Palaeomagnetic analysis

We have chosen to develop three distinct palaeogeographies based on the interpretations noted above. These palaeogeographies are based primarily on palaeomagnetic data, but geological information (e.g. ages of rifting and collision) is used to further constrain the positioning of blocks in the absence of palaeomagnetic data. In essence, we follow the previously published reconstructions of Torsvik et al. (1996), Kirschvink et al. (1997), Pisarevsky et al. (2000), Meert & Van der Voo (2001) and Torsvik & Rehnström (2001) with slight modifications. Euler poles for each of the reconstructions are given in the Appendix along with the palaeomagnetic poles used in the analysis. As with all palaeomagnetically based pre-Mesozoic reconstructions, there is additional uncertainty in determining relative palaeolongitude along with a hemispheric ambiguity (e.g. choice of polarity).

Palaeobiogeographical analysis

To use fossils to aid in palaeogeographical reconstructions we follow the principles described by McKerrow & Cocks (1986), Lieberman (2000) and Cocks & Torsvik (2002). Further, we use phylogenetic relationships and phylogenetic biogeographical methods to analyse the palaeobiogeographical data, following the approaches of Young (1990), Fortey & Cocks (1992), Lieberman & Eldredge (1996) and Lieberman (1997, 2000, 2003).

More than 115 species of olenellid trilobites were considered in the biogeographical analysis. These species occur in the early Cambrian Fallotaspis, Nevadella and Bonnia–Olenellus zones. Taxa evaluated occur on many of the major early Cambrian cratons including Armorica, Australia, Avalonia, Baltica, East Antarctica, eastern, SW and NW Laurentia (with Laurentia including the Precordillera terrane), north Africa and Siberia, and they have been subjected to phylogenetic analysis by Lieberman (1998, 1999, 2001, 2002). Eight species are from Siberia, 14 are from northern Africa–Armorica, four are from Avalonia, 26 are from SW Laurentia, 31 are from NW Laurentia, 26 are from eastern Laurentia, 10 are from Baltica, two are from Antarctica, and one is from Australia. (Some species occur in more than one region.) Taxa analysed and their areas of occurrence have been described by Lieberman (2003) or are available from B.S.L. on request.

The biogeographical analysis used a modified version of Brooks Parsimony Analysis that has been described in detail by Lieberman & Eldredge (1996) andLieberman (1997, 2000). The method converts phylogenies of organisms, in conjunction with their geographical distributions, into two separate data matrices to reconstruct biogeographical patterns. One, the vicariance matrix, is analysed to retrieve repeated episodes of vicariance precipitated by geological processes that isolate formerly contiguous regions. Such processes include continental rifting and sea-level fall (in the case of trilobites). The other, the geo-dispersal matrix, can be analysed to retrieve episodes of congruent range expansion or geo-dispersal (Lieberman & Eldredge 1996) precipitated by geological processes that join formerly separated regions (which in trilobites include continental collision and sea-level rise). The vicariance and geo-dispersal matrices generated from this analysis have been described by Lieberman (2003) or are available from B.S.L. on request.

The data matrices are analysed using the parsimony algorithm paup 4.08b (Swofford 2001), with results expressed as most parsimonious vicariance and geo-dispersal trees, with each tree rooted using an outgroup; the closer two regions sit on a tree, the more recently they shared a common history. For example, a close relationship on the vicariance tree implies that regions became separated relatively recently. Similarities between the vicariance and geo-dispersal tree indicate that sea-level rise and fall had an important influence on biogeographical patterns (Lieberman & Eldredge 1996; Lieberman 1997, 2000). By contrast, differences between the two trees suggest that such repeating processes had less of an affect on biogeographical patterns and instead suggest that tectonic processes such as continental rifting or collision may have had a more fundamental effect. Parsimony based analytical methods have been frequently and successfully applied to the analysis of palaeobiogeographical patterns (e.g. Fortey & Cocks 1992; Lieberman & Eldredge 1996; Lieberman 1997; Waggoner 1999). Such methods, when implemented using paup 4.08b (Swofford 2001), can also be combined with jackknife, bootstrap and Bremer branch support analyses (Bremer 1994) to consider the degree of support for various parts of the biogeographical tree. Further, tree length frequency distributions (Hillis 1991) and cladistic permutation tail probabilities (Faith 1991; Swofford et al. 1996) also can be generated to consider how strong the biogeographical signal is in the data.

Palaeobiogeographical patterns

The analysis of the two data matrices using the exhaustive search option of paup 4.08b (Swofford 2001) produced one most parsimonious vicariance tree (Fig. 4) and six most parsimonious geo-dispersal trees; the strict consensus of the geo-dispersal trees is poorly resolved and therefore is not shown. Tree length frequency distributions and cladistic permutation tail probabilities each suggest a strong biogeographical signal in the vicariance matrix, with values differing from random data minimally at the 0.01 and 0.001 levels, respectively (Lieberman 2003). The general lack of resolution in the consensus geo-dispersal tree indicates that there were few congruent episodes of range expansion between different regions (as a result of either sea-level rise or continental collision) by the trilobites considered in this analysis. Further, the lack of similarity between the vicariance and consensus geo-dispersal trees suggests that tectonic events, rather than repeated episodes of sea-level rise and fall, most profoundly structured the biogeographical patterns in the trilobites studied.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

 The most parsimonious vicariance tree showing palaeobiogeographical patterns in early Cambrian olenellid trilobites. The tree shows the relative time at which regions become isolated from one another as a result of the emergence of geographical barriers. The closer two regions sit on the tree, the more recently the geographical barriers emerged between those regions, isolating their respective trilobite faunas. The tree is rooted using an outgroup.

There are three major biogeographical groupings: Australia and East Antarctica; Baltica, eastern and NW Laurentia; and Siberia, north Africa–Armorica, SW Laurentia and Avalonia. The various parts of the vicariance tree are well supported based on a variety of tests conducted using paup, with the East Antarctica–Australia grouping being the least well supported (Lieberman 2003). This, however, is because of the relatively limited number of basal redlichiine trilobites from these regions that were sampled in this analysis. Additional analyses based on trilobites (Palmer & Rowell 1995) and also archaeocyathans (Debrenne & Kruse 1986) strongly reinforce the biogeographical grouping between East Antarctica and Australia. This part of eastern Gondwana is thus resolved as a distinct biogeographical region. Western Gondwana and peri-western Gondwana also group together, along with Siberia and SW Laurentia. Laurentia itself is a polyphyletic biogeographical region, with parts of the craton possessing trilobite faunas sharing a closer biogeographical history with Baltica, and other parts sharing a closer biogeographical history with western Gondwana, peri-western Gondwana and Siberia.

The notion that tectonic events are driving the palaeobiogeographical patterns is consistent with the palaeomagnetic observations described above. Despite the lack of consensus regarding the exact palaeogeography, all of the models indicate that the opening of the Iapetus Ocean was the major tectonic event in the latest Neoproterozoic. In fact, it was the ubiquitous presence of passive margin sequences worldwide that led to the suggestion by Bond et al. (1984) that the Neoproterozoic heralded the breakup of a supercontinent. Of secondary importance was the closure of the Mawson Sea during the final stages of Gondwana assembly.

Palaeobiogeographical patterns and timing of divergence events

The vicariance tree indicates several vicariance events in early Cambrian trilobites, and there are three major palaeobiogeographical groupings: (1) eastern Gondwana; (2) Baltica, and eastern and NW Laurentia; (3) Siberia, the northern margins of western Gondwana and SW Laurentia. Early vicariance events separated these three regions and their trilobites from one another either all at once or in an order that cannot be resolved. Later, vicariance separated Australian and Antarctic trilobites, Baltic, and eastern and NW Laurentian trilobites, and Siberian, SW Laurentian, Avalonian and north African–Armorican trilobites. The vicariant biogeographical patterns are well supported by a variety of tests described above (Lieberman 2003), and seem resilient and potentially serve as a sound template to consider the relationship between geological changes and evolution. Further, each of the regions considered has trilobites from all of the major biostratigraphical intervals in the trilobitic part of the early Cambrian (see Lieberman 1999, 2002, 2003), suggesting that simple sampling biases related to the available strata preserved are not a likely cause of the biogeographical patterns.

The patterns of vicariance between Laurentia, Siberia and the northern margins of western Gondwana are compatible with a distribution of trilobites across an originally unified Laurentia, Siberia, western Gondwana and peri-western Gondwana, forming the core elements of Pannotia. These trilobite faunas would have subsequently differentiated via vicariance as Pannotia broke apart. The absence of resolution in the geo-dispersal tree further suggests that trilobites probably did not disperse in significant numbers between the different parts of Pannotia (especially Laurentia and western Gondwana) after they became separated. As the breakup of Pannotia is constrained to 550–600 Ma, these palaeobiogeographical patterns suggest that divergence events in trilobites occurred during the Neoproterozoic, preceding the Cambrian radiation by some substantial period of time.

These results match the conclusions of Fortey et al. (1996, 1997), who argued that biogeographical patterns in trilobites indicated a Neoproterozoic origin for the group and suggested that lineages within the Trilobita were actually diverging in the Neoproterozoic. On the basis of the phylogenetic position of trilobites as euarthropods (e.g. Briggs et al. 1992; Wills et al. 1998) this suggests that the Cambrian radiation was well under way in the late Neoproterozoic. It also suggests that early Cambrian trilobites are a potentially useful source of information regarding latest Neoproterozoic tectonic events.

Palaeogeography

There are three end-member tectonic models discussed in this paper. Our preferred high-latitude Laurentia option (Fig. 1a and b) has palaeomagnetic support from two studies in North America (Symons & Chiasson 1991; Meert et al. 1994) and weaker support from a study of the Sept Îles Igneous Complex (Tanczyk et al. 1987). The model does require a relatively rapid transition (minimally c. 11 cm a⁻¹) from a high-latitude Laurentia at c. 565 Ma to a more equatorial position by Mid-Cambrian time (c. 508 Ma, Tapeats sandstone). If the documentation by McCausland & Hodych (1998) of a low-latitude position for the Skinner Cove volcanic rocks of western Newfoundland is correct and if this block was attached to eastern Laurentia (the relationship of this allochthonous block to the eastern Laurentian margin is contentious) then their pole, if representative of Laurentia, requires even higher drift rates (of the order of 35–50 cm a⁻¹). The advantage, however, of our preferred high-latitude Laurentia option at 580 Ma is that it results in a favourable geometry for the final assembly of Gondwana and the closure of the Mawson Sea between Australo-Antarctica and the bulk of Gondwana (see Fig. 1b compared with Fig. 2b).

The low-latitude Laurentia option circumvents all issues related to rapid plate motions, but it precludes a reasonable alternative explanation for the high-latitude results from the Callander and Catoctin studies (Pisarevsky et al. 2000;Meert & Van der Voo 2001). The primary argument for the low-latitude option for Laurentia was to maintain a close relationship to Siberia required by the model of Pelechaty (1996). However, recent publications by Vernikovsky & Vernikovskaya (2001) and Khain et al. (2003) suggest that Siberia probably rifted away from the arctic margin of Laurentia starting at around 800 Ma. Unfortunately, there are no palaeomagnetic data from Siberia to unambiguously document its relationship to Laurentia and geological comparisons are equally contentious (Meert & Torsvik 2004). Therefore, our reconstruction between Siberia and Laurentia differs slightly from that advocated by Pisarevsky et al. (2000) in that we attempt to minimize the offset between the present-day arctic margin of Laurentia and the Siberia. In essence, we attempt to harmonize the evidence for early (c. 800 Ma) rifting with the suggestion by Pelechaty (1996) and Pisarevsky et al. (2000) that Siberia and Laurentia were conjoined until c. 550 Ma.

A number of recent studies have important implications for the possibility of a low-latitude supercontinent at 580 Ma. Popov et al. (2002) presented results from Upper Vendian sediments from the Winter Coast of Baltic Russia. Their palaeomagnetic pole, if correct, would place Baltica at low latitudes in an inverted position (see Fig. 2a). Popov et al. (2002) argued that the Fen Complex pole of Meert et al. (1998) is a Permian remagnetization. However, additional palaeomagnetic data from the Lower Cambrian Dividal Group (Torsvik & Rehnström 2002; Rehnström & Torsvik 2004) along with preliminary data from the 590 Ma Alnö complex of Sweden (Walderhaug et al. 2003) lend further support to the primary nature of the Fen Complex pole. We also note that our palaeobiogeographical patterns, especially the close vicariance relationship between eastern Laurentian and Baltic trilobites, seem to argue strongly against these new palaeomagnetic data, which place Baltica at low latitudes while Laurentia was located at high latitudes.

There are other problematic issues related to the Pisarevsky et al. (2000) low-latitude position of Laurentia at 580 Ma. Most evidence suggests that the opening of the Iapetus Ocean between Laurentia and the South American blocks commenced post-580 Ma (Cawood et al. 2001). Furthermore, Fitzsimons (2000) and Boger et al. (2002) suggested that final Gondwana assembly took place around 550 Ma via the collision of Australo-Antarctica with the remainder of Gondwana. Assuming that both scenarios are correct, a rigid palaeoreconstruction would result in significant overlap between Australo-Antarctica and western Gondwana. We note, however, that by taking into account the errors in palaeomagnetic data this misfit can be alleviated. Still, such a reconstruction would require placing East Antarctica near the tip of present-day SW Laurentia, an alignment at odds with this and other palaeobiogeographical studies and also several palaeomagnetic analyses (e.g. Torsvik et al. 1996; Cawood et al. 2001).

Kirschvink et al. (1997) attempted to reconcile the apparent rapid drift required by the high-latitude Laurentia model shown in Figure 1 by proposing an inertial interchange true polar wander (IITPW) event during the interval from 523 to 508 Ma. A variation of the IITPW model was proposed by Evans (2003). The model was presented without any detailed palaeogeography, but does posit a series of Neoproterozoic–early Palaeozoic true polar wander episodes including at least one inertial interchange event. The IITPW model is somewhat more rigid than the two preceding tectonic models in that the relative palaeolongitudes shown in these figures also need to be fixed; this is unlike conventional early Palaeozoic and Precambrian palaeomagnetically based reconstructions. Assuming the Kirschvink et al. (1997) model, Figure 3a represents the configuration of the continents prior to the IITPW event. Although Kirschvink et al. (1997) claimed that this represented their 540 Ma reconstruction, in fact it was based partly on 575–565 Ma palaeomagnetic data from Laurentia and Baltica; this explains the similarities of Figure 3 to the reconstruction in Figure 1. The main difference between the pre-IITPW model of Kirschvink et al. (1997) and the one shown in Figure 1 is that in the former Baltica is placed in the northern hemisphere (by inverting the polarity of the palaeomagnetic pole) and Siberia is also placed well away from both present-day northern Africa and the arctic margin of Laurentia. The model thus requires a wide Iapetus ocean between Laurentia and Baltica prior to the Fen Complex pole at 580 Ma (Meert et al. 1998). The pre- and post-IITPW (earliest Mid-Cambrian, c. 510 Ma) palaeogeographies show significant differences (Figs 3 and 6). Kirschvink et al. (1997) showed an overlap between Laurentia and Gondwana, which they attributed to an incomplete dataset. The overlap, however, between the two continents is severe and would require more than 40° of latitudinal displacement between western Gondwana and eastern Laurentia to generate a Iapetus ocean consistent with geological data (Cawood et al. 2001). In addition, Baltica is displaced significantly from Avalonia compared with conventional views of Cambrian palaeogeography (see Cocks & Torsvik 2002).

Our preferred palaeogeographies at 540 and 510 Ma based on palaeomagnetic data are shown in Figures 5 & 7. Figure 5 is also based on data from Torsvik & Rehnström (2001) and Cawood et al. (2001); the Iapetus Ocean is near its maximum width and the Ægir Sea separates Baltica from Siberia. Our reconstruction at 510 Ma (Fig. 7) avoids the problems of continental overlap that the model of Kirschvink et al. (1997) suffers from, and is in fact rather similar to those palaeogeographies advocated by Torsvik & Rehnström (2001) and Cocks & Torsvik (2002), each of whom suggested a Iapetus Ocean of moderate width.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

 (a) Our preferred palaeogeographical reconstruction at 540 Ma (after Torsvik et al. 1996; Torsvik & Rehnström 2001) showing the opening of the Iapetus ocean between a fully united Gondwana and Laurentia. (b) rotation of (a) to show the rest of Gondwana. Abbreviations as in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

 Post-IITPW reconstruction for 510–505 Ma based on the analysis of Kirschvink et al. (1997). Abbreviations as in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

 Our preferred 510 Ma reconstruction. Abbreviations as in Figure 1.

Relationship of palaeobiogeographical patterns to palaeogeographical and tectonic models

In several respects the recovered palaeobiogeographical patterns match results from previous studies of early Cambrian palaeobiogeography. For instance, the close palaeobiogeographical relationship between Siberia and north Africa–Armorica matches a pattern found by Lieberman (1997). In addition, the relatively close relationship between Avalonia and north Africa–Armorica was also found in studies of early Cambrian trilobites conducted by Burrett & Richardson (1980) and Fortey & Cocks (1992). Burrett & Richardson (1980), Debrenne & Kruse (1986) and Palmer & Rowell (1995) also found a close palaeobiogeographical relationship between Antarctica and Australia. Finally, Fortey & Cocks (1992) identified a close palaeobiogeographical relationship between Baltica and eastern Laurentia, a result reiterated by this analysis. The polyphyly of Laurentia in Figure 4 differs from some previous palaeobiogeographical studies (e.g. Burrett & Richardson 1980; Fortey & Cocks 1992; Lieberman 1997). Part of the differences between this aspect of the results from Burrett & Richardson (1980), Lieberman (1997) and the current study may be attributable to the fact that those studies focused on a more restricted and later part of the early Cambrian. The difference between the position of Laurentia in this study and in those by Burrett & Richardson (1980) and Fortey & Cocks (1992) may be because those workers could not incorporate phylogenetic information into their palaeobiogeographical analyses. Notably, however, the polyphyly of Laurentia recovered herein agrees with some aspects of the analysis by Waggoner (1999) of Vendian Ediacaran palaeobiogeography. For example, Waggoner (1999) also found that SW Laurentian faunas grouped with those from western Gondwana whereas NW Laurentian faunas grouped with Baltic faunas.

The biogeographical patterns resulting from analyses of the trilobites can be compared instructively with the three models of late Neoproterozoic and early Cambrian tectonics and palaeogeography described above: for example, the IITPW event hypothesized by Kirschvink et al. (1997). On the basis of their reconstructions, one might predict to find close palaeobiogeographical relationships between Baltic faunas and Siberian faunas, and also potentially between Baltic faunas and Avalonian and north African–Armorican faunas. This is because all of these regions, but especially Baltica and Siberia, lie near one another both before and after the inferred IITPW event of Kirschvink et al. (1997). None of these predicted relationships, however, was retrieved by the palaeobiogeographical analysis (Fig. 4).

Further, there are other aspects of the resulting palaeobiogeographical patterns that are counter what might be predicted if the Kirschvink et al. (1997) version of true polar wander had occurred. This is because each of those studies predicted rapid movements of cratons and thus faunas; in accord with this one might predict that palaeobiogeographical patterns among most cratons should show little or no resolution. Instead this is manifestly not the case (see Fig. 4), as palaeobiogeographical patterns of vicariance are well resolved and also strongly supported. It is, of course, still conceivable that IITPW did occur, but if so it must have happened before, or after, the evolutionary and palaeobiogeographical patterns in the trilobites were produced. This minimally constricts the timing of any IITPW event and also constrains the validity of the hypothesis as a general driver and pacemaker of the Cambrian radiation (contra the arguments of Kirschvink et al. 1997), as trilobites are a key component of the Cambrian radiation fauna and appear to be unaffected, at least in a palaeobiogeographically informative manner, by such hypothesized changes. This information, in conjunction with the results from palaeomagnetism described above, casts further doubt on the validity of inferred early Cambrian or late Neoproterozoic IITPW events.

The palaeobiogeographical patterns derived from the trilobites also potentially conflict with the palaeogeographical model proposed by Pisarevsky et al. (2000), which has Laurentia in a low-latitude position (Fig. 2a and b). For example, their model would predict a much closer association between Baltic faunas and faunas from Avalonia and north Africa–Armorica than the pattern actually recovered (see Fig. 4). Further, the model of Pisarevsky et al. (2000) would also predict a closer association between Siberian faunas and faunas from eastern and NW Laurentia, and again such a palaeobiogeographical relationship was not recovered (see Fig. 4).

Overall the results from the palaeobiogeographical analysis (Fig. 4) accord well with our preferred palaeogeographies resulting from analysis of the palaeomagnetic dataset (Figs 1, 5 and 7) such that there is a strong degree of congruence between the two datasets. For example, each analysis recognizes a close association between most of East Antarctica and Australia, and these cratons probably had a common history through the Neoproterozoic and early Cambrian. Further, Baltica lies near parts of Laurentia, especially eastern Laurentia, throughout the Neoproterozoic and into the early Cambrian. The close relationship posited in the palaeobiogeographical analysis between Avalonia and north Africa–Armorica also finds support from the palaeomagnetic dataset as these regions are near one another in the late Neoproterozoic and early Cambrian. The palaeobiogeographical data also agree with aspects of the position of Siberia especially in the close palaeobiogeographical relationship between Siberia, north Africa–Armorica and Avalonia. For this reason, we suggest that the high-latitude Laurentia model for late Neoproterozoic palaeogeography (Fig. 1) and the model for early Cambrian palaeogeography with the opening of the Iapetus Ocean between present-day eastern Laurentia and western Gondwana taking place around 540 Ma (Figs 5 and 7) are best supported by the available palaeobiogeographical and palaeomagnetic data.

There is, however, some disagreement between the palaeobiogeographical results and the palaeomagnetic results, and this involves the grouping of SW Laurentian faunas with faunas from Avalonia, north Africa–Armorica and Siberia. This is because the palaeogeographical geometry predicted by palaeomagnetic studies should have led to a clustering of SW Laurentian faunas with the rest of Laurentia instead of, or more closely than, parts of western Gondwana and Siberia. The results from the palaeobiogeography and palaeoreconstructions could be compatible, however, if Amazonia served as a faunal link or bridge between SW Laurentia and other parts of western Gondwana. Unfortunately, age appropriate deposits in Amazonia are not available to test this in greater detail. Another solution that would make the one potential divergence between the palaeobiogeographical and palaeomagnetic results more compatible would be to rotate Laurentia and Baltica approximately 20° counterclockwise. This, however, is currently not permissible with the available palaeomagnetic data. It is noteworthy that the palaeogeographies from both the IITPW and low-latitude Laurentia models do not provide a better fit for this aspect of our palaeobiogeographical patterns. To address the potential source of disagreement between our palaeobiogeographical and palaeomagnetic datasets, at this time we suggest the need for more extensive faunal and palaeomagnetic sampling, especially in Siberia, north Africa, Avalonia and parts of SW Laurentia that have not been intensively sampled, for example, the Caborca region of Mexico.