The first stratigraphic account of the Carboniferous section at Joggins, and for the Maritimes Basin in general, was the pioneering work of Brown and Smith (1829). Their stratigraphic framework, and subsequent modifications (Fig. 2; Gesner 1836, 1843; Lyell 1843; Dawson 1878), underscore the geologic parallels between once-contiguous Nova Scotia and British Isles (Calder 1998).

The Carboniferous strata exposed along the eastern shore of Chignecto Bay were first measured by William Logan during the summer of 1843 (Logan 1845; reprinted in Poole 1908). Logan measured about 15 km of virtually continuous coastal section from Minudie south to Shulie, a succession that, by his calculation, totalled 4442.4 m (14 570 feet, 11 inches). Incredibly, this feat was achieved during a five-day stop en route from London to the Gaspe Peninsula (Rygel and Shipley 2005) as he began the search for coal-bearing strata in Lower Canada; it constituted the first field project of the fledgling Geological Survey of Canada.

Logan subdivided his section into eight lithologically distinct divisions, numbered from youngest (1) to oldest (8). The red beds of Lower Cove were assigned to Division 5, and the overlying coal measures of the Joggins and Springhill Mines formations to Divisions 4 and 3 respectively. The sandstone-dominated strata of the underlying Boss Point Formation were assigned to Division 6. Logan placed the lower contact of Division 5 at the top of "South Reef" (Fig. 3), which we consider to represent the uppermost sandstone body of the Boss Point Formation. He placed the upper contact 5 ft, 6 in (1.65 m) below the stratigraphically lowest coal (no. 45) of Division 4, a bed that we propose herein as the basal datum of the revised Joggins Formation (see below).

Logan's "Recapitulation" of Division 5 reads as follows: "dominated by red shale, reddish grey sandstone with occasional "drift plants" and "concretionary limestone", and minor greenish grey sandstone, totally 2082 ft 0 in". The thickness of the Lower Cove red beds (634.6 m) obtained by Logan agrees remarkably closely with the thickness of 635.8 m measured in this study (a mere 1.26 m difference). Despite minor modifications and combinations over the years, Logan's (1845) measurements and subdivisions form the basis for all later stratigraphic studies of the Carboniferous basin fill of the Cumberland Basin.

The stratigraphic relationship of the Lower Cove red beds to other units within the Basin (other than those in direct contact in the coastal exposures) has long been enigmatic. Brown and Smith (1829) did not distinguish the Lower Cove beds from their "Millstone Grit" and overlying "Coal Measures" (Fig. 2). Dawson (1855) originally assigned Divisions 5-8 to the "Lower or Older Coal-Formation", broadly analogous to the Lower Carboniferous, but later reconsidered and correlated these strata with the "Millstone-grit Series" (Dawson 1868). Grouping of Logan's Divisions into more broadly recognized lithostratigraphic units represented an attempt to correlate these stratal packages beyond the confines of the basin, but required an understanding of local sedimentation patterns within the tectonically active Cumberland Basin.

During the first half of the Twentieth Century, Bell (1912, 1914,1927,1938,1944,1958) studied the paleontology and regional stratigraphy of the Carboniferous of Nova Scotia. Well aware of the lithostratigraphic complexities of the Cumberland Basin, Bell felt that the basin fill could be more naturally subdivided on the basis of unconformities as indicated by the floral and faunal record. Bell's regional stratigraphy, grounded in his steadfast belief in unconformity-bound groups, became the definitive work of the Twentieth Century, but his combined use of bio- and lithostratigraphic definitions proved impractical. Influential advocates of a purely lithostratigraphic approach to the regional stratigraphy of the Maritimes Basin have been Belt (1964,1965) and Kelley (1967), and their viewpoint was followed, at least in part, by Ryan et al. (1991).

In his first round of stratigraphic revisions, Bell (1912,1914, 1938) grouped Logan's Divisions 3-5 and basal Division 2 into his "Joggins Formation". His grouping of the Lower Cove red beds (Division 5) with the overlying coal-bearing strata (Division 3-4) was based, in part, on the lack of contrary (or indeed any) macrofloral evidence within the red beds. Bell (1914, 1927) correlated the contact between coal-bearing strata and conglomerate at Spicer's Cove (Fig. 1) on the south limb of the Athol Syncline (Fletcher 1908) with the contact between Divisions 5 and 6 at Lower Cove. This correlation, later shown to be incorrect by Ryan et al. (1990a, 1990b), contributed to the impracticality of mapping his Shulie and Joggins formations inland (see map notes of Bell 1938). Consequently, Bell (1938) later advocated for an undivided Cumberland Series; later still, Bell (1944,1958) defined the Cumberland Group, comprising Divisions 1-5, and the Riversdale Group, comprising Divisions 6-7

In the mid Twentieth Century, efforts were renewed to map the basin fill. Central to this work was the goal of establishing stratigraphic relationships between the coastal section and the economically important coal measures of Springhill. In the course of this work, Shaw (1951) recognized the red beds of Lower Cove as a lithologically distinct subunit (10a) of his "coal-bearing facies" and considered the red beds to be coeval with basin-margin conglomerates (his map unit 9, the Polly Brook Formation of later authors) near Springhill (Fig. l). Copeland (1959) concurred, but grouped the Lower Cove red beds with the basin-margin conglomerates (his "Facies A").

On the strength of their interpretation that the coal-bearing strata of Spicer's Cove were correlative with the Joggins coals and were underlain by conglomerates exposed at the south of Spicer's Cove, Bell, Shaw and Copeland correlated the Lower Cove red beds with basin margin conglomerates. Palynostratigraphic studies (Hacquebard and Donaldson 1964; Dolby 1991) have demonstrated, however, that the Spicer's Cove section is younger than that of Lower Cove, a conclusion supported by the macrofloral record (R.H. Wagner, personal communication, 2000). A growing body of evidence further suggested that the contact between the Lower Cove red beds and the underlying Boss Point Formation was likely conformable (Hacquebard 1972; Howie and Barss 1975).

Later in the century, a subsequent round of mapping and stratigraphic investigations in the Cumberland Basin was undertaken by the Nova Scotia Department of Natural Resources (Ryan et al. 1990a,b,c, 1991; Calder 1995). The lithostrati-graphic distinctiveness and regionally mappable boundaries of the Lower Cove red beds were acknowledged by Ryan etal. (1990a) on their geological map, but this unit was nonetheless included with the coal-bearing strata of Division 4 and the basal 51m of Division 3 in their revised and reinstated Joggins Formation (Ryan etal. 1991). Members were not formally defined within this 1433 m-thick unit, although the informal term "Little River Bridge member" was used by Ryan etal. (1991, their fig. 5) and Ryan and Boehner (1994, their figs. 2-13) in reference to the Lower Cove red beds.

Here we formally propose the elevation of the Lower Cove red beds to the status of formation, and coin the name Little River Formation for these beds. Our proposed unit is defined purely on the basis of lithostratigraphy, being a non-coal-bearing, red-bed succession positioned conformably above the Boss Point Formation and conformably beneath the coal-bearing strata of the Joggins Formation. The formation is named after Little River, the mouth of which crosses the type section where it enters Chignecto Bay at Lower Cove. For the sake of clarity, it is important to note that this newly proposed unit bears no relationship to the now historically obsolete "Little River Group" of southern New Brunswick (Dawson 1868, p. 506), despite these strata being approximately time-equivalent. These strata are now assigned to the Lancaster Formation of Alcock (1938), and include the classic "Fern Ledges" of Stopes (1914).

The Little River Formation comprises a lithologically distinctive and regionally mappable unit. It lacks both coal and bivalve-bearing limestone beds, the two defining hallmarks of the overlying Joggins Formation (Ryan et al. 1991). It is also lithologically distinct from the underlying Boss Point Formation, which is characterized by thick, multistorey, well sorted grey sandstones, features absent from the Little River Formation. Similarly, although it is considered to be, in part, coeval with the Polly Brook Formation, it is lithologically distinct from that polymictic conglomerate succession. Both the base and top of the Little River Formation are regionally important surfaces.

The type section comprises outcrops in the intertidal zone and low-relief coastal bluffs on the eastern shore of Chignecto Bay, at Lower Cove, near Joggins, Cumberland County, Nova Scotia (Fig. 3). The section begins at the top (southern edge) of the South Reef of the Boss Point Formation (UTM Coordinates 5066450N, 388000E; NAD 83 datum) and extends 1500 m southward to a point 500 m south of the mouth of Little River (5063200N, 388400E). The top of the section coincides with the base of the lowest coal bed (Coal 45 of Logan 1845) of the Joggins Formation, near the start of the continuous cliff section.

Strata of the Little River Formation are known to occur inland on four stream sections east of the type section (Fig. 1; and see Ryan et al. 1990a). These include, from west to east: 1) exposures on a small tributary to the Maccan River north of the village of Maccan; 2) along Baird Brook at Chignecto; 3) near the contact with the Polly Brook Formation on Saint George's Brook east of Chignecto, which at this locality underlies the Little River Formation; and 4) on Styles Brook south of Stanley.

The basal contact of the Little River Formation is placed at the top of the highest multistorey sandstone bed of the Boss Point Formation, which at the type section is the top of the South Reef, coinciding with the base of Division 5 of Logan (1845). Inland, to the east, the base of the formation progressively onlaps against polymictic conglomerates of the Polly Brook Formation (Ryan etal. 1990a). The upper contact of the formation is placed at the base of the stratigraphically lowest coal bed, which at the type section is coal 45 of Logan (1845), 1.65m above the base of Logan's Division 4. This upper contact defines the base of the revised Joggins Formation (see Davies et al. 2005). The onset of grey mudrock and sandstone in the Joggins Formation provides a secondary criterion in mapping incompletely exposed sections. The formation has been mapped by the first author inland from the type section, its boundaries indicated on the map of Ryan etal. (1990a), where the formation, then unnamed, was designated "abundant redbeds"

The type section at Lower Cove, constituting a remeasure-ment of Division 5 of Logan (1845), is 635.8 m thick. The formation can be traced 30 km to the east along the north limb of the Athol Syncline to Styles Brook (Ryan et al. 1990a), where it eventually pinches out between the underlying Polly Brook Formation and overlying Joggins and Springhill Mines formations (Fig. 1). Because of many similarities, the Lower Cove beds may be laterally equivalent to fault-bound strata exposed across Chignecto Bay to the west on Maringouin Peninsula, New Brunswick, assigned to the informal Grande Anse formation (St. Peter and Johnson 1997), as well as to undesignated strata along strike of the Grand Anse beds at Minudie Point.

Inland, the Little River Formation is in part laterally time-equivalent to coal-bearing strata of the Joggins Formation and is inferred to be laterally equivalent to the Polly Brook Formation on the south limb and in the east of the Athol Syncline (Figs. 1, 4). Although it is probable that the Grande Anse formation in southeast New Brunswick may represent a fades of the Little River Formation, the correlation of the two units is problematic due to uncertainties in the stratigraphic relationship of the Grande Anse formation with adjacent geologic units. The lower contact of the Grand Anse section is in faulted contact with the Boss Point Formation at the coast, although the contact is reported to be conformable in the axis of the Hardledges Syncline south of the Shepody-Beckwith Fault (Johnson 1996). At its upper contact, the Grand Anse formation is in faulted contact with the Mississippian Windsor Group. Johnson (1996) suggested correlation of the Grande Anse with the Polly Brook Formation and lower Joggins (Little River of this paper) Formation.

Age determination of the Little River Formation is problematic due to its sparse paleontological record and, in particular, the complete absence of key age-diagnostic marine index fossils (Calder 1998; R.H. Wagner, personal communication, 2004). Furthermore, although palynology offers an alternative means to date strata, difficulties exist in disentangling evolutionary signatures from paleoecological effects. With these uncertainties in mind, comprehensive palynostratigraphic studies of the type section place the Little River Formation within the upper Namurian (Kinderscoutian) to basal Westphalian (basal Langsettian) according to Dolby (1991,2003), or entirely within the upper Namurian (Marsdenian to Yeadonian) according to J. Utting (personal communication, 2004). Extensive revision by R.H. Wagner of the macroflora of the overlying Joggins Formation demonstrated a floral assemblage consistent with the Langsettian of Europe, and Langsettian floral elements are present along with problematic "hinterland" taxa in the underlying Boss Point Formation (Utting and Wagner 2005).

The miospore Cananoropolis mehtae until recently was known only from strata in the informal "Coal Mine Point member" of the Joggins Formation (Ryan et al. 1991), and was used therefore as local index taxon. Its occurrence in red beds of the Grande Anse section on Cape Maringouin, New Brunswick, was cited as evidence of a late Langsettian age for these strata (Dolby 1999). Subsequent sampling of the formation at Lower Cove, however, yielded a much earlier, albeit solitary specimen of Cananoropolis mehtae (Dolby 2003), suggesting that it may be less useful as a local index taxon than previously assumed, and therefore giving cause to reconsider the late Langsettian age for the Grande Anse formation. Its presence within the Lower Cove and Grande Anse strata further suggests that the two formations are possible correlatives.

Twelve prominent (>2m-thick) channel bodies crop out on the wave cut platform as prominent "reefs", and vary in width from 31-534 m. Channel bodies have been assigned numerical designations (Fig. 3) that correspond to the stratigraphic position in metres of the top of their uppermost bed in the measured section. Channel bodies are described using the bounding surface and lithofacies terminology of Miall (1996). Smaller channel bodies (<2 m thick) also occur, but are dealt with separately below.

Major channel sandstone bodies of the Little River Formation represent the deposits of alluvial drainage channels. Channels are characterized by moderate incision, subsequent infilling and, where the rivers were shallow and broad, a final abandonment phase. These contrast markedly with channel deposits of similar size associated with red mudrock beds of the overlying Joggins Formation, which are incised and confined within shallow valleys (Davies and Gibling 2003). The channel forms of the Little River Formation facilitated aggradation, whereas sediment bypass and cannibalization typified the Joggins paleochannels. The flashy flow and rapid aggradation of the Little River Formation drainage channels are indicated by a sigmoidal barform preserved at 95 m (Fig. 5), which may represent a transitional antidune, and by sandy wings, which record periods of maximum discharge when flow topped levees (Friend et al. 1979). Sand-size clay grains, associated with this and other channel complexes, may represent mud aggregates or finely ground intraformational mud clasts indurated by drying and reworked from floodplain deposits.

The presence of petrocalcic paleosol horizons indicates pronounced rainfall seasonality and greater soil longevity than in the overlying Joggins Formation, which is characterized by weakly developed, immature paleosols (Smith 1991). Local gleyed horizons represent wetting or ponding on the floodplain, which may have been facilitated by induration of the dessicated mudrocks. Alternatively, rare Spirorbis-vich grey mudstone beds may indicate cryptic transgressive events. Although the ecological affinity of this fauna is poorly defined in the fossil record with respect to marine or freshwater environments, it is locally associated with brackish facies in the overlying Joggins Formation (Duff and Walton 1973; Skilliter 2001; Falcon-Lang 2005). Although the red beds generally reflect oxidizing, well drained conditions with sufficient sediment input to prevent extensive pedogenesis, drab haloed root compressions indicate that neither moisture deficit nor disturbance prevented plant life.

The reproductive and growth strategies of the most common plant groups within the Little River red beds show adaptation consistent with the sedimentological record of periodic moisture deficit and aggrading sediment. The persistence of calamiteans, perhaps the most ecologically resilient of the Late Paleozoic plant groups (Calder et al. in press; DiMichele et al. in press), was aided by their prolific vegetative propagation by adventitious roots and underground rhizomes (Gastaldo 1992). Medullosan pteridosperms and cordaitaleans were assisted through dry periods by a substantial root system, which allowed them to tap deep groundwater sources (Falcon-Lang and Bashforth 2004). Although sparse, the presence of lycopsid foliage and aerial stems in the upper two-thirds of the section indicates that standing water was available for their reproduction (Phillips and DiMichele 1992), at least temporarily or in restricted areas of the landscape.

Stratigraphic relationships indicate that the red beds of the Little River Formation were largely coeval with the conglomerates of the Polly Brook Formation at the southern basin margin, which represent extensive alluvial fan deposits derived from the Cobequid Highlands massif to the south (Calder 1991,1994). Paleoflow indicators within the Little River Formation at Lower Cove, however, indicate a dominant source to the northeast of the type section (potentially the Caledonia Highlands), which argues against the Little River Formation as distal deposits of the Polly Brook Formation alluvial fans. The immature, feldspathic composition of certain horizons within the channel sandstone bodies at 90 and 160 m suggests, however, that the Little River Formation may include distal deposits derived from fans along the Caledonia Highlands of southern New Brunswick. Equivocal support for this hypothesis comes from the Grand Anse strata to the north and east of the type section, which exhibit a coarser, more feldspathic composition than at Lower Cove. As discussed earlier, however, the age and fault boundaries of the Grande Anse cast uncertainty on correlation of the two units.

The clear evidence of prevailing dry conditions during deposition of the Little River Formation speaks convincingly of climate as an underlying control. However, the stratigraphic relationships within such an active basinal setting also require consideration of topographic factors that may have come into play in imparting the well-drained character of these red beds. Evidence supporting a dry or dry seasonal climate includes in situ pedogenic carbonate development, absence of coal beds, flashy paleoflow recorded within channel bodies, and the pervasive reddening of mudrocks and some sandstone bodies. The possibility of topographically related drainage effects comes indirectly from possible relationships with basin margin alluvial fan systems. Although calcareous rhizoliths are present in the Joggins Formation (Davies and Gibling 2003; Falcon-Lang et al. 2004), no in situ source beds are present in that interval, nor within the Joggins Formation. This raises the possibility of reworking from an upstream source, namely the Little River Formation red beds. The absence of bivalve-bearing beds, which record flooding events (Davies and Gibling 2003), suggests that base level was not felt within the Little River environment, with the possible exception of a single Spirorbis-rich interval. This further suggests that gradient may have enhanced drainage and so have been a factor in the development of the red beds. However, the general lack of evidence of incision by channel bodies within the Little River Formation, which would be expected if base level was lowered, does not support the enhancement of drainage by topographic relief.

In summary, while there is cause for considering topographic factors in draining the Little River drylands, the development of calcareous soils, complete absence of coal beds and supporting sedimentological evidence clearly required a climate regime with pronounced dry intervals. Topographically enhanced drainage may have played a contributing role.