Journal of the Geological Society, London, Vol. 161, 2004, pp. 969–981. Printed in Great Britain.

Early Mississippian lycopsid forests in a delta-plain setting at Norton, near Sussex, New Brunswick, Canada H . J. FA L C O N - L A N G Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK (e-mail: [email protected]) Abstract: Mississippian lycopsid forests in growth position are extremely rare, and their community-scale ecology remains enigmatic. This is a significant gap in our knowledge, not least because they represent the precursors of Pennsylvanian ‘Coal Forests’. In this paper, nearly 700 in situ fossil trees are described from 13 entisol or inceptisol horizons in the mid-Tournaisian Albert Formation (Horton Group) at Norton, near Sussex, New Brunswick, Canada. These trees, almost all of which are lycopsids of the Protostigmaria– Lepidodendropsis-type, are rooted mostly in the flood-disturbed interdistributary wetland deposits of prograding wave-dominated deltas. Tree mapping on extensive palaeosol surfaces indicates the existence of extremely dense forest vegetation. Scaled up to standard forestry units, densities of 10 000–30 000 trees per hectare are inferred. A significant inverse linear relationship between tree diameter and density for the five most extensive palaeosols indicates that inter-tree competition led to natural self-thinning as the forests matured. Forest maturation also led to a reduction in tree-spacing heterogeneity. Flood events regularly killed whole stands at Norton, burying trees in sandstone sheets, and preventing establishment of climax vegetation. Charcoal remains demonstrate that wildfire was another important disturbance process. Keywords: Tournaisian, deltaic environment, lycopsids, palaeoecology, charcoal.

Sandstone-cast lycopsid trees in growth position are a very common feature of Pennsylvanian-aged strata in North America and Europe, and have been studied for over two centuries (Gastaldo 1986a). These spectacular upright specimens of Lepidodendron sensu latu and Sigillaria with attached Stigmaria rootstocks (Scott & Calder 1994) provide an evocative record of the vast tracts of peat-forming tropical rainforests that stretched in a nearly continuous belt from Kentucky to the Urals (DiMichele & Phillips 1994). As trees are preserved in their original growth context, they provide a wealth of important palaeoecological data (Calder et al. 1996). In contrast, Mississippian-aged fossil trees in growth position are extremely rare. Furthermore, almost all of these in situ fossils have been described in well-drained alluvial-plain deposits rather than wetland facies, and therefore dominantly consist of droughtadapted gymnosperm trees (Long 1979; Galtier et al. 1994). Mississippian wetland facies containing erect lycopsid trees that represent the precursors of Pennsylvanian ‘Coal Forests’ (Pigg 1992, 2001) have only rarely been described, and never with palaeoecological detail (Jennings 1975; Jennings et al. 1983; Beeler 1986; Scheckler 1986a, b; Martel & Gibling 1991). This paper fills this gap by describing numerous, superbly exposed, Early Mississippian fossil lycopsid forests from poorly drained delta-plain facies in eastern Canada (Fig. 1a), five of which contain tens to hundreds of individual trees. Multidisciplinary studies encompassing sedimentary facies analysis, tree distribution mapping, megafloral identification, plant taphonomy and palynodebris analysis are combined to shed light on the previously enigmatic nature of Mississippian forested wetlands.

B

N

Prince Edward Island New Brunswick

Fi

dy

f

yo Ba

A

e2

r gu

n Fu

cotia

Figure 1B

A Atlantic Ocean

200 km

aS Nov

Canada Phillips Inlet Canada West

Fig. 1. Location map. (a) Canada. (b) Map of the Maritime Provinces locating the position of SE New Brunswick.

2a). Site 1 is located at 166–167 km, Site 2 at 171 km, Site 3 at 173–174 km, Site 4 at 175 km and Site 5 at 176 km (Fig. 2b). At all these sites, bedding is subvertical (83–908), and strikes subparallel to the road (c. 010–0208). In addition, large-scale isoclinal folds and normal faults with throws exceeding 10 m locally occur. Consequently, it is extremely difficult to correlate adjacent road cuts precisely (Pickerill et al. 1985). The composite sedimentary succession exposed belongs in its entirety to the Lower Mississippian Albert Formation (Horton Group; van der Poll 1995). The Albert Formation at Norton was deposited on the NW

Geological setting The lycopsid fossil forests are exposed in five consecutive roadside sections along the eastbound carriage of Highway 1 between the 166 and 176 km marker signs, adjacent to Norton, near Sussex, New Brunswick (458379N, 658429W, Figs. 1b and 969

970

H . J. FA L C O N - L A N G

Fig. 2. Geological context. (a) Geological map of the complex half-graben Moncton sub-basin, SE New Brunswick, showing the distribution of the Mississippian Horton Group (after van der Poll 1995). (b) Detailed geological map of the study area near Norton. Circled numbers indicate five roadcut sites along Highway 1 on the basin margins (after Pickerill et al. 1985).

not related easily to adjacent successions in the Moncton subbasin (St. Peter 1993). However, structural and lithofacies data indicate that the sections studied in this paper most probably correlate with the Dawson Settlement Member (Fig. 3; Griener 1962; Carter & Pickerill 1985a; Ferguson & Fyffe 1985). Palynological data indicate a mid-Tournaisian (Courceyan; c. 355 Ma) age for the Norton beds, with the strata being no younger than the base of the Spelaeotriletes cabotii miospore zone (Utting 1987; Martel et al. 1993; St. Peter 2003).

margins of the Moncton sub-basin (Fig. 2a), one of several, narrow, interconnected, NW–SE-trending half-grabens that rifted in SE Laurasia during the Late Devonian following the oblique convergence of Gondwanaland (Calder 1998; Pascucci et al. 2000). This formation thickens from 170 m near the basin margin to 1800 m at the basin centre, and is characterized by a spatially and temporally complex pattern of sedimentary facies (Gussow 1953; Smith & Gibling 1987). Facies associations include alluvial fans, well-drained alluvial plains, hypersaline saltpans, algal lagoons, wave-dominated deltas (Greiner 1962; Carter & Pickerill 1985a), and brackish embayments, which were connected to the open Tethys Ocean via a narrow strait (Tibert & Scott 1999). As a consequence of intertonguing facies, the five lithostratigraphic members of the Albert Formation erected for the Sackville subbasin successions (Carter & Pickerill 1985b) are

Sedimentary facies and plant assemblages Fossil plant-bearing successions at Norton are composed of nine sedimentary facies organized into three associations, Facies Association A being deposited in wave-dominated delta-front settings, Facies Association B in poorly drained delta-plain

Mississippian Pennsylvanian

NORTH Stephanian

Pictou Group

Moncton Formation

Weldon Member

Westphalian

Gautreau Member

Riversdale Group Namurian

Hopewell Group

Visean

Windsor Group

Tournaisian

Horton Group

Dev. Famennian

Metamorphic basement

SOUTH Hillsborough Member

Albert Formation

Hiram Brook Member Round Hill Member Frederick Brook Member Dawson Settlement Member

Memramcock Formation

Undivided

Fig. 3. Stratigraphy of the Moncton subbasin, New Brunswick, illustrating the intertonguing nature of lithostratigraphic members in the Albert Formation along a north–south transect (after Carter & Pickerill 1985b; van der Poll 1995; St. Peter 2003). Star indicates the probable position of the Norton sites in the regional stratigraphy.

M I S S I S S I P P I A N LY C O P S I D F O R E S T S

settings, and Facies Association C in well-drained alluvial-plain settings (Fig. 4).

Facies description Facies Association A comprises Facies 1–4, which are organized into punctuated coarsening-upward successions, 4–7 m thick. Facies 1 consists of well-laminated, pyrite-rich, medium to dark grey, organic-rich mudstone units, 4–60 cm thick, that exhibit millimetre-scale siltstone laminations. Fish fragments, including

971

the palaeoniscid Elonichthys and abundant isolated scales, occur at some intervals (Miller & McGovern 1997). An ichnocoenosis comprising Paleodictyon, Cochlichnus, Gordia, Helminthoposis, Palaeophycus and Planolites has also been described from this facies at Norton by Pickerill (1990). Prone lycopsid stem fragments, up to 12 cm diameter, rarely occur. Most are decorticated, but some are identifiable as Lepidodendropsis corrugata Dawson (Fig. 5a); a few specimens exhibit isodiametric branching. Facies 2 consists of light grey, heterolithic siltstone and fine-

Fig. 4. Sedimentary logs of two measured sections near Norton (Sites 2 and 4) that contain abundant upright lycopsid trees.

972

H . J. FA L C O N - L A N G

Fig. 5. Allochthonous fossil plants, Albert Formation, near Norton. (a) Well-preserved Lepidodendropsis corrugata Dawson axis; scale bar represents 5 mm. (b) Partially decorticated lycopsid axis; scale bar represents 20 mm. (c) Leafy lycopsid terminal shoot; scale bar represents 40 mm. (d) Archaeocalamites stem; scale bar represents 25 mm.

grained sandstone units, 30–76 cm thick, that locally contain symmetrical ripples and arthropod traces, Cruziana problematica Schindewolf and Ruzophycos carbonarius Dawson. Soft-sediment deformation is locally present. Facies 3 consists of fine- to very coarse-grained sandstone sheets, 14–167 cm thick, that contain ubiquitous symmetrical ripples and have a planar to undulatory geometry. Partially decorticated lycopsid stem fragments are common (Fig. 5b), oriented subparallel to the symmetrical ripple crests, together with coalified and charred plant debris. Three tree stumps belonging to putative gymnosperms occur in growth position at two intervals (see below). Diplichnites traces occur on top of another sandstone sheet (Armitage et al. 2002). Facies 4 consists of 122–187 cm thick, erosive-based, coarseto very coarse-grained, broadly channelized sandstone units that contain both symmetrical ripples and trough cross-beds, up to 30 cm thick. Units may show a slight upward coarsening. Rill marks, tool marks and flute casts occur on the base of one unit. Prone lycopsid stem fragments (4–13 cm diameter), a few attributable to Lepidodendropsis, are common. Facies Association B comprises 8–11 m thick successions of Facies 5–7. Facies 5 consists of 120–227 cm thick, erosive-based pebbly arkosic sandstone beds that in some cases exhibit a broadly channelized form, but more commonly have flat bases over tens of metres. Units fine upwards from pebbly, very coarsegrained sandstone to fine- to medium-grained sandstone at the top. Trough cross-beds, up to 30 cm thick, and plane bedding are the dominant sedimentary structures, with ripple cross-lamination common only near the top. Inclined strata, 1–2 m high, were observed in a few channels, the dip of the inclined beds being approximately perpendicular to that of the trough cross-beds. Mud intraclasts, up to 35 cm in diameter, are also common. Decorticated lycopsid axes, up to 19 cm in diameter, charcoal and coalified plant debris occur locally, especially in the channel base. Facies 6 consists of 3–46 cm thick, poorly to well-laminated,

rooted, medium grey mudstone beds that locally contain common lycopsid remains including autochthonous Protostigmaria eggertiana Jennings rootstocks (see below), prone Lepidodendropsis stems, and terminal leafy shoots (Fig. 5c). Rare, prone sphenopsids stems, Archaeocalamites scrobiculatus Schlotheim (Fig. 5d), and gymnosperm pinnules, Aneimites acadia Dawson, also occur (Bell 1960). Facies 7 consists of fine- to coarse-grained sandstone sheets, 7–81 cm thick, that exhibit ripple cross-lamination, gently erosive bases, and a lenticular geometry over tens to hundreds of metres. Many units show a slight upward coarsening and preserve abundant sandstone-cast lycopsid trees in growth position; most trees are decorticated but a few conform to Lepidodendropsis (Fig. 6c and h). Trees may be tilted locally (typically 60–908; i.e. subvertical), the direction of tilt showing a strong preferred orientation, parallel to the palaeoflow direction indicated by the ripple cross-lamination. In some cases, horseshoeshaped scours occur on the upper surface of the sandstone sheets, positioned around the upright trees (Armitage et al. 2002). Autochthonous P. eggertiana rootstocks locally occur on top of the sandstone sheets (Fig. 6b). Facies Association C comprises 2–9 m thick red-bed successions of Facies 5, 8 and 9. Facies 5 has been described above. Facies 8 consists of blocky, red mudstone, 8–71 cm thick, containing centimetre-scale siltstone beds and rain imprints. Green–red mottling typically characterized by dendritic reduction zones (root traces) occurs in some intervals, whereas in others scattered pedogenic carbonate nodules are present. Coalified rootlets occur locally. Desiccation cracks are very common. Facies 9 consists of 9–84 cm thick sheet sandstone units that differ from Facies 7 only in the abundance of large polygonal desiccation cracks and the presence of carbonate nodules and cement. One unit preserves abundant autochthonous P. eggertiana rootstocks in growth position (Fig. 6a and d). These rootstocks occur as compressions on the palaeosol surface and are overlain, but not cast, by pyrite-rich mudstone of Facies 1.

M I S S I S S I P P I A N LY C O P S I D F O R E S T S

973

Fig. 6. Fossil plants in growth position, Albert Formation, near Norton. (a) Coalified compressed Protostigmaria tree base, palaeosol 4.192; scale bar represents 35 mm. (b) Protostigmaria showing mould preservation, palaeosol 2.112; scale bar represents 20 mm. (c) Two closely spaced, upright specimens of Lepidodendropsis, palaeosol 2.112; scale bar represents 50 mm. (d) Compressed Protostigmaria exposed immediately below tree base, palaeosol 4.192; scale bar represents 60 mm. (e) Leaves (arrow) still attached to upright trunk (left), palaeosol 2.105; scale bar represents 10 mm. (f) Poorly preserved tree mould showing radiating roots, palaeosol 4.139; scale bar represents 100 mm. (g) Partially decorticated lycopsid surface preserved on tree mould, palaeosol 4.139; scale bar represents 5 mm. (h) Upright lycopsid trunks (arrows) buried in splay deposit, palaeosol 2.112; scale bar represents 50 mm.

Prone Lepidodendropsis stems and common charcoal fragments occur in the palaeosol.

Palaeoenvironmental interpretation Coarsening-upward successions of Facies Association A are interpreted as prograding delta-front deposits. Originally thought to be freshwater lacustrine deltas (Carter & Pickerill 1985a), the ichnofauna, together with body fossil assemblages of foraminifers, ostracods, and fish indicate that the deltas were marginal to moderately deep brackish bays (Tibert & Scott 1999). Organic-

rich mudstone units are interpreted as prodelta deposits (Facies 1). Heterolithic units containing minor soft-sediment deformation are interpreted as delta-slope sediments deposited below wavebase (Facies 2). Symmetrically rippled sandstone sheets are interpreted as shallow to emergent, wave-dominated shoreface and beach deposits; putative gymnosperm trees colonized beach ridges (Facies 3). Erosive-based, broadly channelized units containing symmetrical ripples and trough cross-beds are interpreted as wave-reworked distributary channel mouth bar deposits (Facies 4). The abundance of symmetrically rippled sheets and absence of well-developed mouth bar deposits strongly suggest

974

H . J. FA L C O N - L A N G

that the delta fronts developed in a wave-dominated setting (Elliot 1986; Martel & Gibling 1991). Facies Association B is interpreted as formed in poorly drained delta-plain settings that were densely vegetated by lycopsid trees. Pebbly sandstone channel bodies that contain lowangle stratification are interpreted as the deposits of sinuous, fast-flowing, distributary channels, up to 2.5 m deep, inclined strata representing the deposits of laterally accreting point bars (Facies 5). Grey, rooted mudstone beds containing abundant plant debris are interpreted as the suspension deposits of vegetated interdistributary wetlands (Facies 6). Lenticular sandstone units containing upright trees are interpreted as crevasse splays that periodically buried interdistributary forests in sedimentary sheets, up to 81 cm thick (Facies 7). Delta-plains were probably fluvially dominated, protected from wave-dominated delta-front conditions by beach ridges (Elliot 1986). The very coarse-grained nature of the sediment indicates that the hinterland catchment was relatively small and steep (Orton & Reading 1993), consistent with the known tectonic setting (Calder 1998). Facies Association C is interpreted as well-drained alluvialplain deposits. In addition to sinuous channel deposits (Facies 5), red desiccated mudstone beds are interpreted as well-drained vegetated floodbasins (Facies 8) that formed under a seasonally dry climate as indicated by pedogenic carbonate nodules (Goudie 1983). Climatic seasonality is supported by studies of palynofacies (Van der Zwan et al. 1981), palaeosols (Wright 1990) and tree-rings in fossil woods (Falcon-Lang 1999a) at other sites in the SE Laurasia region. Climatic seasonality also probably operated during the deposition of Facies Associations A and B, its effects being masked by high water tables during the deposition of these successions. Finally, desiccated sheet sandstones are interpreted as crevasse splays that periodically inundated the well-drained floodbasins (Facies 9). The three facies associations are organized into cycles, 7– 15 m thick, that represent the gradual infilling of brackish embayments by prograding deltas (Facies Associations A and B) followed by floodplain aggradation above base-level resulting in well-drained alluvial plains (Facies Association C). These progradational–aggradational sequences are punctuated by flooding events, transgressions of varying magnitudes that mark a rapid return to prodelta–delta-front conditions. Cyclicity was probably driven by tectonism rather than glacial eustasy, as indicated by evidence for rapid base-level rises (flooding surfaces) without corresponding evidence for marked base-level fall such as valley incision or mature soil development (see Davies & Gibling 2003). This interpretation is supported by the large-scale sandstone dykes and soft-sediment deformation, which demonstrate the occurrence of syndepositional seismic activity (Armitage et al. 2002). An alternative explanation for the cyclicity, autocyclic delta lobe switching (Demko & Gastaldo 1996), fails to account for the largest transgressive events (sudden transitions from well-drained alluvial plains to below wave-base brackish bays) or for sandstone dyke emplacement. Cyclicity in the Horton Bluff Formation of Nova Scotia, which was deposited coevally with the Albert Formation in a very similar geological context, has also been attributed to tectonism (Martel & Gibling 1991).

Fossil forests Abundant trees in growth position (n ¼ 676) were observed on multiple palaeosol horizons (n ¼ 13) in all three facies associations. Each palaeosol was allocated an unique identification number recording the site and its level in metres (e.g. palaeosol

4.192 is located at 19.2 m in the measured section of Site 4; Fig. 4). All tree-bearing palaeosols were very poorly developed, composed of thin, medium grey mudstone–siltstone layers that showed either no alteration (apart from being pervasively rooted) or at most a weak horizonation defined by a light grey upper layer (interpreted as a leached epipedon). These palaeosols would be classified as entisols and inceptisols using modern soil taxonomy nomenclature (USDA Soil Survey Staff 1999), soils typical of very immature fluvial exposure surfaces (Retallack 1990). Nearly pure stands of lycopsid trees occur on 11 palaeosol surfaces (1.071, 2.068, 2.088, 2.105, 2.112, 3.143, 3.153, 4.027, 4.130, 4.139, 4.192). These mostly occur in interdistributary delta-plain settings during progradational phases (Facies Association B), and are buried by crevasse splay sandstone sheets. However, one example (4.192) was located on top of red alluvial-plain strata immediately beneath a major flooding surface (boundary of Facies Associations A and C; Fig. 4), and contained a single putative gymnosperm stump in addition to abundant lycopsids. Although calcrete nodules occur immediately below this fossil forest, the trees are rooted in a thin hydromorphic entisol superimposed on top of the underlying well-drained soils. These fossil trees probably represent the remains of wetland forests that established in response to rising water tables during the early part of a transgressive phase. Two final forest horizons (4.028, 4.032) containing putative gymnosperms occur in wave-dominated and possibly brackish-water shoreline settings, which formed during transgressive phases (boundary of Facies Associations A and B; Fig. 4).

Identity of fossil trees More than 99% of the trees observed in growth position are lycopsids (n ¼ 672). These mostly consist of either rootstock compressions or mould-casts, or sandstone-cast upright trunks that lack a preserved rootstock. Only about 14 in situ lycopsid specimens at Norton showing both rootstock and trunk in organic connection are exposed. The rootstocks are exposed in natural horizontal cross-sections that cut through the specimens at variable levels. Many compressed specimens (n ¼ 87) preserve a complete coalified tree base, which is circular to slightly oval in plan view (Figs. 6a and 7a; 8–156 mm diameter). Subtle marginal lobes are visible in the largest and best-preserved specimens, with 6–14 lobes being typical where measurable (mode: 10 lobes; n ¼ 11); fewer lobes characterize smaller specimens. Internal mould preservation of a few (n ¼ 3) specimens shows that shallow furrows separate lobes (Figs. 6b and 7b). Abundant, 5–8 mm diameter roots, oval to crescent-shaped in cross-section, radiate horizontally from the bases of all specimens (Figs. 6d and 7); roots are typically unbranched although a few bifurcations were observed. The radial extent of the roots appears closely related to tree base diameter, with small (1–2 cm diameter) tree bases having roots that extend for 5–10 cm whereas the largest tree bases (up to 16 cm diameter) have roots radiating for 100–130 cm (Fig. 8). In compressed specimens horizontally sectioned at a level a few millimetres to centimetres below the coalified tree base (n ¼ 346), it is evident that roots initially depart vertically to subvertically from the rootstock base before turning to a subhorizontal orientation after a few centimetres (Fig. 6d); however, some roots maintain a vertical trajectory reaching a depth of 5–10 cm below the rootstock base. All these rootstock specimens are referable to Protostigmaria eggertiana Jennings, the only diagnostic character of this species

M I S S I S S I P P I A N LY C O P S I D F O R E S T S

A lobate coalified tree base

roots

root scars

5 cm fractured surface

B root scars

roots

depressions

2 cm Fig. 7. Field sketches of Protostigmaria rootstocks. (a) Coalified tree base, palaeosol 4.192. (b) Mould preservation of tree base, palaeosol 2.112.

34

33

32

C

975

not preserved at Norton being the precise arrangement of lateral roots on the rootstock (Jennings 1975; Jennings et al. 1983). Upright lycopsid trunks (n ¼ 210), which may or may not be attached to identifiable protostigmarian rootstocks, show preserved heights of up to 81 cm (Fig. 6c and h). Trunks exhibit a pronounced basal flare in the lower 30 cm of the trunk. Diameter ranges from 11 to 97 mm, above the flared zone, and at the point of maximum flare diameter is c. 40–50% greater. Only 29 trunk specimens are exposed longitudinally, and of these 34% are decorticated and 66% preserve leaf scars in various states of preservation (Fig. 6c). The best-preserved specimens exhibit leaf scars with a rounded outline that are borne in helical rows, locally possess leaf cushions and lack parichnos tissue scars. These fossils are attributable to Lepidodendropsis corrugata (Dawson) Bell. Very young specimens with whole-plant heights of up to 32 cm retain attached leaves with fine parallel venation (2–3 mm diameter; up to 31 mm long); leaves depart at angles of 40–458 from the vertical trunk (Fig. 6e). More mature stems are leafless. Numerous basal trunk moulds (n ¼ 26) occur at a few additional horizons that are generally larger than the material discussed so far (5–22 cm diameter). Although in most cases these moulds preserve no diagnostic characters (Fig. 6f), two examples exhibit poorly preserved, partially decorticated lycopsid leaf scars (genus indeterminate; Fig. 6g). These trunk moulds probably represent the largest and most mature specimens of the Protostigmaria–Lepidodendropsis-type trees preserved elsewhere, although it is possible that they are the remains of different lycopsids. Only four of the 676 in situ tree specimens observed cannot be attributed to lycopsids. These non-lycopsids consist of coalified stumps, up to 13 cm in diameter, that exhibit a complex rooting system. Up to 5–7 primary roots, 2–3 cm in diameter occur and exhibit greater than 3–4 orders of branching (Fig. 9). Roots exhibit a longitudinally oriented, coarsely fibrous texture. These stumps cannot be identified precisely, but their gross morphology is suggestive of an indeterminate growth pattern inconsistent with lycopsid and sphenopsid biology (Stewart & Rothwell 1993). The only Mississippian group commonly characterized by indeterminate growth was the gymnosperms. Very few in situ gymnospermous stumps have been described previously in Mississippian tropical strata, all examples being Pitus stumps from Scotland (Long 1979; Galtier et al. 1994). In the absence of further diagnostic features, and given the paucity of comparative material, the Norton tree stumps are classified as putative gymnosperms in the remainder of this paper.

Tree diameter, density and basal area B

25 cm A Fig. 8. Field sketch showing size and distribution of Protostigmaria specimens on a typical 4 m2 area of palaeosol 4.192.

The diameter and density of fossil trees were measured on the five most extensive palaeosol surfaces (2.105, 2.112, 3.153, 4.139, 4.192; Table 1). Diameter measurements taken from palaeosol 4.192 (where only tree bases were preserved) were corrected by a 40% reduction to account for the pronounced basal flare. This allowed the data to be directly compared with those for the other sites. Tree mapping (Fig. 10) was facilitated by drawing a metre-scale grid over each surface with chalk. Tree diameter ranged from 0.6 to 22.0 cm for all sites with the mean diameter for each palaeosol ranging from 2.8 to 12.4 cm (n ¼ 213 measurable trees). For each palaeosol, tree diameter populations were negatively skewed, possessing a higher proportion of small diameter trunks relative to large diameter trunks. Mean tree density varied from 1.048 to 2.906 trees per m2 (range 0–11 trees per m2 ) for the five palaeosols (Table 1; Fig. 10).

976

H . J. FA L C O N - L A N G

The basal trunk area for the Norton forests, calculated from the tree diameter and tree density data, ranges from 16.3 to 126.7 m2 ha
1 for the five most extensive palaeosols (Table 1). Basal trunk area, which is a proxy for forest biomass (Cao et al. 2000), shows a significant inverse relationship with tree density (R2 ¼ 0:891; Fig. 11b) and a significant positive relationship with tree diameter (R2 ¼ 0:988; Fig. 11c). Forest biomass sharply increased with decreasing tree density and increasing tree diameter.

A

5 cm

Tree spatial distribution

B

10 cm

Fig. 9. Field sketches of putative gymnospermous stumps and roots: (a) palaeosol 4.028; (b) palaeosol 4.032.

The data indicate that a significant inverse linear relationship exists between tree diameter and tree density when plotted on a log10 scale (R2 ¼ 0:950). Although the data are limited to only five surfaces, palaeosols bearing larger trees generally appear to have lower tree densities (e.g. 3.153) compared with more densely forested palaeosols (e.g. 4.192) that contain relatively smaller trees (Fig. 11a). Significantly, palaeosols that bear the larger more widely spaced trees are generally slightly more mature (inceptisols) than those that bear smaller and more densely distributed trees (entisols).

The heterogeneity of tree distribution was also evaluated for the two most extensive palaeosols (2.105 and 4.192) using the Hayek & Buzas (1997) method as modified by DiMichele et al. (1996). In this method, the means and variances of tree populations in adjacent quadrats of varying sizes (1 m 3 1 m, 1 m 3 2 m, 1 m 3 3 m, 1 m 3 4 m) were calculated (Table 2). Only a subset of each palaeosol dataset could be analysed because in certain areas the quadrat size exceeded the outcrop size. Hence, data in Table 2 differ slightly from those in Table 1, which are based on calculations encompassing the entire palaeosol area. Log-transformed means and variances of tree occurrences for each of the quadrat sizes were plotted for each palaeosol, with the means on the abscissa and the variances on the ordinate (Fig. 12). In this method, if trees were randomly distributed, the mean and variance of tree occurrence would be equal at all quadrat sizes, resulting in a line with a slope of unity. In contrast, trees showing an ordered or clumped distribution would give plots with either lower or higher slopes, respectively (DiMichele et al. 1996). The slope for palaeosol 2.105 data is 1.69 and that for palaeosol 4.192 is 1.42. These data indicate that trees have a clumped distribution on two palaeosols with high tree density and low tree diameter. Qualitatively these palaeosols are characterized by 4–7 m diameter clumps composed of 20–35 large diameter trees (10–16 cm) separated by tracts, up to several metres wide, dominated by smaller diameter individuals (,5 cm) (Fig. 10). Statistical analysis of tree spatial distribution cannot be performed on palaeosols with lower tree density and high tree diameter (3.153 and 4.139) because far fewer data are available. However, qualitative observations indicate that trees on these latter palaeosols have a more random distribution compared with palaeosols 2.105, 2.112 and 4.192.

Charred palynodebris in palaeosols Mudstone fragments from each of the five well-exposed palaeosols were processed for .200 ìm diameter palynodebris using standard techniques (Pearson & Scott 1999). Charcoal fragments,

Table 1. Descriptive data for the five most extensive fossil forest horizons Fossil forest Site Site Site Site Site

2.105 2.112 3.153 4.139 4.192

No. of trees

Bedding surface area (m2 )

Tree density per m2 (range)

Tree density per hectare

Mean tree diameter (cm) (range)

127 81 11 8 433

72 35 10.5 7 149

1.847 (0–8) 1.943 (0–6) 1.048 (0–3) 1.142 (0–3) 2.906 (0–11)

18472 19428 10476 11428 29060

4.30 (1.4–6.2) 3.77 (0.9–9.7) 12.41 (5.0–22.0) 8.21 (5.0–10.0) 4.44 (0.9–15.6)*

Tree heights calculated from stem diameter using the Niklas (1994) relationship. *Tree height and basal area data for palaeosol 4.192 based on a tree diameter data reduced by 40% (see text for explanation).

Maximum tree height (m) (range) 5.26 (1.77–7.19) 4.68 (1.09–10.23) 12.27 (6–18.02) 8.99 (6–10.47) 3.68 (1.09–14.39)*

Tree basal area (m2 ha
1 ) 26.82 21.69 126.73 60.51 16.39*

M I S S I S S I P P I A N LY C O P S I D F O R E S T S 8

9

10

6

7

5

3

4

2

977

0

1

F 50 cm

Fossil forest exposed for a further 56 m

Edge of bedding surface

E

D

C

B

paleosol 4.192

A

A

B

log10 basal trunk area

log10 tree diameter

1

0.8

R

0.6

2

=

0.

95

03

0.4

C 2.2

2.2

2

2

log10 basal trunk area

1.2

1.8

1.6

R 1.4

2

=

0. 89 11

1.2

0.2 3.9

4

4.1

Fig. 10. Fossil forest map for a small proportion of palaeosol 4.192 showing clumped distribution of Protostigmaria specimens.

4.2

4.3

4.4

4.5

1.8

76

98

2

1.6

R

=

0.

1.4

1.2

1 3.9

1

4

4.1

4.2

4.3

4.4

4.5

0.2

log10 tree density

log10 tree density

0.4

0.6

0.8

1

1.2

log10 tree diameter

Fig. 11. Fossil forest dynamics. (a) Relationship between log mean tree diameter and log tree density; (b) relationship between log basal trunk area and log tree density; (c) relationship between log basal trunk area and log tree diameter (see Table 1).

Table 2. Summary statistics for repeated quadrat analysis of mean tree number per quadrat in a subset of the palaeosol 2.105 and 4.192 data Quadrat size (m 3 m)

Number of trees analysed

Palaeosol 2.105 131 132 133 134 Palaeosol 4.192 131 132 133 134

Stumps per quadrat Mean

Variance

Standard deviation

348 348 348 348

3.16 6.36 9.54 12.73

3.11 8.18 15.37 23.53

1.76 2.86 3.92 4.85

111 111 111 111

2.37 4.65 6.66 9.53

4.95 14.39 30.82 50.10

2.22 3.79 5.55 7.07

When examined using scanning electron microscopy, almost all of the charred material consists of secondary tissue composed of unornamented, elongate cells with exceptionally thick radial walls, arranged in irregular rows. Some of this material occurs as partially flattened cylinders, up to 14 mm in diameter. Previous studies of the Albert Formation in this region have illustrated similar charcoal (e.g. Carter & Pickerill 1985a, fig. 6b). In terms of gross morphology, fragments resemble the periderm of some putative permineralized Lepidodendropsis specimens (Iurina & Lemoigne 1975) and, more generally, charred Pennsylvanian lycopsid periderm fragments (Falcon-Lang 1999b, 2000). Although the Norton charcoal cannot be identified to generic level on anatomical details alone, its facies association with a nearly pure stand of Protostigmaria–Lepidodendropsis trees suggests it was derived from this plant.

Further explanation is given in the text and by DiMichele et al. (1996).

Palaeoecological synthesis

up to 14 mm in diameter, formed the only identifiable component of the resultant palynodebris. Charcoal was particularly common in palaeosol 4.192, which lies on the boundary of Facies Associations A and C (Fig. 4).

Very dense lycopsid forests dominated a variety of interdistributary flood-basin settings at Norton during periods of delta-plain progradation and, more rarely, during transgressive phases. The generally immature nature of the palaeosols indicates that most vegetation colonized newly deposited fluvial sediments.

978

H . J. FA L C O N - L A N G 2

54

9

1x4 0 .0

1.5

-0 .2 13 1

1.6 y=

1x4

=

1x3

ol 4. 19 2, y

1x2

Pa la es

1x1

1. 42 67 x

5, .10 l2

1x2

so eo

1

Pa la

log10 variance

94

4x

+

1x3

0.5

1x1

log–log scale, the ‘self-thinning’ threshold has a slope ranging from
1.5 to
1.7 for almost all modern stands (Cao et al. 2000). Maturing stands may initially have a steeper slope below the self-thinning threshold, but turn parallel to this line as they approach it (Fig. 13). Significantly, the Norton forests are characterized by a relatively steep slope on the log10 diameter–log10 density plot. This, together with negatively skewed tree diameter data, provides further confirmation that these were all immature stands that had not yet reached their self-thinning threshold (Silvertown & Doust 1993). Even more intriguingly, the most mature Norton forests had mean tree diameters that exceed the self-thinning threshold established from modern forest observations for a given forest density (Fig. 13). The development of uniquely dense stands for a given tree diameter may have resulted from the unusually sparse microphyllous canopy possessed by these lycopsids (Jennings et al. 1983; Gensel & Pigg 2002). It is likely that all the Norton forests would have had open canopies with plenty of light gaps regardless of their relative maturity and extraordinary density (see DiMichele & Phillips 1994). Therefore, light competition would not have been a significant factor in determining the selfthinning threshold for these structurally unique forests. The inferred reduction in tree density with increasing forest maturity (Fig. 11a) probably resulted from competition for limited soil nutrients below the self-thinning line (Silvertown & Doust 1993).

0 0

0.2

0.4

0.6

0.8

1

1.2

log10 mean Fig. 12. Plot of the mean v. variance (log-transformed) for the number of trees per quadrat at four quadrat sizes for palaeosols 2.105 and 4.192. A slope of unity indicates random tree distribution whereas lower and higher slopes indicate ordered and clumped distributions, respectively (see Table 2).

Forest biomass, canopy height and structure Relationships between tree density, diameter and basal area indicate that forest maturation (as measured by increasing forest biomass) proceeded through the development of progressively more widely spaced, but larger, trees (Fig. 11a and b). Forest basal area increased from 16 to 127 m2 ha
1 as forests matured. 2

Self-thinning

gth

1.4

old

sh

re

1.2 1

n Nor to

0.8

t fores

log10 tree diameter

1.6

nin

0.6

s

The inverse relationship between tree diameter and spacing indicates that as vegetation matured so forest density decreased. This interpretation is further supported by the slight increase in palaeosol maturity with decreasing fossil forest density. Based on this assumption, palaeosol 4.192, which contains the densest fossil forest composed of the smallest diameter trees, probably preserves a snapshot of a very early post-colonization stage. Scaled up to standard forestry units, tree distribution data indicate forest densities of up to 29 060 trees per hectare. Palaeosol 3.153, which contains the largest diameter and most widely spaced trees, probably represents the most mature lycopsid forests preserved at Norton. Scaled up to standard forestry units, this surface would have supported up to 10 476 trees per hectare (Table 1). The inverse relationship between tree diameter and tree density observed at Norton is a universal characteristic of present-day, even-aged, developing monocultures, and occurs as a result of competition for limited resources (light, nutrients, water) between adjacent plants (Silvertown & Doust 1993). Empirical data and model results have shown that there is an upper limit to tree diameter for a given density. As trees increase in diameter and this upper threshold is approached, tree density is reduced by competition-induced mortality. When plotted on a

hin

lf-t

se

1.8

0.4

A

B

C

0.2 0 0

1

2

3

4

log10 tree per hectare Fig. 13. Log-transformed relationship between tree density and tree diameter for Norton forests (see Fig. 11a), showing the self-thinning threshold. Stands of different densities may initially have steeper trajectories, but turn toward the self-thinning threshold as they approach it (e.g. medium-density stand A: 1000 trees per hectare; high-density stand B: 4000 trees per hectare; extremely high-density stand C: 10 000 trees per hectare; after Cao et al. 2000).

5

M I S S I S S I P P I A N LY C O P S I D F O R E S T S

Such values are substantially lower than those calculated for Pennsylvanian lycopsid forests (up to 992 m2 ha
1 ), indicating that Tournaisian forests did not rival the forest productivity attained at the later zenith of lycopsid evolution (DiMichele et al. 1996). In comparison with Pennsylvanian lycopsid forests, most of the Norton forests evidently possessed rather low canopies. The height of individual lycopsid trees can be estimated using the empirical biomechanical relationship that exists between trunk diameter and tree height (Niklas 1994). Although arborescent lycopsids have no precise modern structural analogues, their dense periderm-supported trunks are probably represented most accurately by the ‘woody trunks’ category of Niklas (1994), which is described by the equation log10 H ¼ 1:59 þ 0:39(log10 D)
0:18(log10 D)2 where H is the maximum buckling height, and D is the trunk diameter above the zone of basal flare. These biomechanical calculations suggest that mean tree heights varied from 5 to 12 m for different palaeosols. The youngest forested palaeosols were characterized by heterogeneous clumps of more mature trees, up to 8–14 m high, surrounded by tracts of more juvenile trees, up to 1–5 m high (mean 5 m). More mature forested palaeosols contained unimodal tree populations, which formed a canopy 6– 18 m high (mean 12 m). Pennsylvanian lycopsid canopies were, in comparison, up to 30 m high (Niklas 1994). The clumpy nature of the most juvenile forested areas at Norton can probably be explained by one of two competing hypotheses. One possibility is that the tracts of very small trees represent secondary regeneration following localized stand disturbance caused by, for example, fires, floods, storms or insect attack (Silvertown & Doust 1993). Although there is direct geological evidence for the occurrence of the former two disturbance processes at Norton (see below), the absence of fallen tree remains in the canopy gaps and the very immature nature of these forest stands makes this hypothesis unlikely. Alternatively, and more probably, heterogeneous tree distribution patterns may have arisen through the progressive colonization of the palaeosol. Experiments suggest that arborescent lycopsid propagules had a rather limited dispersal range (Habgood et al. 1998). Therefore, colonization may have proceeded from a series of isolated centres, resulting in the formation of mature clumps that progressively colonized areas further afield (Silvertown & Doust 1993).

Vegetation disturbance The predominantly immature nature of the lycopsid forests at Norton probably resulted from regular ecological disturbances. These periodically killed whole stands, created widespread unvegetated substrates available for colonization, and prevented the establishment of climax vegetation. There is evidence for three major disturbance processes at Norton. The most important cause of disturbance during delta progradational phases was flooding. All the upright trees within Facies Association B are sandstone-cast within thin crevasse splay sandstones (up to 81 cm thick). Although many floodplain forests today are adapted to regular flooding (Kozlowski 1984), partial burial appears to have been lethal to the lycopsid forests at Norton. This may have been due to the small size of the trees, which were apparently partially uprooted by floods (as indicated by the pronounced tilt of some trunks), and the inability of their determinate rooting system to adventitiously seek shallower aerated substrates following burial.

979

Flooding was also the dominant disturbance process affecting Pennsylvanian lycopsid delta-plain forests (Demko & Gastaldo 1992; Calder et al. 1996). Sandstone-cast lycopsid trees of this age showing similar downstream tilting and distortion to the Norton specimens have been described (Gastaldo 1986b), with some examples additionally preserving complex scour and mound features around the trees (Rygel et al. 2004). A secondary disturbance process was fire, as indicated by locally abundant lycopsid charcoal. Fires appear to have been particularly important in the forest community on palaeosol 4.192, which developed on top of red beds during a transgressive event. Fire frequency is greater on drier substrates (Falcon-Lang 2000), and abundant charcoal may suggest that this lycopsid forest established close to its edaphic limit. A final disturbance process was drowning caused by rapid, fault-induced basin subsidence events. The forest on palaeosol 4.192 appears to have been killed by rapid drowning by deep (below wave-base) brackish bays and subsequently was slowly buried (Fig. 4). This may explain why only rootstocks are preserved on this surface, the aerial portions of the trees having had ample time to rot away on the bottom of the bay prior to significant sedimentation. A slower rate of drowning is recorded by palaeosols 4.028 and 4.032, where vegetation repeatedly established in retrograding wave-dominated beach ridge facies (Fig. 4). Significantly, these are the only horizons where putative gymnosperms rather than lycopsids dominated. It is possible that these brackish bay margin communities may have utilized a mangrove-type strategy as postulated for some Pennsylvanian gymnosperms (Raymond & Phillips 1983).

Tournaisian tropical vegetation biomes Compared with Pennsylvanian times (DiMichele & Phillips 1994), early Mississippian tropical wetland ecosystems are poorly understood. Most of the tropical zone during this period was subjected to varying degrees of rainfall seasonality caused by the Gondwanan monsoonal effect (Wright 1990; Falcon-Lang 1999a). Within this summer-wet biome, well-drained soils were particularly widespread with wetland niches restricted to lake margin and fluvio-deltaic areas (Falcon-Lang 2000). Consequently, most of our knowledge of Tournaisian tropical vegetation is centred on the drylands, which were dominated by diverse, fire-prone gymnosperm–fern woodlands that locally contained very large Pitus trees (Scott & Rex 1987). In contrast, remains of wetland communities have received far less attention. Allochthonous assemblages dominated by lycopsid trees of Lepidodendron calampsoides Long have been recorded from Tournaisian fluvio-deltaic facies across the British Isles (Scott & Rex 1987), and coeval fluvio-deltaic facies in eastern North America have yielded various species of Lepidodendropsis (Dawson 1868; Bell 1960; Jennings et al. 1983). However, lycopsid wetland forests in growth position have previously been described only twice, and never with palaeoecological detail. A few tens of upright Lepidodendropsistrees have been described rooted in peat mire deposits in lower delta-plain associations in the Price Formation of Virginia (Jennings 1975; Jennings et al. 1983; Scheckler 1986a, b). Attempts to reconstruct this species as a whole-plant are currently under way, based on detailed new collections (Gensel & Pigg 2002). The only other documented Tournaisian site containing lycopsid trees in growth position is located in the type section of the Horton Bluffs Formation, Nova Scotia. Here a small number of Lepidodendropsis trees occur rooted in wave-dominated fluviodeltaic facies (Martel & Gibling 1991). Although details of trunk size

980

H . J. FA L C O N - L A N G

were not given, lycopsid forests were of similar density (locally up to 4.5 trees per m2 ) to those of the Albert Formation (Bell 1960, p. 11). Consequently, the numerous Protostigmaria–Lepidodendropsis forests described in this paper represent a significant addition to our knowledge of Mississippian wetland ecosystems. Such studies are not only important for our understanding of the nature of terrestrial communities at this time, but are also essential if we are to fully appreciate the evolution of the Pennsylvanian ‘Coal Forests’. Limited data indicate that Lepidodendron sensu latu replaced Lepdidendropsis as the main forest tree in tropical wetlands by late Mississippian times (Beeler 1986). The Norton forests therefore represent ecological and evolutionary precursors of the Pennsylvanian ‘Coal Forests’.

Conclusions (1) Early Mississippian (Tournaisian) delta-plain deposits in the Albert Formation of New Brunswick, Canada, contain the remains of extremely dense lycopsid forests (up to 30 000 trees per hectare). (2) A linear relationship between tree diameter and density indicates that as forests matured competition for nutrients resulted in self-thinning. However, forests never achieved climax because of regular flood-disturbance. (3) Compared with modern forests, Tournaisian lycopsid forests were extremely dense for a given trunk diameter, a phenomenon related to their microphyllous canopy that minimized light competition. (4) The Albert Formation fossil forests are important because they shed light on the previously enigmatic nature of early Mississippian lycopsid communities, the precursors of the Pennsylvanian ‘Coal Forests’. This study was funded through a Matthew Fellowship (New Brunswick Museum) and an NERC Fellowship (NER/I/S/2001/0738) held at the University of Bristol (UK). I thank P. Gensel, R. Miller, M. Rygel and M. Gibling for encouragement, sharing unpublished data and critically reading early versions of this manuscript. The reviews of Bill DiMichele and Bob Gastaldo greatly improved this paper.

References Armitage, I., Gingras, M. & Keighley, D. 2002. Lacustrine Sedimentology, Ichnology, and Sequence Stratigraphy of Early Carboniferous Rift Basins: Albert Formation, Sussex, New Brunswick. Atlantic Universities Geoscience Conference, Field Trip 4. Beeler, H.E. 1986. Palaeoecology of Upper Mississippian coal swamps. American Journal of Botany, 70, 695. Bell, W.A. 1960. Mississippian Horton Group of Type Windsor–Horton District, Nova Scotia. Geological Society of Canada Memoir, 314. Calder, J.H. 1998. The Carboniferous evolution of Nova Scotia. In: Blundell, D.W.J. & Scott, A.C. (eds) Lyell: The Past is the Key to the Present. Geological Society, London, Special Publications, 143, 261–302. Calder, J.H., Gibling, M.R., Eble, C.F., Scott, A.C. & MacNeil, D.J. 1996. The Westphalian D fossil forest at Table Head, Sydney Basin, Nova Scotia: sedimentology, paleoecology and floral response to changing edaphic conditions. International Journal of Coal Geology, 31, 277–313. Cao, Q.V., Dean, T.J. & Baldwin, C. 2000. Modelling the size–density relationship in direct-seeded slash pine stands. Forest Science, 46, 317–321. Carter, D.C. & Pickerill, R.K. 1985a. Algal swamp, marginal and shallow evaporitic lacustrine lithofacies from the late Devonian–early Carboniferous Albert Formation, southeastern New Brunswick, Canada. Maritime Sediments and Atlantic Geology, 21, 69–86. Carter, D.C. & Pickerill, R.K. 1985b. Lithostratigraphy of the Late Devonian– Early Carboniferous Horton Group of the Moncton subbasin, southern New Brunswick. Maritime Sediments and Atlantic Geology, 21, 11–24. Davies, S.J. & Gibling, M.R. 2003. Architecture of coastal and alluvial deposits in an extensional basin: the Carboniferous Joggins Formation of Eastern Canada.

Sedimentology, 50, 415–439. Dawson, J.W. 1868. Acadian Geology. Macmillan, London. Demko, T.M. & Gastaldo, R.A. 1992. Paludal environments of the Mary Lee Coal zone, Pottsville Formation, Alabama—stacked clastic swamps and peat mires. International Journal of Coal Geology, 20, 23–47. Demko, T.M. & Gastaldo, R.A. 1996. Eustatic and autocyclic influences on deposition of the lower Pennsylvanian Mary Lee coal zone, Warrior Basin, Alabama. International Journal of Coal Geology, 31, 3–19. DiMichele, W.A. & Phillips, T.L. 1994. Palaeobotanical and palaeoecological constraints on models of peat formation in the Late Carboniferous of Euramerica. Palaeogeography, Palaeoclimatology, Palaeoecology, 106, 39–90. DiMichele, W.A., Eble, C.F. & Chaney, D.S. 1996. A drowned lycopsid forest above the Mahoning coal (Conemaugh Group, Upper Pennsylvanian) in eastern Ohio, USA. International Journal of Coal Geology, 31, 249–276. Elliot, T. 1986. Deltas. In: Reading, H.G. (ed.) Sedimentary Environments and Facies. Blackwell Scientific, Oxford, 113–154. Falcon-Lang, H.J. 1999a. The Early Carboniferous (Courceyan–Arundian) monsoonal climate of the British Isles: evidence from growth rings in fossil woods. Geological Magazine, 136, 177–187. Falcon-Lang, H.J. 1999b. Fire ecology of a Late Carboniferous floodplain, Joggins, Nova Scotia. Journal of the Geological Society, London, 156, 137–148. Falcon-Lang, H.J. 2000. Fire ecology of the Carboniferous tropical zone. Palaeogeography, Palaeoclimatology, Palaeoecology, 164, 339–355. Ferguson, L. & Fyffe, L.R. 1985. Geological Highway Map of New Brunswick and Prince Edward Island. Keystone Press, Saint John. Galtier, J., Brown, R.E., Scott, A.C., Rex, G.M. & Rowe, N.P. 1994. A Late Dinantian flora from Weaklaw, East Lothian, Scotland. Special Papers in Palaeontology, 49, 57–74. Gastaldo, R.A. 1986a. Implications on the palaeoecology of autochthonous lycopsids in clastic sedimentary environments of the Early Pennsylvanian of Alabama. Palaeogeography, Palaeoclimatology, Palaeoecology, 53, 191–212. Gastaldo, R.A. 1986b. An explanation for lycopod configuration, ‘Fossil Grove’ Victoria Park, Glagow. Scottish Journal of Geology, 22, 77–83. Gensel, P.G. & Pigg, K.B. 2002. Reconstruction of the Lepidodendropsis– Protostigmaria plant from the Mississippian Price Formation of Virginia, USA. Botany 2002, Botany in the Curriculum: Integrating Research and Teaching. University of Wisconsin, Madison, 27. Goudie, A.S. 1983. Calcrete. In: Goudie, A.S. & Pye, K. (eds) Chemical Sediments and Geomorphology. Academic Press, London, 93–131. Greiner, H.R. 1962. Facies and sedimentary environment of the Albert Shale, New Brunswick. AAPG Bulletin, 46, 219–234. Gussow, W.C. 1953. Carboniferous stratigraphy and structural geology of New Brunswick, Canada. AAPG Bulletin, 37, 1713–1816. Habgood, K.S., Hemsley, A.R. & Thomas, B.A. 1998. Modelling of the dispersal of Lepidocarpon based on experiments using reconstructions. Review of Palaeobotany and Palynology, 102, 101–114. Hayek, L.A. & Buzas, M.A. 1997. Surveying Natural Populations. Columbia University Press, New York. Iurina, A. & Lemoigne, Y. 1975. Anatomical characters of the axes of arborescent lepidophytes of the Siberian Devonian, referred to as Lepidodendropsis kazachstanica Senkevitsch1961. Palaeontographica B, 150, 162–168. Jennings, J.R. 1975. Protostigmaria, a new plant organ from the Lower Mississippian of Virginia. Palaeontology, 18, 19–24. Jennings, J.R., Karrfalt, E.E. & Rothwell, G.W. 1983. Structure and affinities of Protostigmaria eggertiana. American Journal of Botany, 70, 963–974. Kozlowski, T.T. (ed.) 1984. Flooding and Plant Growth, Academic Press, London. Long, A.G. 1979. Observations on the Lower Carboniferous genus Pitus Witham. Transactions of the Royal Society of Edinburgh: Earth Sciences, 70, 111–127. Martel, A.T. & Gibling, M.R. 1991. Wave-dominated lacustrine facies and tectonically controlled cyclicity in the Lower Carboniferous Horton Bluff Formation, Nova Scotia, Canada. In: Anadon, P., Cabrera, L. & Kelts, K. (eds) Lacustrine Facies Analysis. International Association of Sedimentologists, Special Publications, 13, 223–243. Martel, A.T., McGregor, D.C. & Utting, J. 1993. Stratigraphic significance of Upper Devonian and Lower Carboniferous miospores from the type area of Horton Group, Nova Scotia. Canadian Journal of Earth Sciences, 30, 1091–1098. Miller, R.F. & McGovern, J.H. 1997. Preliminary report of fossil fish (Actinopterygii: Palaeonisciformes) from the Lower Carboniferous Albert Formation at Norton, New Brunswick (NTS 21 H/12). In: Carroll, B.M.W. (ed.) Current Research 1996. New Brunswick Department of Natural Resources, Resource Report, 97-4, 191–200. Niklas, K.J. 1994. Predicting the height of fossil plant remains: an allometric approach to an old problem. American Journal of Botany, 81, 1235–1243. Orton, G.J. & Reading, H.G. 1993. Variability of deltaic processes in terms of

M I S S I S S I P P I A N LY C O P S I D F O R E S T S sediment supply, with particular emphasis on grain size. Sedimentology, 40, 475–512. Pascucci, V., Gibling, M.R. & Williamson, M.A. 2000. Late Paleozoic to Cenozoic history of the offshore Sydney Basin, Atlantic Canada. Canadian Journal of Earth Sciences, 37, 1143–1165. Pearson, T. & Scott, A.C. 1999. Large palynomorphs and debris. In: Jones, T.P. & Rowe, N.P. (eds) Fossil Plants and Spores: Modern Techniques. Geological Society, London, 20–25. Pickerill, R.K. 1990. Non-marine Paleodictyon from the Carboniferous Albert Formation of southern New Brunswick. Atlantic Geology, 26, 157–163. Pickerill, R.K., Carter, D. & St. Peter, C. 1985. Albert Formation—Oil Shales, Lakes, Fans and Deltas. Geological Association of Canada, Excursion 6. ` Geological Association of Canada, Fredericton, N.BO Pigg, K.B. 1992. Evolution of isoetalean lycopsids. Annals of the Missouri Botanical Garden, 79, 589–612. Pigg, K.B. 2001. Isoetalean lycopsid evolution: from the Devonian to the Present. American Fern Journal, 91, 99–114. Raymond, A. & Phillips, T.L. 1983. Evidence of an Upper Carboniferous mangrove community. In: Teas, H. (ed.) Biology and Ecology of Mangroves. Tasks in Vegetation Science, 8, 19–30. Retallack, G.J. 1990. Soils of the Past: an Introduction to Paleopedology. Unwin Hyman, Winchester, MA. Rygel, M.C., Gibling, M.R. & Calder, J.H. 2004. Vegetation-induced sedimentary structures from fossil forests in the Pennsylvanian Joggins Formation, Nova Scotia. Sedimentology, 51, 531–552. Scheckler, S.E. 1986a. Floras of the Devonian–Mississippian transition. In: Broadhead, T.W. (ed.) Land Plants: Notes for a Short Course. University of Tennessee, Studies in Geology, 15, 81–96. Scheckler, S.E. 1986b. Old red continent facies in the Late Devonian and Early Carboniferous of Appalachian North America. Annales de la Socie´te´ Ge´ologique de Belgique, 109, 223–236. Scott, A.C. & Calder, J.H. 1994. Carboniferous fossil forests. Geology Today, 7, 214–216. Scott, A.C. & Rex, G.M. 1987. The accumulation and preservation of Dinantian

981

plants from Scotland and its borders. In: Miller, J., Adams, A. & Wright, V.P. (eds) European Dinantian Environments. Wiley, Chichester, 181–200. Silvertown, J.W. & Doust, J.L. 1993. Introduction to Plant Population Biology, 3rd. Blackwell Scientific, Oxford. Smith, W.D. & Gibling, M.R. 1987. Oil shale composition related to depositional setting: a case study from the Albert Formation, New Brunswick, Canada. Bulletin of Canadian Petroleum Geology, 35, 469–487. St. Peter, C. 1993. Maritimes Basin evolution: key geologic and seismic evidence from the Moncton Subbasin of New Brunswick. Atlantic Geology, 29, 233–270. St. Peter, C. 2003. Revisions to Lower Carboniferous Horton Group stratigraphy in New Brunswick, Canada. Geological Society of America, Abstracts with Programs, 35, 19. Stewart, W.N. & Rothwell, G.W. 1993. Palaeobotany and the Evolution of Plants, 2nd. Cambridge University Press, Cambridge. Tibert, N.E. & Scott, D.B. 1999. Ostracodes and agglutinated foraminifera as indicators of palaeoenvironmental change in an Early Carboniferous brackish bay, Atlantic Canada. Palaios, 14, 246–260. USDA Soil Survey Staff 1999. Soil Taxonomy: a Basic System of Soil Classification for Making and Interpreting Soil Surveys, 2nd. US Department of Agriculture, Handbook, 436. Utting, J. 1987. Palynologic investigation of the Albert Formation (Lower Carboniferous) of New Brunswick, Canada. Palynology, 11, 75–98. Van der Poll, H.W. 1995. Upper Palaeozoic rocks, New Brunswick, Prince Edward Island, and Iles de la Madeleine. In: Williams, H. (ed.) Geology of the Appalachian–Caledonian Orogen of Canada and Greenland. Geological Survey of Canada, Geology of Canada, 6, 455–492. Van der Zwan, C.J., Boulter, M.C. & Hubbard, R.N.L.B. 1981. Climatic change during the Lower Carboniferous in Euramerica, based on multivariate statistical analysis of palynological data. Palaeogeography, Palaeoclimatology, Palaeoecology, 52, 1–20. Wright, V.P. 1990. Equatorial aridity and climatic oscillations during the early Carboniferous, southern Britain. Journal of the Geological Society, London, 147, 359–363.

Received 5 November 2003; revised typescript accepted 14 April 2004. Scientific editing by Duncan Pirrie