalves (Fig. 8B). These deposits form dune-scale trough cross-beds.

Temnospondyl stem-amphibians from the Economy Member are referable to two groups, Capitosauroidea and Trematosauroidea, distinguishable on the basis of jaw fragments. The symphyseal portion of a large capitosauroid mandible (FGM998GF8) has a blunt anterior margin and a pair of tusks, followed behind by a row of smaller postsymphyseal teeth (Fig. 9A). The latter feature argues against referral of this jaw fragment to Mastodonsauridae, but it is present in certain other capitosauroids such as Parotosuchus (Schoch and Milner 2000). Additional capitosauroid remains include mandibular fragments (e.g., YPM VPPU 021692; Fig. 9B) and a small piece of the left postorbital region of a skull roof (YPM VPPU 023501).

Welles (1993) referred the trematosauroid cranial remains to the slender-snouted lonchorhynchine Cosgriffius, which was first reported from the Wupatki Member of the Moenkopi Formation (of Olenekian age) in Arizona. The remains reported by Welles include an incomplete left mandibular ramus with 78 preserved tooth positions, many of them still with partial tooth crowns (YPM VPPU 021691) (Fig. 9C). There are no diagnostic features to suggest that the material from the Economy Member is referable to Cosgriffius (as proposed by Welles) rather than another lonchorhynchine taxon. A fragment identified by Welles (1993) as a left premaxilla (YPM VPPU 023686) is not clearly identifiable as such. The posterior region of a left mandibular ramus (YPM VPPU 021692) assigned to Cosgriffius by Welles (1993) belongs to a capitosauroid based on the long, transversely broad and concave postglenoid region, the shape of its posterior end, and the ornamentation on its lateral surface (Fig. 9C).

Complete or partial jaws with teeth, as well as a few other bones, document the presence of procolophonid parareptiles. The smallest dentaries hold a few simple conical teeth (e.g., NSM012GF032.007), whereas larger ones (e.g., FGM000GF86) have transversely broad, bicuspid teeth more posteriorly. It is not clear whether this difference indicates the presence of multiple taxa or merely different growth stages of a single form. Li (1983) documented ontogenetic change from conical, sometimes labiolingually compressed tooth crowns to transversely wide bicuspid crowns in the Early Triassic procolophonid Eumetabolodon brachycephalus from Inner Mongolia, China. The neural arch of a dorsal vertebra (YPM VPPU 023500; width across prezygapophyses: 40 mm) probably belongs to a rather large procolophonid based on its ‘swollen’ configuration.

Archosauromorph reptiles are represented by at least four groups. An incomplete cervical vertebra (YPM VPPU 022000) represents the first record of a long-necked tanystropheid from North America (Fig. 10). It most closely resembles anterior to mid-cervical vertebrae of Tanystropheus spp. from the Anisian to Norian of Europe, the Anisian of the Middle East, and the Ladinian or Carnian of China (Wild 1973, 1980; Rieppel et al. 2010). The neural spine forms a low dorsal ridge, and the postzygapophyses extend posteriorly beyond the posterior end of the centrum. The vertebral centrum is distinctly elongated (length: 76 mm, anterior height: c. 7 mm), more so than in more basal tanystropheid taxa (Sennikov 2011), and we thus identify it as cf. Tanystropheus sp.

An isolated tooth crown (NSM012GF032.001) is labiolingually wide (c. 7 mm) and has three distinct cusps (with worn or abraded apices) that are linked by low ridges. As preserved, the central cusp is the smallest and one of the outer cusps is larger than the other two cusps. The tooth is proportionally narrower transversely than teeth of the archosauromorph Trilophosaurus (Spielmann et al. 2008) and most closely resembles teeth of the enigmatic Tricuspisaurus thomasi from Late Triassic (Norian) fissure fillings in Wales. The latter taxon has variously been classified as a trilophosaurid (Robinson 1957) or a procolophonid (Fraser 1985).

A well-preserved left pubis with an attached fragment of the ilium (YPM VPPU 021806) has a distinct pubic ‘apron’ and most likely represents an archosauriform. A partial left mandibular ramus of a small reptile with nine preserved alveoli and teeth (NSM012GF032.002; preserved length: 32 mm) is possibly referable to the Archosauriformes based on the distinctive ornamentation of fine ridges and grooves along the ventrolateral surface of the dentary. The dentary is long and low, and its medial surface is concealed by the splenial, which is separated from the dentary by a long, slender coronoid more posteriorly. A similarly long and splint-like coronoid is present in various archosauriform reptiles, but also in the crocodylomorph Sphenosuchus (Gower 2003). The small teeth have simple conical crowns and show thecodont implantation. There are two possibly pathological swellings on the dorsolateral margin of the jaw behind the tooth row and on the ventromedial edge further posteriorly, respectively.

Finally, tall, slender and apically recurved tooth crowns with serrated cutting edges (e.g., YPM VPPU 019909; 4 serrations per mm) presumably have archosauriform affinities but cannot be assigned to a particular taxon.

Huber et al. (1993) established the Economian Land Vertebrate Faunachron based on the tetrapod remains from the Economy Member of the Wolfville Formation and considered it Anisian in age. The currently available tetrapod material from the Economy Member, however, is inadequate for a definitive biochronological assessment on its own. Huber et al. (1993) highlighted the presence of what they called cf. Cosgriffius sp. because Cosgriffius campi occurs in the Wupatki Member of the Moenkopi Formation in Arizona (Welles 1993), which has been dated as late Early Triassic (Olenekian) by Lucas and Schoch (2002). Previously reported lonchorhynchine trematosaurids all date from the Early and early Middle Triassic, but two clades of Trematosauridae are now known to range into the Late Triassic (Schoch and Milner 2000; Sues and Schoch 2013). Thus, it is conceivable that the lonchorhynchine from the Economy Member, which is not definitely referable to Cosgriffius, is geologically younger. Tanystropheus has a reported stratigraphic range extending from the late Anisian to the late Norian in Europe (Wild 1980). None of the identifiable tetrapod taxa from the Economy Member has been recorded from the overlying Evangeline Member of the Wolfville Formation, which has yielded at least one taxon (Metoposaurus bakeri) indicating a Carnian age (see below).

Whereas all the identifiable tetrapod taxa from the Economy Member have long ranges that are not diagnostic at the level of age (stage) or even epoch, a sequence of members comparable to that of the Wolfville Formation occurs in the Timezgadiouine Formation of the Argana Basin in the western High Atlas of Morocco, which was deposited at a tropical palaeolatitude similar to that of the Wolfville Formation (Olsen and Et-Touhami 2008). There, the lower part of the Timezgadiouine Formation, the Aglegal Member, has yielded capitosauroids but lacks metoposaurs and is succeeded by the Irohalene Member with abundant metoposaurs and no capitosauroids (Jalil 1999). The Aglegal Member has also yielded unambiguous examples of the pseudosuchian ichnotaxon Chirotherium barthii, which is restricted to Middle Triassic strata elsewhere (Klein et al. 2011). This suggests that the partitioning of the temnospondyl assemblages into an older capitosauroid- trematosaurid and a younger metoposaurid assemblage has temporal significance within the Pangaean tropics, corresponding in a broad way to the similar partitioning of the Early to Middle Triassic temnospondyl assemblages from the Moenkopi Formation and the Late Triassic ones from the Chinle Formation, which were also deposited at tropical palaeolatitudes. However, no such temporal partitioning is evident in the higher-latitude Germanic Basin tetrapod assemblages where capitosauroids have a range apparently spanning the entire Triassic Period (Schoch and Milner 2000; Schoch 2011).

The upper part of TS II is made up of the Evangeline Member, the type section of which consists of the coastal outcrops at Evangeline Beach, Kings County (defined in Appendix; Fig. 4E–F). Skeletal remains of tetrapods are quite common in this member, not only as dissociated bones and skulls in the calcite-cemented, pebbly intraformational conglomerate beds as in the Economy Member (found along with unionoid clams) but also as bones and skeletons in the better-sorted sandstones and mudstones. The most important sections for tetrapod fossils in the Evangeline Member are the type section, Burntcoat Head (Burntcoat, Hants County) and near the community of Noel Shore (Hants County) (Fig. 11). Other less productive sites for skeletal remains include Boot Island (Kings County) and Longspell Point (Kingsport, Kings County).

William Take first discovered tetrapod bones in the Evangeline Member and brought them to the attention of Donald Baird in November 1958. At first in collaboration with Take (Baird and Take 1959), Baird undertook subsequent palaeontological reconnaissance of the strata of this unit. After 1985 the present authors and others have continued this work.

Although scraps of bone are not uncommon, the dissociated and fragmentary nature of most skeletal remains often renders their anatomical and taxonomic identification difficult, and sometimes impossible. It was only through persistent collecting over more than four decades that sufficient material has been gathered to permit a new scientific evaluation of the tetrapod assemblage from the Evangeline Member. Very rarely, partial tetrapod skeletons and associated but disarticulated bones of tetrapods occur both in sandstones and mudstones. We offer here what is perhaps a conservative assessment of the tetrapod biodiversity in the Evangeline Member, based only on specimens that show derived features diagnostic for particular clades.

To date, temnospondyls are represented only by the metoposaurid Metoposaurus bakeri (originally assigned to “Buettneria” by Case 1931). It occurs as a natural mould of much of a skull roof from Noel Head (YPM VPPU 021742; Baird 1986; Hunt 1993) (Fig. 12). As in specimens of Metoposaurus bakeri from Texas, the elongate lacrimal is excluded from the anterior margin of the orbit by the contact between the prefrontal and jugal (Hunt 1993; Sulej 2007). YPM VPPU 021742 represents an immature specimen based on its small size (length of skull roof along the midline c. 190 mm), as well as the relative proportions and pattern of sculpturing of the skull roof, especially in the postorbital region. In addition to this skull roof, the mould of a clavicle with characteristic external sculpturing and isolated vertebral centra provide evidence for the presence of metoposaurid temnospondyls in the Evangeline Member. Metoposaurids have long, dorsoventrally strongly flattened skulls and small orbits that are placed far forward on the cranium. These animals were obligatory aquatic predators. Despite reviews by Hunt (1993) and Sulej (2007), a rigorous phylogenetic assessment of metoposaurid diversity is still needed. For example, Baird (1986), Hunt (1993) and Schoch and Milner (2000) assigned the species bakeri to Metoposaurus, whereas Sulej (2007) still followed Case (1931) in retaining this taxon in “Buettneria” (a preoccupied generic name now replaced by Koskinonodon; Mueller 2007).

By far the most common tetrapod fossils in strata of the Evangeline Member are tooth-bearing jaws of procolophonid parareptiles (Fig. 13). Sues and Baird (1998) formally named and described three taxa diagnosed by their respective dentitions: Acadiella psalidodon, Haligonia bolodon, and Scoloparia glyphanodon. They also re-identified a skull and associated postcranial bones (NSM996GF82.1), earlier referred by Baird and Olsen (1983) to Leptopleuron, which is otherwise known only from the Lossiemouth Sandstone Formation (Late Triassic) of Scotland (Säilä 2010), as an immature specimen of Scoloparia glyphanodon.

Scoloparia glyphanodon is by far the most common procolophonid in the Evangeline Member (Fig. 13A). In addition to numerous jaw elements, it is known from several skulls. The more posterior maxillary and dentary teeth of Scoloparia have transversely aligned, chisel-like tooth crowns that bear three cusps when unworn rather than only two as in most procolophonids. The quadratojugal bears a cluster of long, slender, posterolaterally projecting bony spines. The holotype of Scoloparia glyphanodon (NSM996GF83.1) has a ‘nuchal shield’ composed of small, polygonal osteoderms, but no other known specimen preserves this distinctive feature (Sues and Baird 1998). Phylogenetic analysis places Scoloparia as a member of Leptopleuroninae, the most derived clade of Procolophonidae, which also includes the Late Triassic (Norian-Rhaetian) Hypsognathus fenneri from eastern North America (Sues et al. 2000). Cisneros (2008) placed Scoloparia glyphanodon as the sister-taxon of the Early to early Middle Triassic Sclerosaurus armatus from the Buntsandstein of Germany and Switzerland based solely on the shared presence of dermal armour. However, this coding ignores the profound differences in the structure of the dermal armour in the two taxa. Unlike Scoloparia, Sclerosaurus has two or three rows of sculptured osteoderms on either side of the midline of the body and lacks a ‘nuchal shield’ (Sues and Reisz 2008).

The holotype of Acadiella psalidodon comprises an associated right maxilla and partial mandibular ramus as well as part of the pterygoid flange (NSM996GF69.1; Fig. 13B). Its posterior maxillary and dentary teeth have anteroposteriorly rather than transversely aligned crowns with serrated apical ridges (Sues and Baird 1998). Haligonia bolodon, known only from a right maxilla (NSM996GF74.1, holotype) and a left dentary (NSM996GF31.1), has upper and lower tooth rows that have four small teeth followed by a greatly enlarged tooth with a bulbous crown at the back of the jaw. As yet too little is known about the cranial structure of Acadiella psalidodon and Haligonia bolodon to assess their respective phylogenetic positions. Procolophonid parareptiles are superficially lizard-like forms usually less than 50 cm long. Their structure of their dentition suggests herbivorous or omnivorous habits.

Archosauromorph reptiles are represented in the Evangeline Member by a rhynchosaur, a trilophosaur, diverse pseudosuchians, and a possible ornithodiran (Figs. 14-16).

Baird (in Carroll et al. 1972) reported the presence of two taxa referable to a group of probably herbivorous archosauromorph reptiles, Trilophosauridae. These identifications were based on fragmentary cranial remains and thus are problematical. One of these specimens, a natural mould of a small dentary, is referable to Teraterpeton hrynewichorum, which Sues (2003) named on the basis of an excellently preserved, nearly complete skull and much of the postcranial skeleton collected from Burntcoat Head by the avocational collectors George and Sandy Hrynewich (NSM994GF041; Fig. 14A). As in the Late Triassic Trilophosaurus spp. from the American Southwest (Spielmann et al. 2008), the skull of Teraterpeton hrynewichorum has only large upper temporal fenestrae and an otherwise solid, deep temporal region. Unlike in Trilophosaurus spp., the edentulous premaxillae and symphyseal portions of the dentaries are elongate. The external naris is distinctly longer than the orbit. Both the maxilla and palatine bear teeth whose crowns have bulbous bases, a single tall cusp, and an anterior (mesial) ‘heel’ whereas the dentary teeth have the reverse configuration with a tall cusp and a posterior (distal) ‘heel’. Phylogenetic analysis indicated that Teraterpeton hrynewichorum is most closely related to Trilophosaurus spp. among known archosauromorph taxa (Sues 2003). Like rhynchosaurs, Teraterpeton was presumably herbivorous. The distinctive structure of its jaw joint and its maxillary and palatine teeth indicate fore-and-aft jaw motion of the mandible. The deep, blade-like ungual phalanges and robust pectoral girdle suggest that Teraterpeton hrynewichorum was a burrower (which may well account for the excellent, articulated preservation of the holotype).

Jaw elements of various sizes (Fig. 14B–C), incomplete fused parietals (NSM012GF032.023), and the excellently preserved basicranial portion of a braincase (NSM012GF032.022; Fig. 14D) document the presence of a hyperodapedontine rhynchosaur in the Evangeline Member (Baird 1963). Skeletal remains of this taxon have most frequently been found at Evangeline Beach. As in other rhynchosaurs, the premaxilla (NSM012GF032.024) is downturned (relative to the maxilla) and forms an edentulous ‘beak’ (Fig. 14B). The maxilla has only a single deep longitudinal occlusal groove separating the lateral and medial rows of teeth (1 or 2 each) (Fig. 14C). The lateral teeth are pyramidal. Baird (in Carroll et al. 1972) compared the material to the Hyperodapedon gordoni from the Lossiemouth Sandstone Formation of Scotland. Later authors (e.g., Lucas et al. 2002) referred the Wolfville material to the genus Hyperodapedon without further discussion. The rhynchosaur from the Evangeline Member is distinguished from most other hyperodapedontines by the absence of lingual teeth on the dentary. During the Triassic, rhynchosaurs were a widely distributed group of presumably herbivorous archosauromorph reptiles. The maxillae typically bear multiple rows of teeth on either side of a longitudinal groove; some derived taxa have additional grooves between the tooth rows.

Archosaurian reptiles from the Evangeline Member comprise various taxa of crocodile-line archosaurs (Pseudosuchia) (Figs. 15, 16A–B) and a possible representative of bird-line archosaurs (Ornithodira) (Fig. 16C). Several isolated osteoderms with a raised anterior ‘bar’ and radiating ornamentation are typical of Aetosauria, a group of heavily armoured, omnivorous or herbivorous pseudosuchians from the Late Triassic. A paramedian dorsal osteoderm of an aetosaurine (NSM012GF032.033; Fig. 15D; greatest width: 59 mm) shows distinct flexure along the midline and well-developed sculpturing composed of closely spaced pits, the more anterior ones of which are elongate and extend anterolaterally. A ventral osteoderm (NSM012GF032.032; Fig. 15E; greatest width: 24 mm) has relatively weak, sparse pitting and resembles those of Coahomasuchus kahleorum from the Dockum Group (Late Triassic) of Texas (William Parker, personal communication). The well-preserved proximal portion of a right femur (NSM012GF032.014; Fig. 15B) is referable to aetosaurs based on the large posteromedial tuber on the proximal end and the structure of the fourth trochanter (Nesbitt 2011). Scattered postcranial remains recovered from a mudstone horizon at Burntcoat probably belong to a small aetosaur. The rectangular osteoderms bear sculpturing composed of minute pits.

Mostly incomplete and dissociated skeletal elements document the presence of paracrocodylomorph pseudosuchians (Figs. 15, 16A–B). A set of articulated posterior cervical and anterior dorsal vertebrae with associated ribs (NSM998GF46.1; Fig. 15A) probably belongs to a paracrocodylomorph. The vertebrae have tall neural spines. A left tibia (NSM012GF032.015; Fig. 15C; length: 131 mm) has a straight shaft and is gently concave posteriorly at its distal articular end. It is probably referable to a paracrocodylomorph. A well-preserved partial braincase (YPM VPPU 020750; Fig. 16A–B) closely resembles that of Postosuchus kirkpatricki from Late Triassic strata in the American Southwest (Weinbaum 2011, fig. 24), especially in the structure of the parabasisphenoid, which is greatly elongated between the basal tubera and has a deep, dorsoventrally extended median recess. However, the parabasisphenoid is less extended vertically, and the prootic has separate foramina for branches V¹ and V^2-3 of the trigeminal nerve (Fig. 16A).

Surprisingly, the Evangeline Member has not yielded any diagnostic remains of phytosaurs to date. These superficially crocodile-like amphibious predators are very common in other Late Triassic deposits in Europe, Morocco, and North America and are typically found in association with the perennially aquatic metoposaurs. Phytosaurs were long considered early crocodile-line archosaurs but Nesbitt (2011) recently placed them outside crown-group Archosauria.

In contrast to crocodile-line archosaurs, bird-like archosaurs (Ornithodira) are very rare in the Evangeline Member. Baird (in Carroll et al. 1972) reported the presence of dinosaurs in this unit. He identified recurved, labiolingually flattened tooth crowns with finely serrated mesial and distal cutting edges as representing three size classes of theropod dinosaurs. The largest tooth crown in our sample has a mesiodistal diameter of 23 mm near the base and carinae with 5 serrations per millimetre. There are no features to suggest that any of these teeth are dinosaurian or even ornithodiran, and at least some of them more likely belong to paracrocodylomorph pseudosuchians.

Baird (in Carroll et al. 1972) reported a partial left maxilla containing a single tooth (NSM004GF012.001) as representing a tiny ornithischian dinosaur. Irmis et al. (2007, fig. 5B–D) published drawings of the jaw fragment (made by PEO) and showed that it lacks any features exclusive to ornithischians. Instead they noted similarities to Revueltosaurus callenderi, which was initially known only from isolated teeth from the Bull Canyon Formation (Late Triassic: Norian) of New Mexico that were considered ornithischian in origin. Recent finds of extensive skeletal remains of Revueltosaurus callenderi from the Chinle Formation (Late Triassic: Norian) of Arizona have now established that this taxon is a pseudosuchian closely related to aetosaurs, not an ornithischian dinosaur (Parker et al. 2005; Nesbitt 2011). A well-preserved left ilium lacking only the postacetabular process (NSM012GF032.021) from Evangeline Beach closely resembles those of aetosaurs but is also similar to that of Revueltosaurus (Parker et al. 2005, fig. 3c) in its possession of a short anterior (preacetabular) process.

To date the only possibly ornithodiran discovery from the Evangeline Member is a specimen comprising an articulated set of incomplete right metatarsals II–IV, each associated with the proximal phalanx of its respective pedal digit (YPM VPPU 021694); this specimen is preserved in part as bone and in part as natural mould that was filled in with plaster to document its former structure (Fig. 16C). On the associated label Baird identified this find as belonging to a small theropod dinosaur. However, there are no diagnostic features to indicate dinosaurian affinities. The metatarsals are long, slender, and appressed for their entire preserved lengths. Although some early crocodylomorphs also have such ‘bundling’ of the metatarsals these elements are generally not appressed for their entire length (Nesbitt 2011), and thus YPM VPPU 021694 more likely represents an ornithodiran.

Non-mammalian synapsids (therapsids) are represented only by the large traversodontid cynodont Arctotraversodon plemmyridon (Hopson 1984; Sues et al. 1992) and the as-yet-undescribed pelvic bones of a large dicynodont (Tim Fedak, personal communication). Arctotraversodon plemmyridon is known from three partial dentaries and three isolated teeth. The holotype of this taxon (YPM VPPU 019190; Fig. 17) is an incomplete right dentary (with an attached fragment of the left one) from the northeast corner of the Burntcoat headland. Arctotraversodon is readily distinguished by procumbent incisors with coarsely serrated mesial and distal cutting edges (Fig. 17A) and the presence of an enlarged mental foramen on the buccal surface of the dentary (Fig. 17B). Referred molariform postcanine teeth have mesiodistally short tooth crowns (Sues et al. 1992). An isolated lower postcanine (NSM983GF2.1) has three rather than two anterior cusps, a derived feature shared with Boreogomphodon jeffersoni from Late Triassic (Carnian) strata in Virginia and North Carolina, which also shares the presence of an enlarged mental foramen on the buccal surface of the dentary (Sues and Hopson 2010). Baird and Olsen’s (1983) report of a dicynodont therapsid was based on the partial braincase (YPM VPPU 020750) reidentified here as that of a paracrocodylomorph pseudosuchian. However, Tim Fedak (personal communication) has found the ilia and sacrum of a large dicynodont, but this material has not yet been fully prepared and studied.

With the exception of Metoposaurus bakeri, all tetrapod remains from the Evangeline Member that are identifiable at lower taxonomic levels to date appear to be restricted to this unit and thus have no biostratigraphic utility. Metoposaurus bakeri itself is a faunal element of the Otischalkian, the oldest of the Late Triassic Land Vertebrate Faunachrons defined by Lucas (1998) in the American Southwest, and the only taxon that may be Carnian rather than Norian in age. Kozur and Weems (2010) reported the presence of a monospecific conchostracan assemblage with Euestheria minuta from Evangeline Beach (based on material collected by PEO). They argued that this species is restricted to what they considered early Carnian (Cordevolian) strata in the more southeastern basins of the Newark Supergroup. Presumably this is the same taxon that Powers (1916) reported as Estheria ovata from the Evangeline Member of Long Island (Evangeline Beach; Baird in Carroll et al. 1972) and Boot Island. Based on these faunal elements alone, a more precise dating than late Middle to early Late Triassic is not possible. However, the position of the Evangeline Member assemblage in the lower part of the basin section, well below strata dated as Norian based on magnetostratigraphy (see below), and below the position of the boundary between TS II and TS III, suggests a Carnian age.

Whiteside et al. (2011) argued that the procolophonid- dominated Late Triassic tetrapod assemblages from the Newark Supergroup existed under semi-arid climatic conditions, whereas more or less coeval communities dominated by traversodontid cynodonts from Virginia and North Carolina appear to have been restricted to a narrow equatorial zone with more humid conditions. The tetrapod assemblage from the Evangeline Member clearly falls into the former category, with procolophonids representing most of the identifiable fossils. The traversodontid cynodont Arctotraversodon plemmyridon is only known from a few specimens, unlike its close relative Boreogomphodon jeffersoni, which tends to be common where present. Whereas the presence of relatively large, thick-shelled unionoid clams argues for perennial river settings, the presence of abundant palaeosol rhizoliths suggests at least seasonally dry conditions. The relative scarcity of metoposaurs and apparent absence of phytosaurs are consistent with the latter interpretation.

The stratigraphically highest outcrops of the Evangeline Member that have yielded bone are at Longspell Point near Kingsport, Kings County (e.g., Baird MS, p. 253); this occurrence is in calcite-cemented, pebbly intraformational conglomerate, similar in lithology to those outcrops elsewhere in the Evangeline Member that have produced bone. However, at Longspell Point the rest of the upper Wolfville Formation, which we also place in the Evangeline Member, is present in superposition. The upper Evangeline Member as seen here has a somewhat different lithological character, largely lacking the intraformational conglomerates, with generally smaller fluvial bedforms and numerous deeply rooted palaeosols.

The fluvial and aeolian strata of the Red Head Member are unequivocally exposed in three areas: its type area along the shoreline outcrops of Economy Mountain (Red Head) from Red Head to Lower Economy; at the southern end of Chignecto Provincial Park near Advocate, Nova Scotia; and in the vicinity of Medford, Nova Scotia. Only the last area has yielded tetrapod remains (Figs. 18–19).

Outcrops of the Red Head Member occur in two areas in the vicinity of Medford: on the mainland shore cliffs and foreshore; and at Paddy Island, on the foreshore and along the adjacent mainland shore (Fig. 19). These two areas are separated by a fault with normal displacement of about 30 m (Olsen et al. 1989). Baird and PEO first discovered the productive areas (Fig. 18A–C) in October 1973 (Baird MS, p. 533). Olsen and Baird (1986) and Olsen et al. (1989, 2005a, b) briefly described these outcrops and their fossil assemblages.

Locality A, in the Medford cliff outcrops (Fig. 18D), has yielded a skull and a few associated postcranial fragments (including a nearly complete interclavicle) of the leptopleuronine procolophonid Hypsognathus cf. H. fenneri (NSM998GF45.1; Sues et al. 2000). Aside from its smaller size, the skull is distinguished from other specimens of Hypsognathus fenneri from the Newark Supergroup in having relatively small quadratojugal ‘horns’, a proportionately shorter snout, and more open cranial sutures, all of which suggest that it is a juvenile. The specimen was discovered by Alton Brown (then of Arco Petroleum) and collected by PEO during a field trip led by Martha Withjack and PEO in 1983.

Localities B and C have yielded a relatively diverse assemblage of trace fossils consisting of both tetrapod burrows and footprints. Cynodontipus was originally considered the track of a hirsute cynodont (Ellenberger 1976), but is now known to represent a tetrapod burrow (unpublished work by PEO, Mohammed Et-Touhami, and Jessica Whiteside). Baird first recorded the presence of this ichnotaxon in 1959, identifying it as a “…problematic double-gouge ichnite…” in his field notes (Baird MS, p. 254). Both terminal burrow bases formed where the animal ended its burrow at a recalcitrant dry mud layer. There are also bedding-parallel traces, where the burrower tunnelled along the interface between dry mud and sand (Fig. 20). The size of burrows varies considerably. The simplest hypothesis is that the burrows were made by Hypsognathus because procolophonids have long been considered burrowing animals (e.g., Groenewald 1991) and this taxon is known from the same horizon. However, especially in view of the variation in size, other burrowing tetrapods may have been involved. Perhaps Cynodontipus-type burrows were made by a variety of tetrapods and represent plesiomorphic digging behaviour at the level of Amniota, using synapomorphy- based trace-maker identification (Olsen 1995a; Carrano and Wilson 2001).

Tetrapod tracks from the Red Head Member include Atreipus acadianus (Fig. 21A), Brachychirotherium cf. B. parvum (Fig. 21A), Evazoum sp. (Fig. 21B), and Rhynchosauroides cf. R. brunswickii (Fig. 21C). Most of these tracks occur as natural casts because the underlying mudstones that bear the actual impressions are destroyed during the discovery process. These ichnites are very similar to those from the lower and middle Passaic Formation in the Newark basin of New Jersey and Pennsylvania, and several could be conspecific. However, pending a modern revision of the Passaic Formation material, the identifications proposed here must be considered tentative with the exception of Atreipus acadianus, the holotype of which was collected from the Red Head Member at locality C (Olsen and Baird 1986).

The smallest tetrapod tracks from the Red Head Member (Fig. 21C) are virtually indistinguishable from those of Rhynchosauroides brunswickii (Ryan and Willard 1947, as Kintneria; Baird 1957; Olsen and Flynn 1989). Rhynchosauroides cf. R. brunswickii from the Red Head Member is a quadrupedal trackway with a manus impression smaller than that of the pes, although very similar in digital proportions and with the pes generally being less distinctly impressed. The impressions of manual and pedal digit IV generally project the most, with digits III, II, and I being progressively shorter and digit V short and with a tendency to be recurved. None of the existing specimens shows enough detail to ascertain the phalangeal formula, but, based on other examples of Rhynchosauroides, the phalangeal formulae for both manus and pes are probably 2-3-4-5-3 (e.g., Baird 1957), which is the plesiomorphic count for Amniota. The manus is distinctly smaller than the pes, however, and this appears to be a synapomorphy for Diapsida. Given what we know about Late Triassic continental tetrapods, the most plausible hypothesis is that the trackmaker is a lepidosauromorph such as a rhynchocephalian (Baird 1957) or a small early archosauromorph. Rhynchosauroides cf. R. brunswickii is the most common ichnotaxon in the Red Head Member.

Possible pseudosuchians are represented by quadrupedal trackways very similar to Brachychirotherium parvum (Hitchcock 1889; Baird 1957; Haubold 1971) (Fig. 21A). Baird (1957, p. 473) diagnosed Brachychirotherium parvum as follows: “… phalanges of pes digit V reduced and included in the metatarso-phalangeal pad; pes digit IV and V clawless; narrow, curved claws on pes digits I to III borne high above the thickly padded plantar surface and divergent laterally; metatarso-phalangeal pads III and IV coalesced.”

To this, Baird added this description, “the pedal metatarso-phalangeal pads form a straight line of low bosses across the posterior edge of the sole; pedal digit V is encased in a single pad with no claw; and at least digits IV and V of the manus lacked claws.” We identify the tracks from the Red Head Member as Brachychirotherium cf. B. parvum because they resemble Brachychirotherium parvum from the Newark basin in overall proportions and share at least the first three nominally diagnostic characters listed by Baird. The tracks from the Red Head Member appear to have separate metatarso-phalangeal pads for pedal digits III and IV but are smaller than any specimen of Brachychirotherium parvum from the Newark basin; it is possible that the difference in pad shape is size-related. The interpretation of the presence of claws on the manus is complicated by the large number of burrow penetrations on the slabs, some of which occur in positions that could be mistaken for impressions left by manual or pedal claws. However, the best-preserved specimens apparently lack claws on manual digits IV and V. The relatively large manus seems to place both the Red Head specimens and Brachychirotherium parvum from the Newark basin in the ‘large manus group’ of chirotheriid tracks.

Both the absence of claws of pedal digits IV and V and manual digits IV and V are characters seen minimally in Crocodylomorpha. The distribution of these character states outside that clade is poorly documented, because of both the inadequate preservation of known manus and pedes for many taxa and the difficulty of determining whether a small ungual bore a claw. However, claws were definitely present on digits IV and V of the manus and pes of phytosaurs (as evident in the ichnotaxon Apatopus from the Newark basin) and possibly on pedal digit V in the early Middle Triassic archosauriform Euparkeria capensis. The minute ungual phalanges of manual digits III through V and unguals of pedal digits IV and V of the Late Triassic paracrocodylomorph Postosuchus alisonae (Peyer et al. 2008) appear too small and blunt to have borne claw sheaths. The absence of claws on digits IV and V of both manus and pes may represent synapomorphies at some level within the Archosauria, perhaps at the level of Pseudosuchia to the exclusion of Phytosauria.

Unfortunately, the coalescence of manual and pedal phalangeal pads in Brachychirotherium parvum and the Red Head specimens of Brachychirotherium cf. B. parvum precludes determination of the phalangeal formula. However, some members of Pseudosuchia can be ruled out as track makers because of the size of the manus relative to the pes. In at least some paracrocodylomorphs, such as Postosuchus alisonae (Peyer et al. 2008), the manus is relatively smaller than seen in the forms in question. In the Aetosauria, the manus appears relatively larger, although confusion concerning the association of the best manus and pes specimens (see Lucas and Heckert 2011) prevents definitive assessment. Thus the simplest inference is that Brachychirotherium parvum and the Red Head specimens of Brachychirotherium cf. B. parvum represent either crocodylomorphs or another group of pseudosuchians.

A non-dinosaurian dinosauromorph is represented by the tracks of Atreipus acadianus, the type material of which comes from the Red Head Member at locality C (Olsen and Baird 1986; Fig. 21A). Atreipus acadianus was produced by a habitually quadrupedal tetrapod with a tridactyl pes and a functionally tetradactyl manus. No hallux is evident, even in the deepest impressions, as is also the case in the other North American ichnospecies of Atreipus. Olsen and Baird (1986) argued on phylogenetic grounds that Atreipus was made by either an ornithischian or a member of the sister-group of Dinosauria, which we would now term a non-dinosaurian dinosauromorph. Haubold and Klein (2000), Carrano and Wilson (2001) and Irmis et al. (2007) concurred with this assessment. Subsequently, with the discovery of the dinosauriform Silesaurus from the Late Triassic of Poland (Dzik 2003), it became apparent that at least North American examples of Atreipus also share a potential synapomorphy with Silesaurus to the exclusion of Dinosauria and non-dinosauriform dinosauromorphs. This feature is the distinct reduction in the length of digit I well beyond that seen in basal saurischian dinosaurs but present in the pes of Silesaurus as well as in deeply impressed prints of Atreipus that always lack traces of a hallux. In contrast, deep impressions of Triassic-Early Jurassic brontozoid tracks (sensu Rainforth 2005; traditionally classified as Anchisauripus, Eubrontes and Grallator) always show a trace of the hallux. This suggests that the trackmaker of Atreipus was a member of Dinosauriformes including Silesaurus but excluding Marasuchus, and may well have been a silesaurid, a group now known to have attained a wide geographic distribution during the Triassic (Nesbitt 2011).

Smaller examples of Atreipus acadianus sometimes lack manus impressions, as well as the metatarso-phalangeal impression of pedal digits II and IV that is usually present in North American specimens of Atreipus. Such tracks are easily confused with brontozoid footprints. However, intergradations between all these forms occur even within individual trackways, and thus we think that brontozoid tracks have not yet been found in the Red Head Member and that there is no evidence of coelophysoid theropods in this unit.

The only possible dinosaurian ichnites from the Red Head Member are functionally tridactyl or even didactyl tracks with an unusually elongate digit IV and an apparent trenchant claw on digit II assigned to the ichnogenus Evazoum (Lockley and Lucas 2013; Fig. 21B). This form has previously been referred to as Coelurosaurichnus sp. B (Olsen et al. 1989), “?saurischian dinosaurian track ‘new genus 1’ ’’ (Olsen and Rainforth 2003) and a ‘new dinosaur-like ichnogenus’ (Olsen et al. 2005a, b). Lockley et al. (2006) noted the similarity between this form and the ichnogenera Evazoum and Kalosauropus. However, in terms of synapomorphy-based reasoning, this similarity is largely based on retention of plesiomorphic features below the level of Dinosauria, in combination with pedal digitigrady and with bipedality. Lockley and Harris (2011) suggested Evazoum was made by a sauropodomorph, but again the resemblances are all symplesiomorphies below the level of the Dinosauria. Without any distinct pedal synapomorphies relating it to a specific group, we can only conclude that Evazoum might represent an early dinosaur, perhaps even a basal saurischian or a gracile sauropodomorph, but non- dinosaurian affinities cannot be ruled out.

The overall assemblage of tetrapod tracks from the Red Head Member is very similar at the level of ichnogenus to that from the middle Passaic Formation of the Newark basin. Most ichnospecies from the Red Head Member are also shared with the middle Passaic except for Atreipus acadianus, which is thus far known only from the Red Head Member. Assemblages from the Passaic Formation also include the presence of Evazoum — CU 170.3 and CU 170.4 were figured and misidentified as coming from the Wolfville Formation by Lockley and Lucas (2013) but are actually from Lyndhurst, New Jersey (Olsen and Rainforth 2003). Based on palaeomagnetic polarity stratigraphy, the Red Head Member is coeval with the middle Passaic Formation (Kent and Olsen 2000), and thus a close correspondence in ichnotaxa is not unexpected.

Outcrops along the shores of Economy Point may pertain to the Red Head Member. Eric Leighton found a single archosauriform tooth crown with serrated edges (NSM014GF023.003) at Thomas Cove (at approximately 45.360042°N, 63.917850°W). The strata along this section lie above the unconformity visible on the Consolidated Beacon Resources Ltd. Line 10 seismic profile described in Withjack et al. (2010) and therefore should belong to the Red Head Member. The abundant mud-cracked bedding-plane surfaces in these strata are consistent with this interpretation, although the tooth-producing unit is a calcite-cemented intraformational conglomerate resembling units in the Wolfville Formation.

Overlying the Red Head Member is the White Water Member (defined in Appendix), which crops out at multiple localities along the shores of the Minas Basin and St. Mary’s Bay. Through most of its extent, the White Water Member consists of rhythmically bedded sedimentary packages called sand-patch cycles (Smoot and Olsen 1988; Olsen et al. 1989), which formed largely in playas with efflorescent halite crusts. These rhythmic sediments are devoid of tetrapod remains except for one poorly preserved, possible dinosauromorph track (cf. Atreipus; Fig. 22). However, in the shore outcrops on the north side of St. Mary’s Bay, especially at Rossway (Figs. 23–24), the sedimentary cycles have much less development of sand-patch fabric and the section is far more fossiliferous, especially with respect to tetrapods; however, all but one locality there have yielded only trace fossils.

Cynodontipus burrows found in the Rossway section may have been produced by procolophonid parareptiles. One particularly informative specimen, found by Jessica Whiteside but unfortunately not collected because of its large size, intertidal location and time constraints, resembles the complex burrow systems reported by Voigt et al. (2011) from the Aglegal Member of the Timezgadiouine Formation of the Argana Basin in Morocco. The Rossway specimen, though, also had clear Cynodontipus-type crescent-shaped scratch marks in positive hyporelief directly attached to the burrows, demonstrating their connection.

One example of Rhynchosauroides sp. was found but not collected by PEO and figured by Olsen et al. (2005a, b; Fig. 24F). It is larger than specimens of Rhynchosauroides brunswickii from the Red Head Member, but additional material is needed for more precise ichnotaxonomic identification. As is the case with the tracks from the Red Head Member, Rhynchosauroides sp. from the White Water Member may represent the tracks of a lepidosauromorph.

The prenarial portion of the cranium of a slender- snouted phytosaur (YPM VPPU 007920), with an associated osteoderm and bone fragments, was found by PEO, Bob Salvia and A. Heimlich in 1974 and represents the only skeletal tetrapod remains yet found from the White Water Member (Fig. 25). This is the only known specimen of a phytosaur from the Fundy basin and the first record of this group from Canada. It is a phytosaurid and was initially identified as Rutiodon sp. (Olsen et al. 1989). However, no synapomorphies have been identified to date that would link the specimen from the Blomidon Formation with the geologically considerably older Rutiodon carolinensis.

One example of cf. Brachychirotherium sp. (Fig. 24E) suggests the presence of Pseudosuchia in the White Water Member, but this specimen is rather indistinctly preserved and not identifiable at a lower ichnotaxonomic level.

Possible silesaurid ornithodirans are represented by several pes impressions. These closely resemble Atreipus milfordensis in their distinctive ‘tulip’ shape (Fig. 24D).However, they all lack definitive manus impressions, and thus identification is somewhat uncertain.

The Rossway locality is noteworthy for having yielded the only definitive brontozoid footprints from strata below the basalt in the Fundy basin, and thus they represent the only pre-basalt evidence for theropod dinosaurs in this region. Two large slabs were found (PEO, Amy Litt, and Louise Roth in August 1976; PEO and Dan Simanek in July 1978), but neither could be collected (Fig. 24A–B). The brontozoid tracks have a proportionately long pedal digit III, with (when well-preserved) well-developed pads, especially the most distal, a relatively narrow pes with low angulation, and a tendency for the metatarso-phalangeal pad of digit IV not to leave an impression. Following the classification of Lull (1953), these tracks would be identified as Anchisauripus sillimani or Anchisauripus tuberosus but, as Olsen et al. (1998) argued, the proportional features traditionally employed to differentiate the ichnogenera Anchisauripus, Grallator and Eubrontes are size-related along a morphological continuum. Identification as theropod tracks is based on their functionally tridactyl nature, with a distinct hallux (not actually seen in these specimens), and full bipedality. However, it is possible that some as yet unknown non-dinosaurian dinosauriforms had a similar foot structure that, based on tracks, might not be separable from that of theropods. However, the fact that identical footprints occur in Early Jurassic strata from which no non- dinosaurian dinosauriforms are known and in which the only plausible makers of brontozoid tracks are theropod dinosaurs lends support to the original identification.

In overall composition, the footprint assemblage from the Rossway locality most closely resembles that from the lower part of the upper Passaic Formation (roughly between the Metlars Member and Member OO; Olsen et al. 1996) in the presence of medium-sized brontozoid tracks but with Atreipus still present. This would correlate with the lower to middle White Water Member on the Medford shore, east at Blomidon, above the level of the Red Head Member track assemblage. There is as yet no independent test for this hypothesized correlation.

The top of the Blomidon Formation consists of a less than 10 m thick but distinctive set of beds here named the Partridge Island Member (see Appendix). Although no evidence of tetrapods has been recovered yet from this unit, it requires some discussion because it is important in understanding the chronology of faunal changes across the Triassic-Jurassic transition. At its type section, the Partridge Island Member consists of about 1 m of interbedded red, purple, grey, and black (variegated) mudstone and minor carbonates that have proven critical to the interpretation of the end-Triassic extinction (e.g., Blackburn et al. 2013). The unit contains pollen and spores at several levels, documenting the initial phase of the end-Triassic extinction (Fowell and Traverse 1995; Whiteside et al. 2007; Cirilli et al. 2009), has three modest iridium anomalies (Tanner and Kyte 2005; Tanner et al. 2008; Kyte et al. 2008), and contains the upper part of palaeomagnetic polarity chron E23r (Deenen et al. 2011), which extends downward into the uppermost portion of the White Water Member (see Appendix). Based on zircon U-Pb dates, the age of the Partridge Island Member is latest Rhaetian at 201.564±0.015 Ma (Blackburn et al. 2013). Although Olsen (1997) and Olsen et al. (2005a, b) interpreted the Partridge Island Member as comprising the base of TS IV, it rests conformably on the underlying Blomidon Formation (contra Kozur and Weems 2010) as shown by the position of E23r (Deenen et al. 2011) (see discussion in Appendix).

Facies typically bearing pollen and spores indistinguishable from the type section characterizes the Partridge Island Member over a large area of the Fundy basin — minimally 460 km² as described by Olsen and Et- Touhami (2008) and contra Kozur and Weems 2010 (see Appendix). However, at some localities such as just east of the The Old Wife in Five Islands Provincial Park, the overlying North Mountain Basalt metamorphosed strata of this unit (which Kozur and Weems 2010 interpreted as a palaeosol — see Appendix). Although locally beds in this member appear promising for the presence of tetrapod tracks, no footprints have been found to date.

There are no exposures of definitive sedimentary interbeds within the North Mountain basalt, with all known possible examples being more parsimoniously identified as void fillings within or between flows (Olsen et al. 2012) that postdate the basalt and hence are considered part of the McCoy Brook Formation (see below). Consequently, there are no occurrences of tetrapods from this basalt formation. However, the basalt does contribute to the chronology of events through the Triassic-Jurassic transition because it is one of the few CAMP units for which there are high- precision U-Pb dates (201.566±0.031 Ma; Blackburn et al. 2013).

Although Donohoe and Wallace (1978) first used the name McCoy Brook Formation it was formally defined by Tanner (1996), who provided a measured type section. The formation consists primarily of red clastic rocks and conformably overlies the North Mountain Basalt. However, a widespread, generally grey, white, green, and purple carbonate-rich unit, the Scots Bay Member (Powers 1916; Tanner 1996), forms the base of the formation at most outcrops and below much of the Bay of Fundy.

Locally, especially along the Minas fault zone, the relationship between the McCoy Brook Formation and North Mountain Basalt is complex not only due to post- depositional faulting but also because of sedimentation during faulting (Olsen et al. 1989; Olsen and Schlische 1990; Tanner and Hubert 1991). Probably related to the syntectonic sedimentation is the fact the McCoy Brook Formation is by far the most tetrapod-rich stratigraphic unit of the Fundy Group, with every major outcrop yielding tetrapod bones, or tracks, or both. From west to east along the north shore of Minas basin and then to the south to Scots Bay, important localities with outcrops of the McCoy Brook Formation are: 1) Wasson Bluff Protected Area; 2) McKay Head; 3) Blue Sac west; 4) Blue Sac east; 5) Five Islands Provincial Park; and 6) coves along the southern shore of Scots Bay. et al.

Wasson Bluff Protected Area. Cliff and foreshore out- crops between Wasson Bluff and Swan Creek contain by far the greatest wealth of early Mesozoic tetrapod remains in the Fundy basin. Dawson (1855, fig. 7) illustrated this section, which he had studied in 1846 and 1850. Describing the Wasson Bluff to Swan Creek section, Dawson (1855, p. 87) wrote, “…the coast exhibits the interesting and complicated appearances which I have endeavoured to represent in [his] Fig. 7.” (We reproduce Dawson’s diagram here as Fig. 26.) His description is easy to follow, and he noted a place of special interest at Wasson Bluff: “Opposite the Two Islands, the fissures of the trap are lined with fine crystals of analcime and natrolite; and the fissures and vacant spaces of the trap conglomerate in the same neighbourhood contain a reddish variety of chabasie [chabasite] in rhombohedrons, often of large size” (Dawson 1855, p. 89). The latter clearly refers to the basalt talus breccias at locality a (Figs. 28-29), which is the most consistently fossiliferous location in the Wasson Bluff area. However, Dawson did not report fossils from this section.

PEO first discovered postcranial remains of a sauropodomorph dinosaur in sandstones of the McCoy Brook Formation at Wasson Bluff in August 1976 as part of a team from Princeton University led by Donald Baird (Fig. 27). Following his find of an incomplete but excellently preserved skull of a sphenodontian lepidosaur in 1984 (Fig. 31A), PEO and Neil Shubin (then at Harvard University) secured funding for more extensive reconnaissance. This effort led to the discovery of several vertebrate-bearing deposits at Wasson Bluff (Fig. 28) between 1985 and 1988; important new material continues to be found there on a regular basis. Seven stratigraphic levels representing as many major distinct depositional environments occur in the section. The oldest is an orange aeolian sandstone overlain by a basalt talus slope breccia with red sandy mudstone deposited in the eastern micro-graben against fault scarps of the North Mountain Basalt (Fig. 28A, locality a). The basal sandstone has yielded a few sphenodontian bones (cf. Clevosaurus), but the stratigraphically highest exposed talus breccias (first discovered by William Amaral, Chuck Schaff, Neil Shubin, and H-DS in July 1985) have yielded hundreds of mostly fragmentary but often well preserved bones (Figs. 29C, 32C). Particularly common are skeletal elements and osteoderms of the crocodyliform Protosuchus micmac, including the holotype (Sues et al. 1996), followed by rare remains of the tritheledontid cynodont Pachygenelus cf. P. monus (Shubin et al. 1991).

A small right maxilla was tentatively attributed to Crocodylomorpha (Shubin et al. 1994), but its affinities remain uncertain because it lacks diagnostic features. Possible dinosaurian remains have also been found. This is the easternmost chabasite locality labelled ‘c’ by Dawson (1855, fig. 7) (Fig. 26).

The crocodyliform Protosuchus micmac is known from an incomplete skull (FGM999GF64; Fig. 32A–B), a well-preserved braincase (MCZ 9115), isolated cranial and postcranial bones including numerous osteoderms (Sues et al. 1996). Diagnostic features separating it from other species of Protosuchus include the consistent presence of two caniniform teeth (rather than one) in each dentary and the slender rather than deep anterior process of the jugal extending below the orbit (Clark 1986; Gow 2000). The type species, Protosuchus richardsoni, is known from the upper part of the Dinosaur Canyon Member of the Moenave Formation (Early Jurassic) of Arizona (Colbert and Mook 1951; Clark 1986). A closely related species, Protosuchus haughtoni, occurs in the Upper Elliot and Clarens formations (Early Jurassic) of southern Africa (Busbey and Gow 1984; Clark 1986; Gow 2000).

The tritheledontid Pachygenelus cf. P. monus is known from incomplete cranial and mandibular elements, as well as isolated postcanine teeth from the talus breccia (Shubin et al. 1991). Its postcanine teeth are virtually indistinguishable from those of Pachygenelus monus from the Upper Elliot Formation of southern Africa (Gow 1980). The upper postcanines have a tall principal cusp that is flanked by a smaller, slightly lingually displaced cusp mesially and distally. A labial cingulum links the two smaller cusps. A previously unreported partial right maxilla (MCZ 9137) preserves a canine, five postcanine teeth and two empty postcanine alveoli. The labiolingually compressed lower postcanines have a principal cusp that is followed by smaller accessory cusps distally, and a lingual cingulum (Fig. 33).

Overlying the talus-slope breccia in the middle mini- graben, as we interpret it, are largely tabular, lacustrine sandstones, mudstones, and limestones of the Scots Bay Member. At one important locality in the western mini- graben (Fig. 28A, locality c), the Scots Bay Member laps onto a faulted basalt surface (Fig. 30). Over most of its extent at Wasson Bluff, the Scots Bay member has a single bed of purple green and white limestone and calcareous sandy mudstone that contains disarticulated remains of the holostean fish Semionotus and unidentified palaeonisciforms, teeth of hybodont sharks, and coprolites. However, where this unit intersects the faulted basalt, there is a talus-slope breccia, the matrix of which is a red and purple mudstone rich in fish bones and scales and with relatively common isolated tetrapod remains. The most common tetrapod fossils are teeth referable to small ornithischians and possible theropods, followed by bones and osteoderms of Protosuchus, and bones of cynodont therapsids including a tritylodontid dentary and limb bones.

Fedak et al. (in press) described a fragment of a right dentary with a molariform postcanine tooth followed by two empty alveoli posteriorly (NSM012GF014.006). This specimen is referable to the tritylodontid cynodont Oligokyphus based on the presence of two longitudinal rows with three distinct cusps each on the molariform. The tooth closely resembles those of Oligokyphus sp. from the Kayenta Formation (Early Jurassic: Pliensbachian) of Arizona (Sues 1985) in the presence of a distinct cingular cusp at the anterior (mesial) end of each row of cusps and pronounced perikymata on the enamel. Originally known only from two isolated postcanine teeth from the Rhaeto- Liassic of southern Germany (Hennig 1922), Oligokyphus has subsequently been recorded from Early Jurassic strata in southwestern Britain (Kühne 1956), Arizona (Sues 1985), and Yunnan (China; Luo and Sun 1994). Fedak et al. (in press) also reported on a right humerus (NSM014GF014.002; Fig. 33C) and a partial right ulna (NSM014GF014.003; Fig. 33D) first illustrated by Olsen et al. (2005a, fig. 20G-H) closely resemble the corresponding bones in Oligokyphus (Kühne 1956). The deltopectoral crest and entepicondylar end of the humerus (length: 37 mm) are damaged, but the element is otherwise well preserved. The humerus lacks an ectepicondylar foramen, a condition shared by Tritylodontidae and Mammaliaformes, but not Tritheledontidae (Martinelli et al. 2005). The ulna (preserved length: 22 mm) lacks its distal end and the radial facet. It has a well-developed olecranon process and a distinct insertion scar for the brachialis muscle just distal to the sigmoid notch. This ulna could belong to a tritylodontid, a tritheledontid or a mammaliaform. Finally, the proximal portion of a small right ischium (NSM014GF014.004; Fig. 33E) has an acetabular facet that is separated from the articular contact with the pubis by a deep, narrow groove. Based on this feature, this bone can also be referred to a tritylodontid, a tritheledontid or a mammaliaform.

Although two different groups of very mammal-like non-mammalian cynodonts are known from the McCoy Brook Formation, no diagnostic skeletal remains of mammaliaforms have yet been recorded from this unit. Their presence would not be unexpected because mammaliaforms occur in more or less coeval strata elsewhere.

The fish-scale-rich, tetrapod-bearing mudstone matrix around basalt clasts is most simply interpreted as a wave- sorted shoreline lag. A remnant of probably the same unit was preserved in a small (metre-scale) pocket attached to a giant basalt clast on the westernmost edge of the middle mini-graben, evidently as part of a rider block on the master fault on the west side of the half graben (Olsen et al. 2005a, b).

Isolated tooth crowns from the fish-scale-rich mudstone of the Scots Bay Member (tooth: NSM014GF014.001, Fig. 34C; impressions of teeth: YPM VP 008668, VP 008691) closely resemble those of the teeth of the basal ornithischian dinosaur Lesothosaurus diagnosticus from the Upper Elliot Formation of southern Africa (Sereno 1991). However, these resemblances are plesiomorphic in nature (Sereno 1991), and the teeth from the Scots Bay Member cannot be identified more precisely than Ornithischia. Some postcranial bones originally assigned to ornithischian dinosaurs on their specimen labels, such as vertebral centra (YPM VP 008693a, b) and possible rib fragments (YPM VP 008694, VP 008695), show no features supporting such an identification. The ornithischian teeth represent the oldest definitive skeletal remains of dinosaurs known to date in Canada. Finally the fish-scale-rich mudstone has yielded small, recurved tooth crowns with finely serrated edges. These crowns are referable either to crocodylomorphs or to theropod dinosaurs (Olsen et al. 2005a, fig. 20F).

Overlying the Scots Bay Member are fluviolacustrine red and brown sandstones and mudstones, which are well exposed in the cliffs and on the foreshore (Fig. 28A, localities b and c). These deposits contain abundant dissociated and occasional articulated material of (in order of abundance) the sphenodontian lepidosaur Clevosaurus bairdi (including the holotype, NSM988GF1.1; Fig. 31A), the crocodyliform Protosuchus micmac (Fig. 32A–B), and the cynodont Pachygenelus cf. P. monus (Fig. 33A–B).

Other than protosuchid osteoderms, well-preserved skeletal remains of the sphenodontian lepidosaur Clevosaurus bairdi are the most commonly found tetrapod fossils in the fluviolacustrine strata (Sues et al. 1994). This taxon is represented not only by a number of isolated dentaries and maxillae but also two incomplete skulls and two partial skeletons, one of which (FGM996GF050) includes much of the skull (Fig. 31B–C). Referral of the specimens to Clevosaurus is based on the presence of a lateral contact between the ectopterygoid and palatine, which excludes the maxilla from the margin of the suborbital fenestra, and the presence of a long, posteriorly extending dorsal process of the jugal (Sues et al. 1994; Jones 2006). As in other clevosaurs, the snout is proportionately short. Clevosaurus bairdi has more teeth than the Late Triassic type species of Clevosaurus, Clevosaurus hudsoni, from southwestern England (Fraser 1988), and its skull appears relatively more robust. Clevosaurus was widely distributed across Pangaea and ranged from the Late Triassic to the Early Jurassic (Sues et al. 1994; Bonaparte and Sues 2006).

The tritheledontid Pachygenelus cf. P. monus is represented by a few incomplete cranial and mandibular elements from the fluviolacustrine strata (Shubin et al. 1991; Fig. 33A–B).

Thick orange-brown aeolian dune sandstones and less common brown interbedded interdune sandstones and red fluviolacustrine mudstones overlie the fluviolacustrine red and brown sandstones (Fig. 28A, locality f). Tetrapod elements are generally rare in these deposits and comprise mostly isolated bones of Clevosaurus; a significant exception is a bone bed with articulated, as well as dissociated, skeletal remains of sauropodomorph dinosaurs preserved with basalt clasts (Fedak 2007) in what is plausibly interpreted as an interdune clastic deposit in close proximity to a fault scarp of basalt.

Locality f has produced several associated sets of skeletal remains of sauropodomorph (‘prosauropod’) dinosaurs. Following PEO’s initial discovery of a cervical vertebra in August 1976, several vertebrae and limb-bone fragments (YPM VPPU 022196) were recovered during two subsequent visits by teams from Princeton University led by Don Baird (Fig. 34). Ken Adams, former Curator of the Fundy Geological Museum, and George Hrynewich later collected much of another postcranial skeleton (FGM994GF69). This specimen is smaller (femur length less than 30 cm) than the original find (estimated femur length about 45 cm). Further work by Tim Fedak at this site in 1997 led to the discovery of a third partial skeleton (FGM998GF9; femur length about 50 cm) from the Princeton quarry. Further excavations recovered a left complete femur, tibia, and fibula, the distal end of a right femur, several vertebrae, a nearly complete ischium, and several bones of the forelimbs. In 1998, a large, articulated specimen was discovered (FGM998GF13_I). During subsequent excavations to recover this material, two additional sets of skeletal remains (FGM998GF13_II and FGM998GF13_III) were found, including much of a dissociated skull. This material has yet to be published in detail. The specimens have previously been cited as cf. Ammosaurus sp. (Shubin et al. 1994). However, their affinities remain uncertain because many of the bones were affected, often severely, by tectonic deformation, and perhaps more than one taxon is represented.

One partial, badly deformed and fragmentary skeleton (NSM005GF009 and FGM998GF46; Fedak 2007), probably a sauropodomorph, was found below the talus-slope breccia on the western side of the middle mini-graben (Fig. 28A, locality g). This specimen is important because it preserves gastroliths (Whittle and Onorato 2000; Fedak 2007). The stones themselves can be identified as gastroliths not only because of their position inside the ribs and gastralia fragments, but more importantly because they comprise mostly rounded quartz clasts, a lithology otherwise completely absent from the McCoy Brook Formation in the Wasson Bluff region. Therefore, transport to the fossil site in the gastric mill of the dinosaur from another region seems the only reasonable explanation for their occurrence. This find provides further evidence that at least some sauropodomorphs apparently used a gastric mill (cf. Wings 2005). The apparent absence of similar gastroliths in other sauropodomoph specimens from Wasson Bluff may reflect postmortem loss or basalt gastroliths that have gone unnoticed due to the abundance of other basalt clasts in the matrix. In addition, a mandibular ramus of Clevosaurus was found associated with the gastroliths (PEO, personal observation), which led Barrett (2000) to argue that ‘prosauropods’ were omnivorous. Possible alternative explanations are that the small reptile bone was deliberately ingested for nutrients (Esque and Peters 1994; White 2011), or that the bone was accidentally associated with the dinosaurian remains (Fedak 2007).

The youngest strata in the middle mini-graben are exposed at low tide in the foreshore (Fig. 28A, locality e) and consist of tabular sandstone beds, which have yielded small brontozoid footprints (Anchisauripus sensu Lull 1953). These tracks unfortunately disintegrated before they reached a repository (Fig. 35).

The western mini-graben contains yet another distinctive sequence (Fig. 28A, locality h), which is difficult to correlate with that in the middle mini-graben except that the section includes the Scots Bay Member at its base (Fig. 36A). The only tetrapod material recovered from this area to date is a small block of purplish calcareous mudstone preserving a patch of abdominal dermal armour probably referable to Protosuchus (FMG007GF1) and found by Gustaf Olsen in 2007 (Fig. 36B). The matrix closely resembles the calcareous stringers found in palaeosols interbedded with the basalt debris flows.

McKay Head. Outcrops between the North Mountain Basalt headland of McKay Head and Wasson Bluff (Fig. 36) have yielded an important footprint assemblage from the foreshore of the type-section of the McCoy Brook Formation. PEO discovered the first tracks in 1978. Subsequently, PEO, his associates, and the collector Eldon George from Parrsboro have collected significant additional material. The deposits appear to represent fluvial crevasse splays, although a marginal lacustrine setting cannot be ruled out. The footprints are recovered only as natural casts (positive hyporelief) in sandstone because the impressions in mudstone disintegrate during the collecting process.

Crocodylomorphs are represented by abundant small to large tracks of Batrachopus, the trackmaker attribution being closely similar to that of Batrachopus deweyii and Batrachopus dispar (Olsen and Padian 1986) (Fig. 37C). Theropod dinosaurs produced small (e.g., YPM VP 008668, YPM VPPU 23631) to minute brontozoid tracks (Anchisauripus and Grallator in the classification by Lull 1953). The media have touted the smallest specimen (also on slab YPM VPPU 023631; Fig. 37D) as the smallest known dinosaur footprint in the world, but tracks of comparable size are also known from other Newark Supergroup deposits (Olsen 1995b; Olsen et al. 1998). Numerous tracks of Anomoepus scambus document the presence of ornithischian dinosaurs (Olsen and Rainforth 2003; Figs. 37H, 38A). Particularly noteworthy is the relative abundance of the ichnotaxon Otozoum moodii, which has been plausibly attributed to sauropodomorphs (Rainforth 2003) and some of which are beautifully preserved. Eldon George donated a particular noteworthy partial trackway of Otozoum to the Nova Scotia Museum of Natural History (Grantham 1996). The absence as yet of large brontozoid tracks such as Eubrontes is striking.

Blue Sac west. Cliff and foreshore outcrops between Blue Sac Road and Moose River have yielded a few protosuchid osteoderms from talus-slope breccias and a significant footprint assemblage from fluviolacustrine strata (Fig. 37). Outcrops of basalt talus-slope breccias of the McCoy Brook Formation occur at the more eastern outcrops and have yielded the osteoderms (rediscovered on 2 July 2014 at 45.401530°N, 64.135475°W). These breccias do not occur in proximity to any known actual basalt flow outcrops, the palaeocliffs of which were presumably lifted to a higher structural level and subsequently eroded. More westerly outcropping strata presumably occur at a higher stratigraphic level than the breccia and closely resemble deposits of the type McCoy Brook Formation and yield well-preserved footprints, the first of which were discovered by PEO in 1978. Small brontozoid tracks (Anchisauripus) have been found along with relatively common footprints of Anomoepus scambus (YPM VP 008665, FGM994GF2; Olsen and Rainforth 2003; Fig. 38A, C) and poorly preserved tracks of Otozoum (YPM VP 008669; Fig. 38B).

Blue Sac east. Beach cliff outcrops east and north east of Blue Sac Road have yielded only tracks of Otozoum from volcanoclastic sandstones in rock falls (Fig. 39). This ichnotaxon is usually so uncommon that this occurrence deserves special mention.

Five Islands Provincial Park. Extensive outcrops ofMcCoy Brook Formation occur in large cliffs and along the foreshore to the west of The Old Wife in Five Islands Provincial Park. Originally these outcrops were thought to represent part of the Blomidon Formation, but on structural grounds Liew (1976) correctly reinterpreted them as lying above the North Mountain Basalt rather than below it. This conclusion corresponds with PEO’s 1973 observation that the succession preserves a footprint assemblage similar to one from the Hartford basin and first discovered by PEO and Baird in 1973 (Baird MS, pp. 531-532). Although strata west of The Old Wife are now accepted as part of the McCoy Brook Formation, their precise level within the formation is not clear. Although a contact with the North Mountain Basalt is exposed within the cliff face, the contact is structurally complex and condensed. Structural continuity between the main cliff outcrop of gently dipping strata and the basalt is not evident, but it appears that the McCoy Brook strata may lie far above the basalt contact, plausibly making them the stratigraphically highest and youngest exposed units withinthe Fundy basin.

Although no definitive tetrapod skeletal remains have been recovered from Five Islands Provincial Park, an important assemblage of footprints has been assembled over the years (Fig. 40). This assemblage differs from those at McKay Head and Blue Sac east and west in being dominated by brontozoid tracks of all sizes, including Eubrontes giganteus (Fig. 40C) and apparently lacking Anomoepus and Otozoum.

Coves on the South Shore of Scots Bay. The largely grey, green and brown limestone, silicified limestone, and sand- stone beds of the type Scots Bay Member have yielded a few tetrapod bones and a slightly richer assemblage of footprints (Fig. 41). To date several sphenodontian bones have been reported (Cameron and Jones 1988), as well as small and large brontozoid tracks (Fig. 41B–E). The assemblage is consistent with, but much less rich than, that at Wasson Bluff.

The tetrapod assemblage from the McCoy Brook Formation closely resembles presumably more or less coeval, post-end-Triassic-extinction assemblages from China, Europe, southern Africa and the American Southwest (Sues et al. 1994). With the possible exception of the sauropodomorph dinosaur, all identifiable tetrapod genera are shared by at least one of these other assemblages.