Previous research

6 Ryan et al. (1991), Ryan and Boehner (1994) and Calder (1998) summarized the history of stratigraphic nomenclature for the Cumberland Basin, and Calder et al. (2005) set out more fully the history of stratigraphic work in the Joggins area. Previous approaches to subdividing the Joggins stratal interval are outlined in Figs. 3 and 4. Despite its magnificence, the coastal exposure provides only a two-dimensional view of the basinal strata, and a full regional understanding is complicated by minimal exposure inland, the presence of Chignecto Bay, and complex facies relationships southward towards the Cobequid Highlands (Fig. 1), which was an active, fault-bounded upland during deposition of the Joggins Formation. In proposing a stratigraphic framework, every researcher since Logan has wrestled with this difficult background.

Display large image of Figure1

7 Logan (1845) divided the Chignecto Bay strata into eight divisions (numbered from the top downwards), placing the predominantly grey, coal-bearing strata of the Joggins cliffs in Division 4. He measured this interval as 2539' 1" (774 m) of grey and red mudstone, sandstone, limestone, coal, carbonaceous shale and minor conglomerate. Logan recognized 45 coal groups (identified with numbers) in Division 4, and recorded an aggregate thickness of 37' 9.5" (11.5 m) of coal and 23' 3" (7.1 m) of limestone in the division. Red, red-green and chocolate shale and sandstone constituted 264 m (34%) of his section, the remainder being grey. In contrast, he noted the absence of coal in the underlying Division 5, which he described as red mudstone and grey and red sandstone, with a few green shale and calcareous nodular horizons. This division, only partially exposed at Lower Cove, was measured at 2082'(635 m) down to the top of "South Reef", a prominent and thick sandstone body within Division 6. Logan's Coal Group 45 lies slightly above the base of Division 4 at Lower Cove, and was described as comprising 10' 2" (3.1 m) of coal, carbonaceous shale and interbedded shale and sandstone, with a basal thin coal 3" (7.5 cm) thick. These strata are underlain by 5' 6" (1.7 m) of underclay and red shale and sandstone, the base of which marks the base of Division 4. The top of Coal Group 1 marks the top of Division 4 near Joggins village, where a 4' (1.2 m) limestone rests upon a 1' (0.3 m) coal. The overlying Division 3 has a lesser proportion of coal (22 seams with only 5' 5" or 1.7 m aggregate thickness), a greater proportion of red beds and, most significantly, no limestone. Division 2 is mainly red sandstone and mudstone.

Display large image of Figure 2

8 Subsequent workers used aspects of Logan's classification. Dawson (1854,1855) divided Division 4 into 27 coal divisions (designated with Roman numerals), and provided more detail for some of Logan's coal groups. Dawson (1868) adopted aspects of British terminology in identifying a Millstone Grit Series and a Middle Coal Formation, a system that was modified by Bell (1912). Bell (1914) subsequently proposed the name "Joggins Formation" for Divisions 5,4,3 and part of 2, an interval 6886' (2099 m) thick, and the name "Shulie Formation" for overlying strata (most of Division 2 and Division l). Bell discussed the rationale for combining the Division 5 red beds and the Division 4 grey, coal-bearing beds into one formation, inferring that a regional disconformity underlies Division 5 within the Cumberland Basin (although it is not apparent in the coastal section). He subsequently retracted the formation names (Bell 1943) due to problems in correlating units inland, arising in part from miscorrelation of the coal-bearing strata of Spicers Cove with those of Division 4 to the north at Joggins. Bell combined the Joggins and Shulie formations as the "Joggins member" of an undivided Cumberland Group. Shaw(l951a,b) mapped coarser and finer rock units within the basin, and assigned the Joggins grey strata to his unit 10, noting that conglomeratic wedges near the Cobequid Highlands thin northward and complicate the basinal stratigraphy. Copeland (1959) followed this approach, and confirmed that coals thin inland within the Joggins-Chignecto coalfield. Belt (1964) referred strata at this level across the region to the "Coarse Clastic Fades", but Kelley (1967) retained the Cumberland Group for the lower part of this interval.

Display large image of Figure 3

9 Ryan et al. (1991) reinstated the Joggins Formation and presented a formal description with the Joggins cliffs as the type section. Ryan et al. (1990) mapped the formation inland, and included Divisions 5,4 and the basal strata of Division 3 within the reconstituted Joggins Formation, for a total thickness of 1433 m. Both Ryan et al. (1991) and Ryan and Boehner (1994) illustrated member boundaries on a column derived from Logan's section (their figs. 5 and 2-13, respectively), but did not elaborate on them in the text. The figures show three informal members: the Little River Bridge member (Division 5), the Coal Mine Point member, and the Bells Brook member (Division 4 and basal strata of Division 3), but these members were not formally defined or shown on the accompanying maps.

10 In the coastal section, Logan's Division 5/6 boundary was taken to mark the base of the Joggins Formation, and its top was placed at the base of a thick sandstone body that ushers in a more sand-dominated interval at the base of their new Springhill Mines Formation (Ryan et al. 1991; Ryan and Boehner 1994). This upper boundary was selected by Ryan et al. (1991) who placed the basal 168' 2" (51 m) of Division 3 within the Joggins Formation. This contact has proven difficult to map.

Modifications to Joggins Formation definition at the type section

11 Remeasurement of the section (Appendix A of this paper, and Appendix A of Calder et al. 2005) has led Calder et al. (2005, their Appendix C) to redefine the Joggins Formation from the boundaries set out by Ryan et al. (1991) (Fig. 4). Below, we describe the boundaries of the revised formation at the coastal type section as they relate to our recent sedimentological investigations.

Open-water facies assemblage

19 This assemblage represents major flooding events, some of which probably inundated most of the western Cumberland Basin. Especially good examples in the lower part of the formation (cycles 2-4) are represented by a thin coal overlain by a dark limestone up to 1 m thick, which is overlain in turn by several metres of grey siltstone capped abruptly or gradationally by sandstone. The limestones are well cemented and stand out on the foreshore. They are locally termed "clam coals" because of their dark colour and the abundance of bivalves with accessory ostracodes, spirorbids, arthropods, disarticulated fish and plant fragments.

20 The overlying siltstones are laminated and platy weathering, and contain discoidal siderite nodules. The siltstones contain bivalves and ostracodes (generally confined to discrete levels), as well as drifted plant material. Agglutinated foraminifera were obtained from some samples (Archer et al. 1995). Capping the siltstones are sharp-based, sheet-like sandstones a few metres thick, which extend across the cliff and foreshore and are characterized by planar bedding and a flaggy appearance. In a few instances, they comprise overlapping mounds up to 100 m in apparent width. The sandstones contain unidirectional ripple cross-lamination, local mud drapes, and lineated plane beds, with wave ripples and rare hummocky cross-stratification indicating wave activity. Trace fossils include delicate grazing and walking traces (Archer et al. 1995) and, less commonly, resting traces (for example of limulids). A few channel bodies cut the planar sandstones, with which they are evidently closely associated. The topmost coarser beds contain roots, which mark the re-establishment of subaerial conditions after the initial flooding event.

21 The assemblage represents the establishment across the basin of a restricted-marine gulf, perhaps similar to the modern Baltic Sea in its partially enclosed nature and variable but generally low salinity (Grasshoff 1975). Open-marine faunal elements have not been observed, but the presence of certain taxa of bivalves, ostracodes, foraminifera and trace fossils suggest at least brackish conditions (Bell 1914; Duff and Walton 1973; Archer et al. 1995; Skilliter 2001; Tibert and Dewey 2005). Strontium isotope data from fish material also suggest marine influence (Calder 1998), although mineralogical and geochemical data from some bivalve shells are consistent with a freshwater setting (Brand 1994). During some flooding events at cycle bases, the presence of a basal coal suggests that peat formation initially kept pace with rising water level. When the rate of sea-level rise exceeded that of peat accumulation, the area was transformed into a shallow bay where faunal-concentrate layers accumulated. The upward change to siltstone indicates the renewed advance of the coastal plain as the rate of sea-level rise decreased. Progradation culminated in shallow-water sands that represent thin shorefaces and small delta lobes derived from the associated channels. Wave activity was prominent, but sedimentological evidence for tidal influence is restricted to mud drapes at a few levels (Skilliter 2001).

22 Drifted lycopsid plants predominate in the basal limestones, whereas overlying siltstones and sandstones contain a mixed suite of drifted gymnosperms (cordaitaleans), sphenopsids (primarily calamiteans), pteridosperms and putative pro-gymnosperms (Falcon-Lang 2003a). The high proportion of progymnosperms and gymnosperms suggests that the basin floor was almost entirely drowned, greatly reducing the area of coastal swamps. Under these conditions, elements of the upland vegetation (Falcon-Lang and Scott 2000) brought to the sea by rivers were preferentially concentrated in the siltstones (an example of the "Neves Effect" of Chaloner 1958). The presence of stigmarian roots within some limestone beds points to near emergent, shallow conditions in some cases, and maximum water depth during deposition of open-water facies was probably only a few metres to a few tens of metres.

Display large image of Figure 6a

Display large image of Figure 6b

Display large image of Figure 7

Poorly drained floodplain assemblage

23 Joggins is justifiably famous for its poorly drained floodplain deposits, which contain the spectacular fossil forest horizons. Sandstone and green/grey mudstone (commonly intensively rooted), are accompanied by coal, carbonaceous shale and minor limestone, with siderite nodules. Bivalves and ostracodes are generally less common than in the open-water assemblage. The assemblage includes thin grey carbonaceous shales that separate red beds near cycle tops from coal and limestone at the base of the next cycle.

24 Especially prominent are sheet-like, heterolithic units of sandstone and mudstone, several metres thick, which extend across the cliffs and foreshore and contain many entombed erect trees. Splendid examples include two forested intervals below the Fundy Seam (404-420 m; Calder et al. in press), two intervals just below the Forty Brine Seam (539-546 m) and the lower reef just north of Coal Mine Point (628-630 m). Charles Lyell was struck by the preserved height of the trees, which he estimated to be up to 7.6 m tall (Lyell 1842,1845). The maximum height observed over the past three decades has been 6 m (Calder et al. in press). Vegetation includes abundant lycopsids and sphenopsids in situ (Falcon-Lang 1999; Calder et al. in press) with a compression macrofloral record comprising cordaitalean gymnosperms, pteridosperms, ferns and putative progymnosperms (Calder et al. in press). Many standing trees are bordered by scour fills of sandstone up to 2 m thick with centroclinal (inward dipping) cross-stratification, and suites of large sandy mounds (vegetation shadows) are common (Rygel et al. 2004).

25 Within this assemblage, Lyell and Dawson made their remarkable discovery of fossils within tree trunks, mainly from the lower reef ("Lesser Reef" of Dawson 1882) at Coal Mine Point. The tree-stump fauna is highly diverse, including eleven tetrapod genera, abundant coprolites, and a variety of invertebrates - land snails (Dendropupa), millipedes, arthropod fragments, and calcareous shells of the annelid Spirorbis (Carroll et al. 1972). Charcoal fragments are abundant within and adjacent to some trunks, testifying to wildfires that swept the forests (Falcon-Lang 1999). The organisms may have taken refuge within hollow trees or may have become entombed accidentally, and some fragments may have been swept in by floods (Lyell and Dawson 1853; Carroll et al. 1972; Scott 2001).

26 The heterolithic units are closely associated with narrow channel bodies up to 3 m thick. A few much larger channel bodies are also present in the assemblage. One large channel body at 580 m has incised 9 m through rooted grey mudstone and contains two vertically stacked storeys. The major sandstone "reef" at Coal Mine Point (637-648 m) is a channel body that contains trough cross-beds, scroll-bar forms and lateral accretion surfaces (Rygel 2005). This channel body contains especially spectacular examples of Diplichnites (large trackways attributed to the arthropod Arthropleura: Ferguson 1975).

27 Many of the coal groups (marked on the section in Appendix A) lie within this assemblage, which includes the main economic seams of the formerly worked Joggins-Chignecto Coalfield: most notably the Fundy (coal 29a), Forty Brine (coal 20), Kimberly (coal 14), Queen (coal 8) and Joggins (coal 7) seams. The bituminous coals are sulphur-rich (Copeland 1959; Skilliter 2001) with prominent mudstone partings and locally high concentrations of Zn, Pb and As (Hower et al. 2000).

28 The strata were deposited in wetlands akin to those of the modern Mississippi Delta (Coleman and Prior 1980; Tye and Coleman 1989), although the geomorphic form of the coastal system is not known. Small distributary channels traversed the coastal plain and brought sand and mud to the adjacent fresh to brackish bays during repeated flood events, depositing characteristically heterolithic sediment as interdistributary crevasse splays and bay fills. These sedimentation events entombed the standing trees (Calder et al. in press) and created scour hollows and vegetation shadows around the trunks (Rygel et al. 2004). At the level of the upper Fundy forest (419 m level in Appendix A), thin sandstone sheets can be traced from the margins of channel bodies on the tidal platform into scour fills around standing trees exposed in the cliffs. The forests were affected by wildfires that may have been instrumental in hollowing out the trunks and providing shelter for early tetrapods and other organisms. The channel body at the 580 m level is interpreted as a large distributary channel, based on its incised nature and aggradational style, whereas the Coal Mine Point channel body represents a meandering river that advanced over bayfills, much as the modern Atchafalaya River of Louisiana advanced rapidly once it had filled Atchafalaya Bay (Tye and Coleman 1989). Thin poorly drained intervals at cycle tops represent incipient drowning of the coastal zone prior to the main transgression.

29 The coals represent planar (groundwater-fed) mires (Hower et al. 2000; Calder et al. in press) where, for prolonged periods, peat accumulated away from detrital supply, although floods periodically generated muddy partings. Several economic seams cap heterolithic units or channel bodies, suggesting that the precursor peat accumulated in freshwater settings following abandonment of a local distributary. These thick coals may be the updip equivalent of marine flooding events, and the high sulphur levels of many coals suggest that marine, sulphate-rich waters influenced the peats, probably after the mires were drowned by sea-level rise. Many drab mudstones are hydromorphic paleosols that formed under variable redoxymorphic conditions, and red and red/grey mottled intervals testify to episodes of soil formation under oxidizing conditions (Smith 1991). The 595-612 m interval of cycle 9 contains stratified red and grey mudstone without coal or invertebrate fossils, suggesting that oxidized mud was washed into clastic-dominated lakes.

Well drained floodplain assemblage

30 This predominantly red bed assemblage comprises red mudstone and sandstone, with minor grey mudstone, rare coal and ostracode-bearing limestone. Although not highly fossiliferous, these strata have recently yielded some unusual fossil discoveries (Hebert and Calder 2004), as outlined below. Cycle 4 contains an especially thick red bed interval.

31 Channel bodies are narrow and up to 6 m thick with an aggradational style of filling, and in places several bodies lie at the same stratigraphic level in the cliffs (Rygel 2005). Most are associated with heterolithic sheets of sandstone and mud-stone that typically thin away from the channel bodies and represent levee and crevasse splay complexes. The channels are filled with grey and red sandstone and mudstone, with local conglomerates composed of reworked carbonate (paleosol) fragments. Within cycles 1 and 3, some large channel bodies contain smaller channel fills, suggesting that the bodies are small dryland valleys. The red muds are poorly stratified and contain scattered calcareous nodules, although petrocalcic horizons were not observed.

32 Standing trees are restricted to poorly preserved stump casts, and narrow hollow fills with abundant underlying roots marking the former positions of trees, since decayed (Rygel et al. 2004). Charcoal and other floral remains in channel bodies are dominated by cordaitaleans, with minor pteridosperms, sphenopsids and lycopsids - the minor constituents typically confined to channel-margin situations (Falcon-Lang 1999, 2003b; Falcon-Lang and Scott 2000). One channel body at 270-274 m, known informally as the "Hebert beds", contains abundant charcoal as well as tetrapod material, shells up to 23 cm long of the unionid bivalve Archanodon, and land snails (Dendropupa) (Falcon-Lang et al. 2004a; Hebert and Calder 2004). Other channel bodies have yielded large arthropod trackways (Diplichnites).

33 The assemblage represents the alluvial plain of a seasonal dryland traversed by suites of narrow channels that probably had an anastomosing planform (Rygel 2005), as indicated by multiple, narrow channels at similar levels connected by sheet sandstones ("ribbon tiers" of Kraus and Wells 1999). The setting may have resembled that of the Channel Country of Australia with its dryland anastomosing systems and water holes (Gibling et al. 1998); Rust et al. (1984) also drew on the Channel Country in interpreting the overlying Springhill Mines Formation. The red floodplain muds are immature, cumulative paleosols that formed under a humid seasonal climate (Smith 1991). The abundance of cordaitalean charcoal in some dryland channels suggests that the seasonally dry floodplains were covered with a fire-prone and ecologically stressed assemblage dominated by gymnosperms (Falcon-Lang 2003b; Falcon-Lang et al. 2004a). Riparian (channel-margin) settings permitted the local growth of vegetation more akin to the wetlands, and wildfires were common, perhaps promoted by elevated levels of atmospheric oxygen (Robinson 1991). The unusual biota preserved within the "Hebert beds" suggest that the parent channels provided intermittent water holes where organisms continued to flourish during seasonal low-stage flow or more prolonged droughts (Falcon-Lang et al. 2004a).

Cyclic patterns

34 The 14 cycles recognized in the 915.5 m of the Joggins Formation range in thickness from 16 to 212 m, averaging 65 m. Nine cycles are 16 to 52 m thick, three are 52 to 99 m thick, and the remaining two are 158 m and 212 m thick, respectively. Facies distribution was used to construct a relative base-level curve for the formation (Davies and Gibling 2003; Fig. 7). Unfortunately, the present lack of firm biostratigraphic boundaries and absolute dates for the section precludes the determination of cycle durations and accumulation rates.

35 Relatively straightforward facies patterns are evident in cycles 1 to 4 and in the basal part of cycle 5. These intervals show a systematic upward succession from open-water facies, which mark major transgressions, to poorly drained and well drained facies, marking regressions. At some levels, thin occurrences of poorly drained facies below open-water intervals herald the base of the next cycle, denoting the onset of base-level rise. Limestones and platy siltstones are prominent, coals are thin and mainly underlie limestones, and the prominent sharp-based sandstones that cap open-water deposits contain many trace fossils. Cycle 5 constitutes the most prolonged period of red bed accumulation, with alternate periods of poorly and well drained conditions and some thin coals in the upper 80 m.

36 The most marked and sustained lithological change within the formation is the relatively abrupt change from well drained to poorly drained floodplain deposits at the base of cycle 6 (Fig. 7), with a suite of prominent coals; future mapping inland may provide justification for identification of a member boundary at this level. Cycles 6 to 8 are of moderate thickness, and usher in a period when the area was dominated by coastal wetlands (represented by the poorly drained assemblage), with only thin intervals of open-water deposits. Coals are numerous and thick, and many of the most prominent fossil forests are found in this interval (Calder et al. in press). Limestones are generally scarce, apart from a thick bed at the base of cycle 8. Cycle 9 commences with a well developed occurrence of open-water facies above the Forty Brine Seam (Skilliter 2001), with limestones, mud drapes, trace fossils (Archer et al. 1995), and a large distributary channel body, passing upward into probable lacustrine deposits of stratified red and grey beds.

37 Cycle 10 (158 m thick) marks the start of a 200 m interval of alternate poorly drained and well drained deposits without open-water sections and limestones (cycles 10-12). Prominent sets of fossiliferous carbonaceous shales (cycle 10) or thick coals (Queen and Joggins seams, cycles 11 and 12) mark cycle bases, but several thin coals delineate minor transgressions within the numbered cycles. Limestones mark the base of cycles 13 and 14 and the Joggins / Springhill Mines formation contact.

38 Following the abrupt onset of wetland conditions at its base, the Joggins Formation records a punctuated set of advances and retreats of the coastal zone (Fig. 7). A long-term balance seems to have been maintained between accommodation creation and sediment supply, such that the study area remained close to the coastal zone for much of Joggins Formation time, with periods of more sustained open-water, wetland or dryland conditions. Thick limestones and thick coals tend to be mutually exclusive (Fig. 7): thin coals underlie many limestones, but thick coals rarely have limestone caps, the most notable exception being the Forty Brine Seam (coal 20). This pattern probably reflects variations in the magnitude and rate of base-level rise. Large base-level rises would tend to flood much of the Cumberland Basin, resulting in reduced sediment flux to open-water areas and the accumulation of fossil-concentrate limestone. Rapid base-level rise would tend to outpace the rate of peat accumulation, resulting in thin peats only. In contrast, thick peats (coals) probably accumulated where modest or slow base-level rise caused prolonged freshwater ponding inland of transgressive shorelines (Rosters and Suter 1993). A rheotrophic (groundwater-influenced), planar character is the hallmark of the coals of the Joggins Formation (Hower et al. 2000; Calder et al. in press).

39 Sand accumulated preferentially in the poorly drained flood-plain assemblage, where coastal bays formed repositories for coarse detritus. In these areas, sand deposition was strongly focused into sheets, scour fills and vegetation shadows where forested landscapes slowed overtopping flood waters (Rygel et al. 2004). In contrast, shoreface and delta-lobe sands of the open-water assemblage are relatively thin, and dryland alluvial plains of the well drained assemblage include thick mud intervals, with sands restricted to small channels, levees and splays.

Sequence stratigraphy

40 Many Carboniferous cycles (or cyclothems) reflect sea-level fluctuations in the order of tens of metres in amplitude caused by the accumulation and melting of ice sheets in high southern latitudes (Crowley and Baum 1991; Maynard and Leeder 1992; Soreghan and Giles 1999). Glacioeustasy in Carboniferous basins has commonly generated stacked Exxon-type sequences with prominent sequence boundaries, valley fills, maximum flooding surfaces and systems tracts (Hampson et al. 1999; Gibling e£ al. 2004). Such expressions of glacioeustasy maybe modified under conditions of unusually rapid subsidence, as at Joggins, where the record of sea-level fall may be suppressed and the record of sea level rise may be strongly augmented, rendering the basin susceptible to basin-wide flooding events marked by fauna-rich horizons. At Joggins in the western Cumberland Basin, rapid subsidence reflects the extensional basin setting, coupled with active withdrawal of Windsor salt (Waldron and Rygel 2005). In consequence, the Joggins cycles display what we consider to be a "tectonically controlled architecture" characterized by multiple flooding surfaces (Davies and Gibling 2003), including coal and fossiliferous limestone at cycle bases that mark important episodes of sea-level rise. The overlying strata lack clear evidence for sea-level fall such as profound valley incision or well developed paleosols, although sharp-based shoreface and delta-lobe sandstones may reflect in part modest falls of sea-level (similar to those documented by Plint 1988). Large channel bodies appear to represent distributary channels and meandering rivers within a coastal plain setting, rather than recording profound basinward facies shifts that could have emplaced proximal (braided or low-sinuosity) river deposits over marine deposits. Small valley fills in cycles 2 and 3 appear to lie within red bed intervals, and need not imply major basinward shifts of facies belts linked to base-level lowering. The Joggins cycles may record glacial-interglacial transitions manifested in an equatorial setting, periods of varied subsidence rate as faults moved and salt migrated, variations in sediment flux, or combinations of all three. However, regardless of the ultimate cause of the cycles, their unusual architectural features are inferred to reflect extremely rapid subsidence, in accord with observations in other high-subsidence settings worldwide (Davies and Gibling 2003).

41 In the absence of sequence boundaries, the Joggins cycles can be categorized as parasequence sets, in which the predominant coastal-plain facies are composed of numerous thin parasequences and bounded by flooding surfaces marked by limestones, coals and carbonaceous shales. Thin drab intervals at cycle tops mark retrogradational parasequence sets that culminated in profound flooding at the start of the overlying cycle. As coastal rivers re-advanced, thick progradational parasequence sets accumulated where tropical wetland deposits filled marine embayments, until a dryland alluvial plain was established. Thereafter, alluvial red beds accumulated, flooding surfaces become fewer and less prominent, and trends of pro-, retro- or aggradation are difficult to establish.

42 Because subsidence was so rapid, a remarkably complete record of environments and the organisms that inhabited them is preserved in the Joggins cycles. In particular, prolonged periods of wetland conditions, during which sedimentation kept pace with subsidence, promoted the repeated generation and burial of forests at many levels (Waldron and Rygel 2005).