New occurrences of Ichniotherium and Striatichnium from the Lower Permian Kildare Capes Formation, Prince Edward Island, Canada: palaeoenvironmental and biostratigraphic implications


Corresponding author.


Abstract:  Tracks and trackways of the vertebrate ichnotaxon Ichniotherium sphaerodactylum and a trace of the invertebrate ichnotaxon Striatichnium bromackerense are described for the first time in association outside of Europe. The tracks are identified as I. sphaerodactylum based on their characteristic rounded digit ends, the ovoid sole-pad of the pedal imprint and the increase in digit lengths from digits I–IV, and the invertebrate trace is identifiable as S. bromackerense based on the band-like systems of distally bifurcated striae. The tracks of I. sphaerodactylum are the largest known to date and represent rare evidence of large-bodied terrestrial vertebrates in the Kildare Capes Formation of Prince Edward Island, Canada. Morphometric differences between ichnospecies of Ichniotherium and other ichnospecies previously collected from Prince Edward Island are examined in a multivariate analysis, and results suggest that the distance between the manus and pes of the same imprint pair, the width of pace and the length of certain digits is useful for species identification. The association between I. sphaerodactylum and S. bromackerense was previously known only from the Bromacker quarry, Tambach Formation, Germany, which is interpreted as a seasonally dry, semi-arid upland environment. The co-occurrence of these traces suggests that the Eldon locality of Prince Edward Island is similar in depositional environment to the Bromacker quarry in Germany, and allows for comparison of these two localities for the first time. As diadectids are thought to be the trackmakers of Ichniotherium, the Eldon locality of the Kildare Capes Formation may, with further work, be considered as another example of the rare, herbivore-dominated palaeoenvironment, which is generally uncharacteristic of the Early Permian of North America.

T etrapod footprints are commonly used as an added measure of taxonomic diversity in terrestrial deposits, where preservation of vertebrate body fossils may not be favoured. The Permo-Carboniferous fossil record of the Atlantic provinces of Canada yields many well-preserved body fossils (Langston 1963; Reisz 1972; Calder 1998), but the taxonomic diversity increases with the consideration of the tetrapod ichnofauna (Sarjeant and Mossman 1978; Mossman and Place 1989; Mossman and Grantham 2000; Keighley and Pickerill 2003; Calder et al. 2004; Van Allen et al. 2005; Falcon-Lang et al. 2007; Falcon-Lang et al. 2010).

The Permian vertebrate ichnofauna of the Pictou Group redbeds of Prince Edward Island includes ichnotaxa previously described as ‘Gilmoreichnus kablikaeHaubold, 1973a and ‘Amphisauropus latusHaubold, 1970 (both of which should likely be reassigned to Amphisauropus kablikae (Geinitz and Deichmüller, 1882) sensuVoigt 2005), ‘Gilmoreichnus cf. G. (Hylopus) hermitanus’ (Gilmore, 1927) (considered a subjective synonym of LimnopusMarsh, 1894, VaranopusMoodie, 1929, DimetropusRomer and Price, 1940, and AmphisauropusHaubold, 1971, by S. Voigt (2005, pers. comm. 2011)), an indeterminate ichnospecies of Notalacerta Butts, 1891, and a specimen described as ‘Ichniotherium cf. I. willsiHaubold and Sarjeant, 1973. These ichnotaxa are thought to pertain to a typical Early Permian tetrapod fauna including seymouriamorphs, captorhinids, ophiacodontids, protorothyridids and diadectids (Mossman and Place 1989; Calder et al. 2004). All of these tracks were made by putatively small-bodied individuals with apparent body lengths (glenoacetabular distances) averaging 54–130 mm (Mossman and Place 1989; Calder et al. 2004). These measurements, when coupled with the fragmentary record of vertebrate body fossils from Prince Edward Island (Langston 1963), may suggest an unusual rarity of large-bodied taxa in the ecosystem, which has been noted to characterize upland assemblages (sensuEberth et al. 2000), which occur far from coastal environments or alluvial plains, such as Richards Spur, Oklahoma, USA (Sullivan et al. 2000) and the Bromacker quarry, Germany (Eberth et al. 2000). Alternatively, this pattern could be explained by a taphonomic bias against the preservation of large-bodied animals, which is an inherent problem with most ichnological studies (Van Allen et al. 2005). Better sampling is needed to address the significance of this unusual body size distribution pattern in the Permian fossil record.

The well-documented Bromacker quarry offers important insights into the complexity of preservation patterns and biases in Early Permian terrestrial sedimentary basins. The Bromacker quarry is unique as it preserves trackways of vertebrates and invertebrates, as well as vertebrate body fossils (Eberth et al. 2000; Voigt 2005; Voigt et al. 2007). The most abundant vertebrate remains recovered are of diadectids (Berman et al. 1998; Berman et al. 2004), which is discordant with the trend observed in most Early Permian North American localities, where the large-bodied synapsid carnivore, DimetrodonCope, 1878, is most abundant (Romer and Price 1940). This trend is also present in tracksites, such as the Brule locality, Nova Scotia, Canada, where all trackmakers are interpreted as carnivorous (Van Allen et al. 2005). This variation has been attributed to palaeogeographic differences between the seasonally drier, upland palaeoenvironment of Bromacker and the lowland–coastal and alluvial plain settings of certain localities in North America (Eberth et al. 2000). Therefore, a thorough palaeoecological understanding of both upland and lowland localities and comparisons between them is critical in investigating the evolution and development of terrestrial ecosystems.

In this article, we describe the first trackway made by a large-bodied (glenoacetabular distance >200 mm) vertebrate trackmaker from the Kildare Capes Formation of the Pictou Group, near Eldon, Prince Edward Island. A qualitative and quantitative analysis of published measurements of tracks and trackways from both Prince Edward Island and Europe suggests that the new tracks represent the first occurrence of Ichniotherium sphaerodactylum (Pabst, 1895) in North America and the second occurrence of trackways of this ichnospecies outside of Germany. This analysis also suggests that the previous identification of another trackway specimen from a different locality within the Kildare Capes Formation as ‘Ichniotherium cf. I. willsi’ by Mossman and Place (1989) is incorrect. Additionally, a rare invertebrate trace fossil, Striatichnium bromackerenseMartens, 1982 (originally named Striatichnium bromackeri), co-occurs with the vertebrate trackways. This marks the second occurrence of S. bromackerense outside of Germany and the first occurrence in Canada and allows for a preliminary palaeoenvironmental comparison of Kildare Capes and Tambach formation localities, which is interpreted as a rare, Early Permian upland depositional setting.

Ichnological abbreviations.  A, distance between manus and pes of the same imprint pair; B, width of pace; C, apparent body length (glenoacetabular distance); D, distance between manus and pes imprints of two successive sets; E, distance between the midline of the manus and of the pes within a manus–pes pair; m, manus imprint; mb, width of manus imprint; ml, length of manus imprint; mI to mV, length of digit I to V of the manus imprint; P, length of pace; p, pes imprint; pb, width of pes imprint; pl, length of pes imprint; pI to pV, length of digit I to V of the pes imprint; S, length of stride; α, pace angulation; β, divarication of manus or pes from midline; +, outward rotation; −, inward rotation; γ, interdigital angle I–V; I–II–III–IV–V, first to fifth digit, numbered from medial to lateral side of imprint.

Materials, locality and age

ROM 49281 is a large sandstone block with an ichnofossil-bearing surface that measures approximately 1100 mm by 630 mm. The block was collected by C. Fortune, S. Fortune and D. Prattico from the base of a large cliff along the south-eastern shore of Orwell Bay (UTM: 20T 508115E 5105935N; Fig. 1), near the town of Eldon, Queens County, Prince Edward Island, Canada. The block was not found in situ and was held privately until subsequent donation to the ROM in 1999. The exact stratigraphic provenance within the 25–30 m thick section exposed along the bay is unknown. The specimen contains eight partial tetrapod foot imprints made by two individuals, one slightly larger than the other (Fig. 2), and several invertebrate traces, preserved as natural casts on the asymmetrically ripple-marked surface of the reddish-brown medium-grained sandstone block. An examination of a nearby locality, Point Prim, where previous trackways have been located, indicates a similar depositional environment to that of the Eldon locality, consisting of crevasse-splay sedimentation within a floodplain or overbank deposit (Mossman and Place 1989; Tanner et al. 2005).

Figure 1.

 Map showing the location of the collection site of ROM 49281, near Eldon, Prince Edward Island, Canada. A, Canada. B, Prince Edward Island. C, Hillsborough Bay area. Geological interpretation after Ziegler et al. (2002).

Figure 2.

 ROM 49281, sandstone block containing multiple ichnofossils. A, original sandstone block with positive hyporelief, a natural cast of the trackway. B, diagram of ichnofossils from latex peel of original block, with negative epirelief. Dark grey indicates Trackway 1, light grey indicates Trackway 2. Grey lines connect pes imprints of Trackway 2.

The continental redbeds of the Pictou Group of Prince Edward Island range in age from the latest Pennsylvanian to Early Permian (Virgilian–Wolfcampian). These ages are based on the analyses of vertebrate fossil material (Langston 1963), plant macrofossil data (Ziegler et al. 2002) and spore data (Barss et al. 1963; Hacquebard 1972). The Pictou Group was divided into four fining-upward fluvial megasequences by van de Poll (1983), which were later given formation names (van de Poll 1989). The strata at the Eldon trackway locality belong to the Kildare Capes Formation, the base of which marks the approximate location of the transition from Pennsylvanian to Permian strata (van de Poll 1989). The Kildare Capes Formation represents a fining-upward megacyclic sequence (megasequence II of van de Poll 1983) estimated to be 350 m thick in total, with small-scale, generally fluvial-dominated cycles within the formation (van de Poll 1989). The strata along the Point Prim peninsula, near the Eldon locality, are interpreted as being near the top of the formation (van de Poll 1983; Mossman and Place 1989). Fossils previously reported from the Kildare Capes Formation include plant macrofossils such as Walchia, Tylodendron, Calamites, Pecopteris and Cordaites, numerous microfossils including bladder spores that are typical of the Permian (Dawson and Harrington 1871; Bain and Dawson 1885; Holden 1913; Darrah 1936; Barss et al. 1963; Hacquebard 1972; van de Poll 1983; Ziegler et al. 2002) and vertebrate traces and body fossils (Langston 1963; Mossman and Place 1989; Calder et al. 2004). Vertebrate material is rare and fragmentary (preserved pieces no larger than 50 mm) and includes xenacanth teeth, a premaxilla of EryopsCope 1877, a diadectid dentary and postzygapophysis and an ophiacodontid pterygoid (Langston 1963).

As the fossil record of the Pictou Group is sparse, analyses of floral and faunal occurrences from Prince Edward Island have been limited when considering the biostratigraphic correlation of global Permian strata and understanding climatic changes during the Permo-Carboniferous (Olson and Vaughn 1970; Mossman and Place 1989; Ziegler et al. 2002). The Carboniferous strata of Nova Scotia are well correlated with those of Western Europe (Calder 1998), and the Pictou Group redbeds have been suggested to correlate with the Lower Rotliegend Group (Mossman and Place 1989). Within North America, the Pictou Group correlates stratigraphically with the Wichita and Clear Fork groups of Texas based on botanical and vertebrate remains and trackways (Langston 1963; Olson and Vaughn 1970; Ziegler et al. 2002). The climate was drier during the Early Permian than in the Pennsylvanian of Nova Scotia, as evidenced by the lack of coal beds and the decrease in floral and faunal diversity throughout the Pictou Group (Ziegler et al. 2002; Tanner et al. 2005). The climate is suggested to be seasonally dry and subhumid, with frequent occurrence of ephemeral streams and fluvial action (van de Poll 1983; Ziegler et al. 2002; Tanner et al. 2005).

Institutional abbreviations.  CMN, Canadian Museum of Nature, Ottawa, Ontario, Canada; ROM, Royal Ontario Museum, Toronto, Ontario, Canada.

Systematic ichnology


Ichnogenus ICHNIOTHERIUM Pohlig, 1892

Type ichnospecies.  Ichniotherium cottae (Pohlig, 1885).

Ichniotherium sphaerodactylum (Pabst, 1895)
Figures 2, 3, 4A, 5; Tables 1–3

Figure 3.

 Trackway comparisons. A, Ichniotherium sphaerodactylum. B, Ichniotherium cottae. C, ROM 49281. D, ‘Ichniotherium cf. I. willsi’. Modified from Mossman and Place (1989) and Voigt (2005). Scale bar for A to C represents 50 mm; scale bar for D represents 10 mm.

Figure 4.

Striatichnium bromackerense and indeterminate horizontal burrow, ROM 49281. A, ROM 49281, latex peel of sandstone block giving negative epirelief, with rectangle indicating magnified area in B and C, rotated 100 degrees counterclockwise. B, latex peel of original sandstone block, with negative epirelief. C, diagram of ichnofossils from latex peel. Grey lines are used for the first striae cluster of S. bromackerense, while thin black lines are used for the second. Heavier black lines are used for the borders of the indeterminate horizontal burrow, and the dotted line indicates the leeward edge of the large ripple mark.

Figure 5.

 PCA diagrams. A, PCA of tracks, PC1 = 84 per cent total variation. B, PCA of trackways, PC1 = 61 per cent total variation.

Table 1.   Imprint parameters of Ichniotherium sphaerodactylum on ROM 49281. All measurements are in mm or degrees.
  1. *Indicates estimated value.

1-RM1, RP133486465*60*156170243853513410916493
1-RM2, RP241535962*133 81
2-LM1, LP140404252*43120*135102
2-RM1, RP1363743595913315275324038433010113375
Table 2.   Trackway parameters of Ichniotherium sphaerodactylum, ROM 49281. All measurements are in mm or degrees.
  1. *Indicates estimated value.

1540540*122410 22  
Table 3.   Average apparent body lengths (measurement C) of previously described ichnotaxa present in the Pictou Group, Prince Edward Island, Canada.
IchnotaxonReferenceC (mm)
Ichniotherium sphaerodactylum (Pabst, 1895)This paper418
Dimetropus nicolasi Gand and Haubold, 1984 Van Allen et al. (2005) 270–384
Amphisauropus latus’ Haubold, 1970 Van Allen et al. (2005) 164–193
Gilmoreichnus (Hylopus) hermitanus’ (Gilmore, 1927) Calder et al. (2004) 130
Varanopus microdactylus (Pabst, 1896) Van Allen et al. (2005) 130
Dromopus agilis Marsh, 1894 Van Allen et al. (2005) 125
Notalacerta Butts, 1891 Calder et al. (2004)  90
Limnopus vagus Marsh, 1894 Van Allen et al. (2005) 82–114
Varanopus Moodie, 1929 indet. Van Allen et al. (2005)  66–78
Amphisauropus latus’ Haubold, 1970 Mossman and Place (1989)  63–74
Gilmoreichnus kablikaeHaubold, 1973a Mossman and Place (1989)  55–62
Ichniotherium cf. I. willsiHaubold and Sarjeant, 1973 Mossman and Place (1989)  54

Emended diagnosis.  Digit V relatively long, 73–100 per cent the length of digit IV on the pes and 58–83 per cent the length of digit IV on the manus. Entire outline of digit I of manus commonly narrowly triangular. Distal portions of digits II–IV of manus typically narrowly pointed and angled or sharply bent medially and oriented subparallel to one another. Average length of manus for each individual trackmaker 44–110.5 mm long. Sole-pad of manus generally poorly delineated, elongate, oval, positioned opposite digits II and III, or sometimes elongated proximolaterally to the level of the mid-width of digit IV. Manus sole-pad smaller and narrower than that of pes. Pes sole-pad mediolaterally expanded and ovoid, measuring approximately twice its proximodistal length, which lies opposite digits II–V. Average length of pes imprints for each individual trackmaker 57–144.5 mm long. Trackway pattern variable, ranging from alternating single imprints to alternating manus–pes imprint pairs. Pes imprints at the most marginally overlapping manus. Pace angulation 70–110 degrees. Derived track parameters show a relatively short stride length. Sp:C = 0.8–1.4:1 and Sp:pl = 2.5–4.4:1. Tail drag marks rare, discontinuous if present. Modified after Voigt (2005) and Voigt et al. (2007).

Referred material.  ROM 49281, sandstone block containing six near-complete and two partial foot imprints belonging to two individual trackmakers (Figs 2, 3, 4A and 5; Tables 1–3) in positive hyporelief, and latex peel of the block giving negative epirelief.

Description.  The specimen consists of pentadactyl quadruped tracks with plantigrade imprints, wide and rounded digit tips, and well-preserved digit impressions transected by short curved grooves. Digit lengths on the manus and pes increase from digit I to IV, and digit V is relatively long, 92–100 per cent the length of digit IV on the pes and 67–83 per cent the length of digit IV on the manus. The average length of manus imprints for each individual trackmaker measures 101–110.5 mm long. Digit tips of digits II–IV of the manus are typically narrowly pointed and angled or sharply bent medially and oriented subparallel to one another. The outline of digit I of the manus is sometimes narrowly triangular. The interdigital angle for digits I–V of the manus is 75–102 degrees. Manus length is approximately 80 per cent of manus width and 75 per cent of pes length. The manus imprint has an ovoid, but generally poorly delineated sole-pad with evenly distributed relief. The manus sole-pad is smaller and narrower than that of the pes. The average lengths of the pes imprints for each individual trackmaker are 133–144.5 mm long. Digit tips of the pes are expanded and rounded, giving them a ‘drumstick’-shaped outline. The interdigital angle for digits I–V of the pes is 75–81 degrees. The pes sole-pad has a mediolaterally expanded ovoid heel pad measuring approximately twice its proximodistal length, which lies opposite digits II–V, is generally clearly separated from the digits and has a mediolateral decrease in relief. Imprint parameters are summarized in Table 1.

The trackway pattern preserved shows alternating manus–pes imprint pairs. The manus imprint is positioned anterior to the pes imprint, with very marginal overlap of the manus imprint by the pes imprint. Pace angulation = 103 degrees, Sp:pl = 4:1, and Sp:C = 0.76–0.8:1. Imprints are oriented inwards approximately 22 degrees for the manus, and approximately 36 degrees for the pes. Tail drag marks are absent. Trackway parameters are summarized in Table 2.

Remarks.  The plantigrade pentadactyl quadruped imprints on ROM 49281 are clearly attributable to the ichnogenus IchniotheriumPohlig, 1892 (Fig. 3A–C) based on the rounded digit ends, increasing digit lengths from digits I–IV on both the manus and the pes, poor delineation of the ovoid sole-pad of the manus imprint, clear separation of sole-pad from digital imprints on the pes, and manus imprints with evenly distributed relief, whereas pes imprints, with greater relief overall, show a mediolateral decrease in relief.

Ichniotherium sphaerodactylum differs from other ichnospecies of Ichniotherium in having an elongate digit V (Voigt et al. 2007), and the imprints on ROM 49281 include specimens with the greatest recorded lengths of digit V on the manus relative to digit IV. The trackways on ROM 49281 differ from I. cottae (Pohlig, 1885) and are consistent with I. sphaerodactylum in having a minimal overlap of the pes on the manus in manus–pes imprint pairs (Voigt et al. 2007). The imprints differ from I. willsi in the inward rotation of the pes imprints (Voigt and Ganzelewski 2010) and the lack of the two-part sole and caudal extension of the heel pad (Haubold and Sarjeant 1973; Tucker and Smith 2004). The imprints on ROM 49281 also differ from I. praesidentis (Schmidt, 1956) in lacking several diagnostic features, including the characteristic, distinct oval pad proximal to the first digit of the manus, the outward rotation of the pes imprints and inversely coupled manus–pes imprint pairs (Voigt and Ganzelewski 2010).

Although these tracks correspond to previous definitions of Ichniotherium sphaerodactylum in terms of overall shape, they have necessitated the alteration in some imprint parameters defining the ichnospecies. The imprints are the largest on average yet reported for I. sphaerodactylum. Previous characterizations of I. sphaerodactylym (e.g. Voigt 2005; Voigt et al. 2007) have used a range of means of imprint lengths for individual trackmakers to diagnose the ichnotaxon. The maximum average length of the manus for each individual trackmaker has been increased from 106 mm (Voigt et al. 2007) to 110.5 mm. The maximum average length of the pes imprint for each individual trackmaker has been increased from 129 to 144.5 mm, and the lengths of all pes imprints present on ROM 49281 (133–156 mm) exceed the previously reported average measurements (Voigt et al. 2007). The pace angulation of 103 degrees in Trackway 2 on ROM 49281 falls within the range previously defined for Ichniotherium sphaerodactylum (70–110 degrees; Voigt et al. 2007). The diagnostic, derived trackway parameters Sp:pl and Sp:C also fall within the ranges previously defined for I. sphaerodactylum by Voigt et al. (2007).

The presence of Ichniotherium sphaerodactylum improves our understanding of the vertebrate ichnological diversity of the Kildare Capes Formation. This formation also includes a specimen of five tracks previously identified as ‘Ichniotherium cf. willsi’ (Fig. 3D) based on the sprawling gait, pace angulation and the presence of a deflated ‘cushion’ in the area of the calcaneum (Mossman and Place 1989). In their description, they noted the possibility that the specimen could represent a variant of ‘Gilmoreichnus’ (Haubold inMossman and Place 1989), an opinion shared by Calder et al. (2004). The similarity of these tracks to those of Amphisauropus sp. from the Brule locality, Nova Scotia, has also been previously noted (Van Allen et al. 2005). However, the presence of claw marks excludes these imprints from identification as a variant of Amphisauropus sp. (Voigt 2005). Although the original specimen of ‘Ichniotherium cf. I. willsi’ has been lost (D. J. Mossman, pers. comm. 2010), an examination of the photographs and measurements suggest that the tracks may represent small specimens of Dimetropus sp. or Varanopus sp. based on the overall size (with the smallest apparent body lengths of all vertebrate ichnotaxa known from the Pictou Group, Table 3), slender clawed digits and a prominent, fairly continuous tail drag (Mossman and Place 1989), which are all uncharacteristic of Ichniotherium (Voigt 2005). We therefore consider ROM 49281 to represent the only definitive occurrence of the ichnogenus Ichniotherium from the Kildare Capes Formation.

At the Bromacker locality, two ichnospecies of Ichniotherium, I. cottae and I. sphaerodactylum, have been associated with their diadectid trackmakers, Diadectes absitusBerman, Sumida, and Martens, 1998 and Orobates pabstiBerman, Henrici, Kissel, Sumida, and Martens, 2004, respectively (Voigt et al. 2007). One dentary (CMN 9918) and one postzygapophysis (CMN 9919) attributed to Diadectidae, described by Langston (1963), were found near Lord Selkirk Provincial Park, which is in close proximity to the Eldon locality. Only the dentary is illustrated, and neither specimen could be located in the collections of the CMN for re-examination. Therefore, at this time, it is not possible to name a specific trackmaker for the Eldon I. sphaerodactylum tracks, although given the similarities between the Bromacker locality and the Eldon locality, it is possible that future work will reveal specimens clearly attributable to a particular diadectid species.

The occurrence of Ichniotherium sphaerodactylum from Prince Edward Island marks the most well-preserved record of the ichnospecies and the first appearance of trackways outside of Germany. A single pes imprint of I. sphaerodactylum recently described from Morocco (Voigt et al. 2011) suggests that Ichniotherium is more geographically widespread than previously thought, but is still rare compared to other Early Permian ichnofaunae (Lucas 2007).

Ichnogenus STRIATICHNIUM Walter, 1982

Type ichnospecies.  Striatichnium natalis Walter, 1982.

Striatichnium bromackerenseMartens, 1982
Figures 2A, 4

  • 1963 ‘Rippenbündel (zu Genus inc. spiralis A. H. Müller gehörend?)’ (‘bundle of ribs (belonging to the genus that includes spiralis A. H. Müller?)’) Steiner and Schneider, p. 724, figs 4, 5.

  • 1969 ‘Fächerförmige, sehr flache, distal meist gegabelte Kratzspuren, Weidespuren, vermutlich zu Tambia spiralis gehörend’ (‘Fan-shaped, very flat, mostly distally bifurcated scratches, grazing traces, probably belonging to Tambia spiralis’) Müller, p. 928, pl. 2.4.

  • 1973b ‘Fiedermarken’ (‘Pinnate marks’) Haubold, p. 251, pls 5–8.

  • 1974 ‘Grundtyp E (Striemenfächer)’ (‘Basic type E (fan of striae)’) Martens, p. 29, figs 14, 21, pls 15.1–15.7, 16.1–16.6, 17.1–17.3, 18.1–18.3, 19–22.

  • 1975 ‘Haupttyp E (Striemenfächer)’ (‘Primary type E (fan of striae)’) Martens, p. 118, pls 4.1–4.3.

  • 1981 ‘Haupttyp E (Striemenfächer)’ (‘Primary type E (fan of striae)’) Martens, p. 27, figs 4.1–4.3.

  • 1981 ‘Striemenfächer’ (‘fan of striae’) Martens, Schneider, and Walter, p. 88, fig. 10.

  • * 1982 Striatichnium bromackeriMartens, 1982; Martens, p. 41, pl. 5.5.

  • 2000 Striatichnium bromackerenseMartens, 1982; Voigt and Haubold, p. 20, fig. 3g.

  • 2002 Striatichnium bromackerenseMartens, 1982; Voigt, p. 46, pl. 1.1.

  • 2005 Striatichnium bromackerenseMartens, 1982; Voigt, Small, and Sanders, p. 347, figs 6d, 7d.

  • 2009 Striatichnium bromackeri Minter and Braddy, fig. 33d.

Emended diagnosis.  Predominantly band-like systems of striae composed of rhythmically repeated, superimposed striations arranged in a fan-like manner, usually with distal bifurcation. Striae approximately 1 mm in width. Distances between the parallel to slightly divergent striae vary, from complete overlap up to 5 mm. The width of these systems varies between 40 and 140 mm. (Modified after Martens (1982) and Voigt et al. (2005)).

Referred material.  ROM 49281, sandstone block containing a trace of Striatichnium bromackerense (Figs 2A and 4) in positive hyporelief and latex peel of the block giving negative epirelief.

Description.  The trace is composed of nearly bilaterally symmetrical clusters of curvilinear, slightly convex-inward striations roughly 1 mm wide in a fan-like arrangement. Striae often appear to exhibit distal bifurcation, although this is difficult to differentiate from slight, low-angle overlap of striae. The band-like system of striae results from the superpositioning of fans of striae in the longitudinal direction. It is possible that two separate but very closely spaced striae clusters comprise the band-like trace present in ROM 49281, with each striae cluster corresponding to a period of movement of the producer, with a brief interruption between them; however, it is also possible that the two striae clusters together represent a single system of superimposed fans created in a single, extended period of movement extending over 145 mm of substrate, although this cannot be determined with certainty owing to a separate crosscutting horizontal burrow. The trace was made on the lee side of a large ripple mark, which is visible along the left margin of the trace (Fig. 4, with leeward edge indicated by dotted line).

The first striae cluster is poorly preserved (Fig. 4, indicated in grey), and possibly worked over by the second striae cluster (indicated in black) and unidentified horizontal burrows (indicated in thicker black lines). Approximately 30 recognizable striations are preserved. The cluster is approximately 4 mm deep when measured from the peel of ROM 49281, and 96 mm wide. The total length of the apparent cluster is indeterminate owing to a crosscutting horizontal burrow towards the distal edge. The striae cluster spans a sector of roughly 80 degrees.

The second striae cluster is better preserved, although still partially crosscut towards the proximal end by the same horizontal burrow that obscures the margins of the first. It measures approximately 95 mm in width and an estimated 80 mm in length. The cluster includes approximately 70 identifiable striations and also spans a sector of roughly 80 degrees. Individual striations are up to 78 mm in length, although they may have been longer before being obscured by the crosscutting horizontal burrow. The striations are fairly superficial, and the cluster forms only an extremely shallow depression in the surface of <2 mm when measured on the latex peel of ROM 49281.

Nomenclatural note. Martens (1982) originally named Striatichnium bromackerense as Striatichnium bromackeri. The suffix of the specific epithet was altered upon the first subsequent publication (Martens 1988, p. 935) to reflect its etymological reference to the type locality, with no change in attribution, and thus S. bromackerense became an incorrect subsequent spelling (ICZN 1999, Article 33.3). The revised suffix was adopted by other nearly all subsequent authors that mention the ichnospecies (Martens 2000, p. 26; Voigt and Haubold 2000, p. 20, fig. 3g; Martens 2001a, p. 56; Martens 2001b, p. 190; Voigt 2002, p. 46, plate 1.1; Voigt, 2005, pp. 23, 27; Voigt, Small, and Sanders 2005, p. 347, figs. 6d, 7d; Buatois and Mángano 2007, table 17.2; not adopted by Minter and Braddy 2009, p. 56, fig. 33d), and should now be considered a correct original spelling as a result of prevailing usage, in accordance with Article 33.3.1 (ICZN 1999).

Remarks.  The ichnogenus Striatichnium is rare, and according to Minter and Braddy (2009), includes four ichnospecies: S. natalisWalter, 1982; S. bromackerenseMartens, 1982; S. irregularisWalter and Hoffman, 2001; and S. biflabellisMinter and Braddy, 2009. The ichnogenus is known from the Early Permian of Germany (Walter 1982; Martens 1982; Voigt and Haubold 2000; Walter and Hoffman 2001; Voigt 2002; Minter et al. 2007) and New Mexico (Braddy 1998; Braddy and Briggs 2002; Minter and Braddy 2009) and the Permo-Carboniferous of Colorado (Voigt et al. 2005). Other reports of specimens resembling or assigned to Striatichnium were rejected by Minter and Braddy (2009) and referred to MonomorphichnusCrimes, 1970 (Brooks, 2001; Brooks et al. 2003; Morrissey and Braddy 2004) or are rejected here (Chakrabarti and Baskaran (1989), an arching, striated trace from the Pleistocene of India that does not resemble any known ichnospecies of Striatichnium; Marriott et al. (2009), which is also best referred to Monomorphichnus based on morphology and the reassignment of similar traces from the same group (Morrissey and Braddy 2004) by Minter and Braddy (2009)). Striatichnium bromackerense was previously known only from a few Early Permian localities in Germany, including the Bromacker horizon within the Tambach Formation, the stratigraphically equivalent Elgersburg Formation and the slightly older Goldlauter Formation (Walter 1982; Martens 1982; Voigt and Haubold 2000), and from the Maroon Formation in Colorado (Voigt et al. 2005).

The striated trace on ROM 49281 is attributable to Striatichnium bromackerense and is easily distinguished from other ichnospecies of Striatichnium. Striatichnium bromackerense differs from S. natalis in its larger size, lesser curvature of striae, the narrower angle at which the striae meet in each striae cluster and the large central zone of alternating overlap of striae (Walter 1982; Martens 1982; Minter and Braddy 2009). The regularly bundled, slightly curved pattern of striae differentiates S. bromackerense from S. irregularis, which exhibits irregularly spaced, linear scratches (Walter and Hoffman 2001). Striatichnium bromackerense differs from S. biflabellis in the lower convex-inward curvature of the striae and the absence of the central nonstriated area characteristic of the latter ichnospecies (Minter and Braddy 2009). The term ‘cluster’ is used by Minter and Braddy (2009) to describe the striated regions on each side of the unstriated midline area of S. biflabellis, whereas here it is used to describe smaller fans of striae arranged anteroposteriorly, along the long axis of S. bromackerense. The oblique-to-perpendicular scratches reported in specimens of S. bromackerense by Voigt et al. (2005) from Colorado are absent in the specimen from Prince Edward Island, and perhaps the markings on the Colorado specimen represent a separate, independent trace superimposed over the striated trace.

Previous characterizations of Striatichnium bromackerense have included the ‘opening angle’ (20–40 degrees; Martens 1982) or ‘dihedral angle’ (50–60 degrees; Voigt et al. 2005) of the fans, a measure that was found to have a maximum of approximately 80 degrees in the trace on ROM 49281. In all other aspects, the trace exhibits features in accordance with all other defining characteristics of S. bromackerense. The extremely wide range of values for this measurement and the changing angle of the striae mediolaterally through the trace cast doubt on the reliability of this measurement as a diagnostic characteristic of this ichnospecies. Thus, we have chosen to eliminate this measurement from the revised characterization of the ichnospecies.

Length of the trace cannot be used as a diagnostic characteristic of Striatichnium, as the length is dependent upon the duration of the tracemaking behaviour. The maximum length of S. bromackerense has been measured at 500 mm (Martens 1982). In some cases, only single, shorter, individual clusters of striae are preserved, as described by Voigt et al. (2005).

An assortment of tracemakers and tracemaking behaviours for the various ichnospecies of Striatichnium have been proposed (e.g. Martens 1975, 1982; Walter 1982; Schneider 1983; Walter and Hoffman 2001; Simon et al. 2003; Wisshak et al. 2004), although most agree with the general interpretation of the trace being produced by some form of arthropod using its appendages to rake the sediments while moving across the substrate in order to feed, most likely subaqueously (Voigt 2002; Minter and Braddy 2009).

Morphometric analysis of vertebrate tracks and trackways

Previous analyses of Ichniotherium have used ratios of measured data to compare and contrast ichnospecies (Voigt 2005; Voigt et al. 2007; Voigt and Ganzelewski 2010). This method is useful for comparing the average size of tracks, but to understand the variability of track and trackway measurements within and between ichnospecies and to visualize these parameters in morphospace, a multivariate analysis is more informative.

Multivariate analyses have had limited use in the description of tetrapod tracks and trackways. Previous studies have used Factor Analyses and Discriminant Analyses to differentiate between different genera of Cretaceous tridactyl dinosaur footprints (Moratalla et al. 1988) and Cluster Analyses to differentiate Carboniferous ichnospecies of Limnopus based on quantitative and qualitative characters (Tucker and Smith 2004). A recent analysis on tridactyl dinosaur footprints used biometric and outline analyses visualized with a multivariate principal components analysis (PCA) to compare tracks from different localities (Moreau et al. 2012). Here, to view measured parameters in morphospace and to understand which ichnotaxa are most similar in shape and size, a PCA is performed.

Data sets (Appendices S1 and S2) comprising the average measurements of 62 trackways were compiled from measurements of the Eldon tracks (Tables 1, 2) and from the published literature. Measurements of the Eldon tracks and trackways were taken following the methods of previous authors (Haubold 1971; Leonardi 1987; Voigt and Haubold 2000). Measurements of I. cottae and I. sphaerodactylum were taken from Voigt et al. (2007), and measurements of I. praesidentis were taken from Voigt and Ganzelewski (2010). Average measurements of specimens previously described as ‘Ichniotherium cf. I. willsi’, ‘Amphisauropus latus’ and ‘Gilmoreichnus kablikae’ from the Kildare Capes Formation at Point Prim, Prince Edward Island, were taken from the published measurements of Mossman and Place (1989). Measurements of specimens previously identified as ‘Gilmoreichnus kablikae’, ‘Amphisauropus latus’ and I. willsi from Europe were taken from Haubold (1971, 1973a) and Haubold and Sarjeant (1973).

The use of measurement data from different authors introduces the possibility of measurement error in the data set. For this study, 89 per cent of the data used were collected by one author (Voigt et al. 2007; Voigt and Ganzelewski 2010), and re-collecting measurements for the other material from Prince Edward Island described by Mossman and Place (1989) is impossible at this time as one of the slabs has been lost. It is likely that any error in measurement in the data set will be outweighed by the taxonomic signal present in the data.

Although certain angles were measured for the description of ROM 49281, they were not included in the morphometric analysis, as those data did not conform to the assumption of linearity required for a PCA (Harvey-Collier test for linearity; Harvey and Collier 1977). Additionally, not all angle measurements are independent of the linear measurements, and so the inclusion of the angle measurements in the analyses would place undue additional emphasis on those variables. To account for large variances in some variables, all PCA data were standardized using a correlation matrix.

As not all measurements could be collected for each track and trackway owing to poor preservation, missing data were estimated using a Bayesian principal components analysis (BPCA, Oba et al. 2003), which has proven most effective at calculating missing data in morphometric analyses at small sample sizes (Brown et al. 2012). The use of BPCA allows for the inclusion of all available data into the analysis, including the ichnotaxon of interest. The amount of missing data in this analysis is low (0.02 per cent), and so the error rate of calculating missing variables is low (Brown et al. 2012). To test that the estimated data are not skewing the results of the analyses, both analyses were also performed without missing data. All estimated data are in bold in Appendices S1 and S2. The track measurements (Appendix S1) were analysed separately from the trackway measurements (Appendix S2).

All morphometric analyses were completed using the program R ( with the packages vegan (Oksanen et al. 2010), lmtest (Zeileis and Hothorn 2002) and pcaMethods (Stacklies et al. 2007).

Results.  The results of the PC analyses with and without estimated data are nearly identical (Appendices S3 and S4), and therefore only the results of the analyses containing data estimated using BPCA will be discussed further, as the data from ROM 49281 contain estimated values. The results of the PC analyses of both tracks and trackways indicate that the majority of the variation in the data set is captured by PC1, which is interpreted as being driven by size (Fig. 5). For the track measurements (Fig. 5A), PC1 accounts for 95 per cent of total variation, and PC2 accounts for 2 per cent. The loadings of PC1 are all variables relating to size, and PC2 is weighted positively by the length of digit V of the pes and negatively by the length of digit IV of the manus. PC3 to PC14 did not show any significant differentiation among the variables. For the trackway measurements (Fig. 5B), PC1 accounts for 95 per cent of the total variation and is driven by all size variables. The variables loading PC2 (2 per cent of the total variation) are positively weighted by the distance between manus and pes of the same imprint pair and the width of pace in both the manus and pes, and negatively weighted by the length of the stride in both the manus and pes. PC3 to PC8 did not show any significant differentiation among the variables.

Remarks.  Here, a PCA on linear measurements of tracks (Fig. 5A) shows that the main differences between ichnospecies of Ichniotherium are based on size, which is typical for linear morphometric analyses. The overlap between track measurements of all four ichnospecies of Ichniotherium is expected, as all are similar in size. Overlap between Ichniotherium and Amphisauropus is also expected, considering the similarities in size and morphology between the two ichnotaxa (Hunt et al. 2005), although the sample size of Amphisauropus included in this analysis is too small to accurately test this hypothesis (Fig. 5A). One of the distinguishing characteristics of I. sphaerodactylum, the elongation of digit V, is heavily weighted on PC2. This explains the placement of the Eldon tracks relative to the other ichnospecies of Ichniotherium, as the Eldon tracks have the longest measured digit V of any specimen included in this analysis.

Differentiation based on trackway measurements (Fig. 5B) is similar to that of track measurements, but the main difference after stride length (a proxy for size, PC1) is the distance between the manus and pes of the same imprint pair, which is one of the characters used to differentiate ichnospecies of Ichniotherium (Voigt 2005; Voigt et al. 2007; Voigt and Ganzelewski 2010). Therefore, the position of the I. praesidentis trackway in the PC morphospace is indicative of the inversely coupled manus–pes imprint pattern of the trackway. Trackways of I. sphaerodactylum are positioned between I. praesidentis and I. cottae, as minimal overlap is present between the manus and pes of the same imprint pair in I. sphaerodactylum, and tracks of I. cottae show substantial overlap (Voigt 2005; Voigt et al. 2007).

In both analyses, the new Ichniotherium tracks from Prince Edward Island plot within the range of variation in I. cottae, I. sphaerodactylum and I. praesidentis (Figs 5A, B). Ichniotherium willsi from Europe is also similar to the other European Ichniotherium tracks and trackways, although a larger sample size would help better understand the variation among these three species. The tracks previously described from Prince Edward Island as ‘Ichniotherium cf. I. willsi’ plot more closely with those of ‘Gilmoreichnus kablikae’ and ‘Amphisauropus latus’ than with any ichnospecies of Ichniotherium.

A geometric morphometric analysis would be interesting for future analyses of these trackways, as the main goal of that methodology is to analyse shape change irrelevant of size. This method has been successful for the analyses of dinosaur trackways (i.e. Rodrigues and dos Santos 2004) and would allow for the quantification of potentially significant characters, such as the characteristic ‘drumstick’ shape of Ichniotherium digits. This could be particularly useful for distinguishing ichnospecies of Ichniotherium, and for distinguishing between other similar ichnotaxa, such as Amphisauropus.

Stratigraphic and palaeoecological significance

The Early Permian ichnospecies Ichniotherium has a Pangaean distribution, with occurrences in the Czech Republic, Germany, Great Britain, France, Italy, Morocco, Poland, the USA and now Canada (Haubold and Sarjeant 1973; Mossman and Place 1989; Hunt et al. 1995; Voigt 2005; Voigt et al. 2005; Gand and Durand 2006; Santi 2007; Voigt et al. 2011; Voigt et al. 2012). The most abundant and best-preserved tracks are from the Bromacker quarry of the Tambach Formation, Upper Rotliegend Group, in Germany (Voigt and Haubold 2000; Voigt 2005). There, two ichnospecies, I. cottae and I. sphaerodactylum, have been associated with body fossils of their trackmakers, Diadectes absitus and Orobates pabsti, respectively (Voigt et al. 2007). Until recently, I. sphaerodactylum and O. pabsti were restricted to the Tambach Formation, but the discovery of I. sphaerodactylum in Morocco suggests that the tracks and associated diadectid trackmaker may be more widespread than previously thought (Voigt et al. 2011).

An invertebrate trace fossil also previously thought to be restricted to the Tambach Formation is Striatichnium bromackerense, a presumed arthropod swimming and/or grazing trace (Martens 1982). Its discovery in the Maroon Formation, Colorado, USA, alongside tracks of I. cottae, suggests faunal similarities between the Tambach Formation and some Early Permian sites in North America and increases the occurrence of lesser known upland Permian palaeoenvironments (Voigt et al. 2005).

The presence of Ichniotherium sphaerodactylum at the Eldon locality in Prince Edward Island represents a geographic range extension for the ichnospecies, which was previously known only from the Tambach Formation of Germany and a single pes imprint from the Khenifra Basin of Morocco (Fig. 6; Voigt et al. 2011). This occurrence in Prince Edward Island also marks the first in present-day North America, and the most western occurrence in the Early Permian equatorial Pangaea. The presence of Striatichnium bromackerense in Prince Edward Island marks its second occurrence in North America, and only its third occurrence worldwide.

Figure 6.

 Palaeogeographic occurrences of Ichniotherium sphaerodactylum in the Early Permian of central Pangaea. A, Eldon locality, Prince Edward Island, Canada. B, Bromacker locality, Germany. C, Khenifra Basin locality, Morocco. Grey represents Early Permian continental area. Map after Voigt et al. (2011).

The occurrence of both Ichniotherium sphaerodactylum and Striatichnium bromackerense in the Early Permian Kildare Capes Formation of Prince Edward Island is significant for a number of reasons. These two ichnospecies occur in the same stratum at one other locality: the Bromacker quarry, Tambach Formation, Germany (Voigt and Haubold 2000), although Ichniotherium cottae and S. bromackerense have been found in association at the Gast footprint locality, Maroon Formation, USA (Voigt et al. 2005). Both the Bromacker and the Gast localities have been interpreted as upland, semi-arid to arid depositional environments, where herbivorous diadectids are the dominant vertebrates (Eberth et al. 2000; Voigt et al. 2005). The semi-arid climatic interpretation of the Pictou Group redbeds (van de Poll 1983; Tanner et al. 2005), the floral and faunal occurrences (Langston 1963; Olson and Vaughn 1970; Ziegler et al. 2002) and similar preservation of I. sphaerodactylum and S. bromackerense suggest that the Eldon locality is ecologically similar to the Bromacker locality. Although the fossil evidence to date is sparse, the presence of diadectid skeletal material at a nearby locality (Langston 1963) and the tracks described here make diadectids the most well-represented terrestrial vertebrate group from the Kildare Capes Formation. Therefore, further sampling of the Kildare Capes Formation may provide additional insights into the palaeoecology of rarely preserved upland environments in the Early Permian.

The Ichniotherium sphaerodactylum tracks from Prince Edward Island are also significant as they preserve definite evidence of large-bodied terrestrial vertebrates in the Kildare Capes Formation and represent the largest animals known from the Pictou Group (Table 3). Although diadectid and ophiacodontid body fossil material has been collected from this formation previously (Langston 1963), the material is extremely fragmentary, and so body sizes for these animals cannot be accurately estimated. The trackways do allow for an estimate of body size and suggest that, even though the vertebrate ichnological record may be biased towards smaller tracks and trackways (Van Allen et al. 2005), larger taxa were in fact present in the fauna.

The co-occurrence of Ichniotherium sphaerodactylum and Striatichnium bromackerense in the Kildare Capes Formation may also assist in the characterization and comparison of Permian ichnofacies (i.e. Hunt and Lucas 1998, 2007). The Eldon locality appears to belong to the inland-distal alluvial fan ichnocoenosis of the Early Permian Batrachichnus ichnofacies of Hunt and Lucas (2005), which is characterized by very few DimetropusRomer and Price, 1940 and an abundance of Ichniotherium (Hunt et al. 1995; Hunt and Lucas 2007). The distribution of Ichniotherium is suggested to be facies-controlled, as it is only recorded in inland and upland palaeoenvironments to date (Hunt and Lucas 1998).

Although efforts have been made to use vertebrate ichnofossils in biostratigraphic correlation (Hunt and Lucas 2007), this practice has been criticized by other authors (Gand and Durand 2006; Santi and Nicosia 2008). Some vertebrate ichnospecies have a large temporal distribution, which is not useful for biochronologic correlations, or others, such as Ichniotherium, are too rare to be of widespread use (Gand and Durand 2006). However, the presence of Ichniotherium in Europe, Africa and now North America suggests it may be more useful as a palaeoenvironmental indicator than previously thought. The earliest Permian Brule locality of Nova Scotia, Canada, which is suggested to be contemporaneous with or slightly older than the Pictou Group redbeds of Prince Edward Island, is dominated by trackways made by carnivores (Van Allen et al. 2005). The depositional setting of the Brule locality is interpreted as an abandoned freshwater distributary channel subject to frequent flooding that was later colonized by a walchian flora (Van Allen et al. 2005). The interpretation of the Pictou Group redbeds is similar, where the area was subject to frequent, ephemeral fluvial action (Tanner et al. 2005). The difference in ichnotaxa between the Brule and Eldon localities could be attributed to climate, as it became drier from the Pennsylvanian to the Early Permian as seasonal rainfall was affected by a shift in the morphology of the depositional basin through tectonic action (Calder 1998; Tanner et al. 2005). This would support the suggestion that diadectids did prefer more upland environments, as preserved in other localities (Eberth et al. 2000), making Ichniotherium useful as palaeoenvironmental marker (Hunt and Lucas 1998), but additional sampling in upland palaeoenvironments of the Kildare Capes Formation and other units are required to demonstrate that this association is robust.


The addition of Ichniotherium sphaerodactylum to the vertebrate ichnofauna of the Kildare Capes Formation provides evidence of rare, large-bodied terrestrial vertebrates, but the taxonomic status of the previously described small-bodied ichnotaxa still remains unresolved. Specimens previously described as ‘Gilmoreichnus kablikae’ and ‘Amphisauropus latus’ by Mossman and Place (1989) should likely be referred to Amphisauropus kablikae, following the synonymization proposed by Voigt (2005). The specimen described as ‘Ichniotherium cf. I. willsi’ by Mossman and Place (1989), is clearly not attributable to Ichniotherium and may represent a variant of Dimetropus sp. or Varanopus sp. Other vertebrate ichnospecies from the Pictou Group of Prince Edward Island include two ichnotaxa from the younger Hillsborough River Formation, identified by Calder et al. (2004): Notolacerta isp., and a specimen identified as ‘Gilmoreichnus cf. G. (Hylopus) hermitanus’, which is now considered a subjective synonym of Limnopus, Varanopus, Dimetropus and Amphisauropus (following S. Voigt 2005, pers. comm. 2011).

The use of PC analyses to visualize the measured track and trackway parameters in morphospace was successful in this study. The characters that distinguish ichnospecies of Ichniotherium from each other in qualitative descriptions match the results of the PC analyses; therefore, the results of the analyses also support the assignment of the new tracks from Prince Edward Island to I. sphaerodactylum.

The association of the rare invertebrate ichnofossil Striatichnium bromackerense with Ichniotherium sphaerodactylum strengthens the correlation between the facies of the Kildare Capes Formation and the Tambach Formation of Germany. Accurate description of the fauna is essential in determining characteristics of upland vs. lowland palaeoenvironments, and the interactions between Early Permian trackmakers and their environment in the earliest well-established terrestrial ecosystems consisting of tetrapod herbivores and carnivores. The occurrence of these new ichnofossils at the Eldon locality adds new data that will help understand palaeoenvironmental changes through the Permo-Carboniferous strata of the Maritimes basin (Calder 1998), and the evolution and development of early terrestrial ecosystems.


Acknowledgements.  The authors sincerely thank the following people: Two anonymous reviewers, K. Angielczyk, and S. Voigt for constructive comments; C. Fortune, S. Fortune and D. Prattico for collecting ROM 49281; G. Hrynewich and H.-D. Sues for facilitating the donation of the specimen to the ROM; H. E. Kristmanson (Aboriginal Affairs and Archaeology, Government of Prince Edward Island) for a permit to conduct palaeontological investigations; K. Seymour (ROM), M. Currie and K. Shepherd (CMN) for specimen access; D. Dufault, B. Iwama and S. Murphy for technical assistance; B. Boyle for specimen photography; H. Haubold, T. Martens, D. J. Mossman and S. Voigt for access to publications and helpful discussion; C. Brown, N. Campione, S. Modesto, L. O’Brien, R. Reisz, M. Vavrek and M. V. H. Wilson for discussions; and R. A. Lyon for translation assistance. This research was supported by Natural Sciences and Engineering Research Council of Canada (NSERC) postgraduate scholarships to KSB and JRH, and an NSERC Discovery Grant to DCE.

Editor. Kenneth Angielczyk