Metamorphosed silty mudstones of the Burgess Shale and Stephen Shale formations record a polymetamorphic history. An early greenschist-facies event associated with burial by Paleozoic strata produced a nearly ubiquitous bedding-parallel cleavage (S1). Tectonic exhumation during the formation of the southern Canadian Rocky Mountains produced a domainal subgreenschist-facies retrograde overprint in which a high-angle crenulation cleavage (S2) was developed. Whereas all rocks have experienced these two events, the degree of deformation and fossil preservation varies with position relative to the Cathedral Escarpment. This paleosubmarine cliff resulted in a zone of reduced deformation within adjacent strata by buttressing them during burial and deflecting deformation during orogenesis. Fossil-bearing strata are composed of a typical greenschist assemblage of muscovite–chlorite–quartz–albite, are devoid of clays, and contain an average of 0.28% organic carbon. This typical metamudstone assemblage is consistent with the typical whole-rock composition of these rocks which tends to be richer in K and Al and poorer in Fe relative to the Post-Archean Average Shale. These mineralogical–compositional characteristics suggest that the premetamorphic clay assemblage was likely illite–smectite–kaolinite, with no evidence of highly reactive species such as nontronite or Na-montmorillonite. This is contrary to the required conditions for taphonomic models involving organic preservation due to clay-related suppression of decomposition-related reactions. Metamorphism of the Burgess Shale has also reduced the total organic carbon content to <20% of initial values. This must be considered in any models that involve interpretation of organic carbon in diagenetic processes (e.g., fossil formation and determination of paleoredox conditions).
Since their discovery in 1909, the fossil beds of the Burgess Shale, British Columbia, Canada (Fig. 1), have received considerable attention due to the exceptional preservation of their diverse and unusual soft-bodied fauna. The fossils lie within the Burgess Shale Formation, a thick, basinal succession composed of calcareous, silty mudstones interbedded with limestones (Fletcher and Collins 1998). Regionally, however, these strata occur as crenulated schists or phyllites, and so the Burgess Shale area must have experienced at least one significant metamorphic episode.
A significant metamorphic overprint would have affected the mineral and textural composition of both the rocks and the soft-bodied fossils within them. The puzzle of the processes that led to the preservation of the Burgess Shale fossils cannot be solved without unraveling the post- fossilization metamorphic overprint. A better understanding of the metamorphic history of the fossil beds of the Burgess Shale is required. Accordingly, this study was undertaken to determine the metamorphic history of the Burgess Shale fossil beds and the effect of metamorphism on the Burgess Shale and its fossils.
The Burgess Shale Formation abuts the basinal edge of platformal carbonate rocks of the Cathedral Formation along a near-vertical contact called the Cathedral Escarpment. This paleotopographic feature marked the cliff-like edge of a regional passive margin in the Middle Cambrian. The escarpment curves across the neighboring Kicking Horse Valley, forming the Stephen–Field Embayment (McIlreath 1977), and is interpreted to be a fossil gravity slide-scar formed by collapse of the platform front (Stewart et al. 1993).
The carbonate platform repeatedly redeveloped throughout the Middle and Late Cambrian. The basin–platform facies transition remained fixed in position during this time and is referred to as the Kicking Horse Rim (Aitken 1971). Subsidence along this paleogeographic feature was evidently controlled by deep-seated faults (Stewart et al. 1993) that formed during Proterozoic rifting and continued to be reactivated through the Cambrian.
Regionally, metapelitic rocks of both basinal and platformal facies have a phyllitic nature. Clearly, the region surrounding the Burgess Shale must have experienced low- to moderate- grade metamorphism during its history. The rocks immediately adjacent to the escarpment, however, have preserved their primary sedimentary features and exhibit little evidence of deformation in the field. All Burgess Shale fossil localities lie within the relatively undeformed rocks of the shale basin immediately adjacent to the Cathedral Escarpment (Collins et al. 1983). This has been interpreted to be a preservational relationship, with the delicate fossils preserved within the escarpment’s pressure shadow during subsequent deformational events (McIlreath and Aitken 1977).
The soft-bodied fossils are preserved as thin, silvery, multilayered films, with thin layers of rock lying between appendages and the exoskeleton (Whittington 1980). Significant discussion has focused on the composition of the silvery films. (Butterfield 1990, 1995) reported that the principal constituent of these fossils is organic carbon and in fact defines Burgess Shale type preservation as “exceptionally preserved fossils whose primary taphonomic mode is one of non-mineralizing organisms preserved as carbonaceous compressions in fully marine sediments” (Butterfield 1995). The shine in this case is attributed to coalification associated with metamorphism. In contrast, (Towe 1996) and (Orr et al. 1998) concluded that aluminum-bearing silicate minerals are the principal constituent of the fossils and that the luster of these fossils is a result of the compression and alignment of clay minerals parallel to the surface of the fossil. These authors note that organic carbon is present, but as a relatively minor constituent in both the fossils and the rocks of the Burgess Shale as a whole. The average total organic carbon (TOC) value reported by (Kelafant 1987) is 0.41%, whereas (Butterfield 1990) reported an average of 0.11% organic carbon.
Disagreement regarding the composition of the fossils correlates with disagreement regarding the mode of fossilization at the Burgess Shale. (Butterfield 1990, 1995) proposed that the clay that enveloped the fossils during burial acted as a catalyst that inhibited microbial decomposition of the tissues (i.e., adsorption of extracellular enzymes on smectites), leaving the organic materials to produce a kerogen film during diagenesis. In contrast, (Orr et al. 1998) concluded that fossilization occurred during decomposition due to chemical interactions between the tissues and the surrounding clays. In their model, variations in composition and reactivity of the various parts of the fossils during decomposition resulted in varying accumulations of clays on the carcass either by accumulation on the template of the tissues, or by direct precipitation from the pore water.
Almost certainly, some of the controversy regarding the mechanisms of Burgess Shale type preservation is due to the obscuring of primary features during post-depositional processes including metamorphism. Organic carbon content of shales decreases significantly during metamorphism, perhaps resulting in a great underestimation of the organic carbon that was present immediately after fossilization. In addition, metamorphism would have resulted in recrystallization and (or) consumption of primary clay minerals. Thus, the phyllosilicates currently associated with the fossils are not the same minerals that initially coated the buried organisms, nor do they display purely primary textures.
Sampling and analytical methods
Samples for this study were collected on Fossil Ridge, the north–south-trending ridge that connects Wapta Mountain to the north and Mount Field to the south (Fig. 1B) and hosts three distinct fossiliferous horizons that have been quarried. From the lowest to highest stratigraphic position, these quarries are referred to as the Walcott Quarry, the Raymond Quarry, and the Tuzoia Beds (Fig. 1C). Samples were collected across the southern half of Fossil Ridge, including samples from each of the three exploited fossiliferous horizons, from platformal metapelitic strata immediately overlying the Cathedral Formation, and from scattered outcrops of the Burgess Shale Formation south of the quarries (Fig. 2).
This sampling pattern was designed to obtain samples at various distances from the Cathedral Escarpment, such that the rocks collected would have experienced varying degrees of protection during deformation due to the buttressing effect of the escarpment. Given that the Cathedral Escarpment cuts obliquely through Fossil Ridge at a relatively shallow angle, however, the distance of outcrops from the Cathedral Escarpment increases southward, but the distance is significantly less than the apparent distance on the ridge face (i.e., a sample taken 1 km south of the exposure of the Cathedral Escarpment on Fossil Ridge is likely to be a true distance of less than a few hundred metres normal to the paleo-cliff face).
X-ray diffraction (XRD) methods were used to investigate the lattice ordering of sheet silicates and further characterize the mineral content of the rocks. Samples were ground initially in a ring mill and further ground when necessary using mortar and pestle. For whole-rock analysis, 1 g of powder was packed into a cavity mount. Clay sample preparation involved mixing the powder in water, disaggregating the material with ultrasonic vibration, and allowing the sediment to settle for 4 h. According to Stokes law for settling solids, the remaining suspended sediment is the clay fraction (∼2 μm). The suspension was decanted, centrifuged, and placed on glass slides. As the samples dried, the clay crystallites settled in a flat orientation, which emphasized the diagnostic basal lattices. Multiple clay-fraction slides were produced for each sample so that each sample could be scanned under the three following conditions: (i) air-dried (as described earlier in the paper), (ii) glycolated by exposing the sample to ethylene glycol atmosphere for 12 h to expand lattices, and (iii) heated to 550°C for 30–60 min to tighten lattices or render certain minerals amorphous.
Both whole-rock and clay-fraction samples were analyzed at the University of Calgary, Calgary, Alberta, using a Scintag XDS 2000 X-ray diffractometer using Cu-Kα radiation and settings of 45 kV and 40 μA. Whole-rock samples were run from 4° through 74° 2θ at a step time of 2 s and a step width of 0.1°. Clay-fraction samples were run from 2° through 28° 2θ at a step time of 2 s and a step width of 0.05°. Diffraction patterns were analyzed using Jade version 3.0 software by Materials Data Incorporated.
Whole-rock analysis (by inductively coupled plasma – mass spectrometry (ICP–MS)) and total organic carbon determinations (by coulometry) were performed by X-RAL Laboratories, Toronto, Ontario. Samples were selected such that obvious weathered surfaces and deeper weathering rinds were removed prior to processing.
Mineralogical characteristics of the Burgess Shale
Major-element analyses of samples from the quarries (Table 1) indicate that the Burgess Shale from this locality has a bulk-rock composition similar to that of the average shale (Post-Archean Australian Shale (PAAS); Taylor and McLennan 1985) with only minor amounts of sulphur and organic carbon, the latter being in agreement with the results of (Kelafant 1987). Petrographic and XRD analysis indicates that all of the Fossil Ridge argillaceous rocks examined are composed of the assemblage chlorite–muscovite–quartz–albite–calcite, with minor amounts of pyrite and graphitic carbon and trace chalcopyrite, a composition similar to that noted by (Allison and Brett 1995). No sulphates were noted in either XRD or petrographic analysis, suggesting that most, if not all, sulphur is contained within sulphide minerals. (Conway Morris (1990) noted the presence of rare barium sulphate nodules in direct association with some fossils.)
The mineralogical association of chlorite–muscovite–quartz–albite in metapelitic rocks is potentially stable from the upper subgreenschist to middle greenschist facies. Accordingly, XRD analysis of clay content and lattice ordering of illite (referred to as illite “crystallinity”) were used to further characterize the metamorphic grade of the Burgess Shale.
Kaolinite and discrete grains of smectite are absent from all samples. This is consistent with metamorphism of mudstones and siltstones at subgreenschist grade or higher (Kisch 1987). Through the subgreenschist and lower greenschist facies, there is no appearance or disappearance of obvious index minerals in relatively silty (i.e., moderately aluminous) metasedimentary rocks such as those of the Burgess Shale Formation. (Weaver 1960) noted, however, that there was a correlation between metamorphic grade and the shape of the 10 Å (1 Å = 0.1 nm) peak in the X-ray diffraction pattern of illite–muscovite. With increasing grade, as the illite crystal lattice becomes more ordered, the 10 Å peak becomes sharper. Weaver quantified the relative sharpness of the 10 Å peak by means of a ratio of the intensity H at 10 and 10.5 Å (Weaver index = H10Å/H10.5Å). The value of the Weaver index of illite crystallinity increases with an increase in grade: values <2.3 correlate to the diagenesis zone, values between 2.3 and 12.1 correlate to the anchizone, and values >12.1 correlate to the epizone (Weaver 1960; Kubler 1964; see Fig. 3 for a correlation between Kubler zones and metamorphic facies). Because of the general correlation between zones and facies, and the greater familiarity of the latter, this paper will refer to metamorphic facies from here on.
At higher grade, as the 10 Å peak becomes sharper, the error in measurement of the Weaver index becomes greater. Accordingly, this easily determined measurement of lattice ordering should not be used for exact determinations of grade in higher grade rocks. It can still be useful, however, for comparing relative grade of samples within a region, analyzing patterns across a study area, and determining generalized conditions of metamorphism. In this paper, the Weaver index is used for these general comparative purposes.
Whereas limestone units occur within both the Burgess Shale and Stephen Shale formations, only argillitic rocks were examined in this study. Some clarification on terminology of these muddy rocks is necessary. Such rocks on Fossil Ridge are described as graded, laminated, calcareous, silty mudstones (Fletcher and Collins 1998). The rocks do not generally break into thin sheets along bedding planes, but rather into thick slabs or conchoidally fractured blocks. Those that break into slabs typically have smooth surfaces with a weakly to moderately sparkly appearance, indicating significant metamorphic recrystallization of clays to micas to produce a slaty cleavage parallel to bedding. Furthermore, as will be demonstrated by the XRD data, these sedimentary strata have been metamorphosed in the greenschist facies despite the general lack of well-defined metamorphic fabrics within the fossil quarries. To more accurately reflect the metamorphic nature of these rocks, the term metamudstone will be used in this paper to describe the metamorphosed, weakly deformed, fine-grained siliciclastic rocks of the Burgess Shale and Stephen Shale formations, which lack a well- defined slaty or phyllitic cleavage.
Characteristics of metamudstone samples from Fossil Ridge are summarized in Table 2. All metamudstone samples examined in this study are essentially similar in mineralogy: silt-sized grains of quartz and albite in a finer grained micaceous and calcareous matrix. Muscovite and chlorite are both common components. Detrital muscovite is present in trace amounts in some samples. Very fine grained pyrite is a trace constituent in all rocks examined. Whereas a minority of such pyrite preserves a framboidal texture, most of the fine-grained pyrite has recrystallized into tiny euhedral cubic crystals. Such recrystallization that results in decreased surface area is typical of metamorphic recrystallization.
Fifteen of the 16 samples (except sample M) contained bedding-parallel cleavage (S1). This fabric, which is generally very subtle in outcrop or hand sample, is clearly defined in thin section. It is characterized by a weak alignment of the fine-grained matrix micas, concentrations of very fine grained opaque material into discontinuous and stylolitic bands (Fig. 4A), and flattening of clastic material (quartz–albite grains, fossil fragments). Bedding-parallel pressure solution and foliation development are amplified adjacent to rigid bodies such as fragments of brachiopod and trilobite shells (Fig. 4B). Some shell fragments in calcite-rich rocks display pressure shadows composed of bedding-parallel ribbons of quartz (Fig. 4C).
Eight of the 16 samples contained a subtly developed crenulation cleavage at a high angle to bedding (S2). It is generally defined by the folding of the bedding-parallel foliation and by discontinuous planar concentrations of very fine grained opaque material and the recrystallization of phyllosilicates in the spaced cleavage planes (Fig. 4D). This cleavage is much more apparent in rocks located slightly farther from the Cathedral Escarpment, such as those of the Mount Stephen trilobite beds, where it is clearly evident in outcrop as a spaced cleavage with a spacing of 2–3 cm.
Whole-rock XRD analysis of Burgess Shale samples indicates that the rocks are composed predominantly of quartz, muscovite, chlorite, and calcite, confirming the petrographic mineralogical analysis and in agreement with the analyses of (Allison and Brett 1995). The relative abundance of these four predominant minerals varies significantly between samples.
The Weaver index values of samples from Fossil Ridge are variable. Values range from 7 to 58, with a mean of 27 and a standard deviation of 16. There is a distinct pattern of Weaver index values relative to their position on the ridge, however (Fig. 5). The seven samples within, or adjacent to, the fossil quarries yield a tight cluster of Weaver indices (19, 22, 24, 24, 25, 27, 27), with a mean of 24 and a standard deviation of 2.6. The remaining samples (i.e., those from outside of the zone of soft-bodied preservation) exhibit a bimodal distribution of Weaver index values (7, 11, 11, 12, 13, 47, 49, 51, 58).
All moderate- to high-crystallinity samples (Weaver index ≥ 13) exhibit no variation between air-dried and glycolated runs (Fig. 6A), which indicates the absence of smectite interlayers in the white mica. All low-crystallinity samples (Weaver index ≤ 13), with the exception of sample H, exhibit a minor decrease in the basal spacing for white mica (10 Å peak) of the glycolated run relative to the air-dried run (Fig. 6B). This indicates the presence of approximately 10% smectite interlayers within illite crystals in these samples.
Interpretation of petrological characteristics
Petrographic and XRD data indicate that the metamudstones of the Burgess Shale Formation on Fossil Ridge consist of at least two populations with distinct characteristics: (i) a higher grade, low-deformation group, which include samples from both inside and outside of the fossil quarries; and (ii) a lower grade polydeformed group that occurs sporadically but only outside of the fossil quarries.
The distribution of rock characteristics is rather counterintuitive. The rocks with the least degree of deformation, those within the fossil quarries, exhibit moderate metamorphic grade. The average Weaver index for this group is 24, which is certainly correlative with greenschist-facies conditions. Rocks occurring outside of the fossil quarries are far more variable in their characteristics, but essentially fall into two clusters. One group has very high Weaver index values (∼50) and generally exhibits only a bedding-parallel cleavage (S1). The other group has low Weaver index values (∼11), indicative of upper subgreenschist facies conditions, includes minor smectite interlayers, and generally contains both a bedding-parallel cleavage (S1) and a weakly developed high-angle crenulation cleavage (S2).
Multiple variables are necessary to account for this complex pattern. Temperature is the main variable in determining the crystalline properties of illite–muscovite (Schaer and Persoz 1976) but is by no means the only one. The following additional factors can influence illite crystallinity: (i) time — the longer a grain has to recrystallize, the greater the potential crystallinity index (Essene 1982); (ii) composition — illite is K poor relative to muscovite, and so a comparative lack of K could retard the transformation of illite to muscovite; (iii) permeability increased permeability can increase crystallinity by increasing diffusion rates (Frey 1987); (iv) grain size — the presence of detrital mica and the increased permeability associated with coarser grained sediments can lead to elevated crystallinity values (Dunoyer de Segonzac 1970); and (v) deformation crystallinity is enhanced by strain (Roberts and Merriman 1985).
In this study, care was taken to include only rocks of similar composition and grain size in the comparison of illite crystallinity. Furthermore, petrographic analysis indicates that coarse detrital micas are very rare in the strata examined. Time is not a likely cause of variation in the sample set because all samples were taken within a 2 km long area and therefore must have experienced approximately the same thermal events over the same time interval. The retarding effect of a lack of K was found to be negligible in subgreenschist- and greenschist-facies rocks (Kubler 1964, 1984). Furthermore, this compositional control is not of concern in this case due to the relatively K-rich composition of the Burgess rocks.
The two remaining variables that could have potentially affected the crystallinity values of the sample set are permeability and deformation. All samples outside of the fossil quarries exhibit increased deformation in the form of foliation development, which would also result in increased permeability relative to the samples in the quarries. Accordingly, variations in illite crystallinity are interpreted to be the result of either variations in degree of deformation, or the recording of different thermal–recrystallization events.
The scenario most consistent with the pattern of illite crystallinity values is a polymetamorphic history. The first event, which was associated with the bedding-parallel cleavage, reached peak metamorphic conditions in the greenschist facies. The greater the degree of development of the early cleavage, the more ordered the white micas became, and therefore the greater the Weaver index values. The second event was a retrograde event that reached subgreenschist- facies conditions and was associated with the development of the crenulation cleavage. Any models or analytical tools that are applied to the rocks of the Burgess Shale Formation must take this metamorphic history into account.
Implications for taphonomic models
Three taphonomic models have been presented to potentially explain the preservation of labile tissues in the Burgess Shale. (Butterfield 1990, 1995) proposed a model of organic preservation involving the catalytic but nonreactive involvement of clays. In contrast, (Orr et al. 1998) proposed a model of inorganic preservation involving chemical reaction between sedimentary clays and reactive organic materials. Most recently, (Petrovich 2001) suggested that organic materials may have been preserved due to the inhibition of bacterial decomposition because of the adsorption of Fe2+ from solution onto the organic material. This adsorbed iron then reacted within the diagenetic environment to form Fe-rich clays that coated the organic remains.
Butterfield’s (1990) organic preservation model stresses the presence of incomplete organic carbon films within most nonbiomineralized fossils, particularly in cuticulated organisms. The basis of this proposed means of preservation is the potential for adsorption of organic compounds by certain clays, particularly those of the montmorillonite–smectite group, under weakly acidic conditions (pH 5–6). (Butterfield 1990) cites evidence that the activity of certain enzymes can be reduced over 90% once adsorbed onto kaolinite or Na-montmorillonite. In addition, the powerful interactive capability of smectites, particularly nontronite (Fe-rich smectite), is stressed as a potential factor in inhibiting post-burial organic decomposition (Butterfield 1995).
In terms of testing this hypothesis of clay-catalyzed organic preservation, (Butterfield 1995) states the following: “The obvious test of the ‘clay-organic’ hypothesis as an explanation for organic-walled fossil preservation is to determine the clay mineralogies of fossil-bearing localities. Unfortunately, most have been exposed to considerable alteration through diagenesis, metamorphism and (or) weathering; smectites are likely to be altered, simply as a function of age (Chamley 1989; Weaver 1989). Thus relatively little can be said about the original clay mineralogies of the chloritized Burgess Shale.” Of course, unravelling premetamorphic compositions and characteristics is one important purpose of metamorphic studies.
First, it is important to stress that regional metamorphic processes (particularly those at relatively low grade) are isochemical, except for the reduction of volatile components such as water, carbon dioxide, and organic carbon. Thus with the exception of loss on ignition (LOI), the current composition of the Burgess Shale should correspond to the primary composition of the unit, assuming that the rock has experienced neither metasomatic reactions due to the presence of chemically reactive hydrothermal fluids, nor recent surficial weathering processes.
In this sense, the term chloritized used by (Butterfield 1995) to describe the Burgess Shale is inappropriate. Chloritization is a metasomatic process associated with many ore deposits, particularly volcanogenic massive sulphides (e.g., Riverin and Hodgson 1980), and involves the removal of alkalis and silica with the addition of iron and magnesium. As can be seen from the bulk-rock compositions of the Burgess Shale, relative to average shale compositions (Table 1), there is no evidence of significant post-depositional composition change. In fact, alkalis are more abundant than in the average shale, whereas iron is generally below average, opposite to the trend expressed by chloritized rocks. The lack of metasomatic activity is also supported by the preservation of fossil material that would have been destroyed had there been substantial interaction with chemically reactive, acidic solutions.
Rather than being chloritized, the rocks of the Burgess Shale have simply experienced regional metamorphism in the greenschist facies which has caused the primary and diagenetic clay minerals to transform into new phyllosilicates such as chlorite and muscovite that are stable at higher metamorphic grade. Because this process is isochemical, the current composition (both mineralogical and chemical) can be used to infer the primary composition of the rock. Thus (Butterfield’s 1995) statement that “... little can be said about the original clay mineralogies... “ is false.
In comparing the bulk-rock compositions of low-carbonate Burgess Shale metamudstones with the average shale composition (PAAS; Taylor and McLennan 1985; Table 1), the most striking differences are the reduced SiO2 content and the elevated Al2O3 and K2O contents of the Burgess samples. This would certainly suggest a relatively high primary clay content in the Burgess Shale and below average content of detrital quartz. Note, however, that the Na2O and FeO contents are generally below average and that, although CaO content is significantly higher than that of the average shale, petrographic analysis indicates that this is due to an above average carbonate content in the metamudstones of the Burgess Shale Formation.
The low iron content of the fossiliferous strata would preclude there being any significant percentage of Fe-rich phases present in the rock now, as well as prior to metamorphism. Thus nontronite, and other Fe-rich smectites, could not have been present in any significant amount prior to metamorphism. Nor is it possible that Na-montmorillonite was present in abundance in the muds during fossilization. The low Na2O content would preclude there being a significant proportion of Na-rich minerals at any point in the history of the Burgess Shale.
Both the bulk-rock composition of fossiliferous strata and their current metamorphic mineral assemblage of muscovite–chlorite–quartz–carbonate–albite with trace amounts of pyrite and organic carbon (a typical lower greenschist facies metamudstone assemblage) indicate that there is nothing remarkable about the mineralogical character of these rocks. Thus it is highly unlikely that there was anything unusual about the initial mineralogical composition of the muds from which these rocks formed. Certainly there are no regional geological data, chemical data, or mineralogical data to support the notion of the addition of a volcanic-ash component to these rocks to produce montmorillonites or allanophane. There is no chemical or petrographic basis on which to conclude that Fe-rich hydrothermal-associated clays such as nontronite were introduced to the Burgess Shale basin in any significant amount, as suggested by (Butterfield 1995). If iron-rich clays such as nontronite or glauconite or Na-rich clays such as Na-montmorillonite were present at all, it could have only been in trace amounts.
Given the average to below average Fe content and the above average K content, it is likely that the original sediment contained only minor smectite and a significant component of illite. During diagenesis additional illite likely developed at the expense of smectite. Such reactions are well studied in modern sediments undergoing diagenesis and likely proceed by the following generalized reaction (Hower et al. 1976):
The iron and magnesium from the breakdown of the smectite is taken up by the formation of chlorite. Accordingly, the chloritic component of the modern rock can provide a gauge for the amount of original smectite in the sediment. (Allison and Brett 1995) document the composition of a representative sample of Burgess Shale to be 40.1% K-mica, 38.5% quartz, 18.8% carbonate minerals, and only 2.5% chlorite. Thus even if all of the chlorite present was derived from the breakdown of smectite (i.e., no detrital chlorite), there could only have been a minor smectite component to the original sediments.
Both major-element composition and mineral assemblage compositions in the Burgess Shale suggest a low initial smectite content. Accordingly, a taphonomic model that is dependent on an abundance of such reactive clay species does not present a likely explanation for soft-bodied preservation at the Burgess Shale.
One additional problematic aspect of the organic preservation model in relation to the significant metamorphic history is the associated reliance on low TOC values. (Butterfield 1995) notes that low TOC values are required for organic preservation so that the clay surfaces do not become saturated with adsorbed organic molecules and cites the low TOC values of the Burgess Shale (TOC ≤ 0.13%; Butterfield 1990) as support for the model. It can certainly be concluded, however, that the current organic carbon content of the metamudstones is less than the content during deposition and early diagenesis.
How much organic carbon was removed during decarboxylation and microbial methanogenesis associated with thermal maturation is uncertain. (Raiswell and Berner 1987), however, documented an exponential loss of organic carbon relative to vitrinite reflectance in post-Devonian shales. Lower greenschist facies metamorphism would correspond to a minimum vitrinite reflectance of 5% (Kisch 1987). At a vitrinite reflectance of only 3.5% (the limit of the projected data of Raiswell and Berner 1987) organic carbon content is less than 20% of its initial value. Projecting their curve to a greenschist-facies equivalent yields an estimated organic carbon preservation of 15%. The average organic carbon content of the four samples reported here (Table 1) is 0.28%, suggesting, perhaps, an approximate initial organic carbon content of > 1.5%.
Competing models for Burgess Shale type preservation as presented by (Towe 1996) and (Orr et al. 1998) also involve the injection of clays into body cavities of the organisms and replication of the outlines of the carcass when the gel-like clay stabilized. Initially, the organisms acted solely as templates for clay binding. Orr et al. noted, however, that different anatomical parts have different elemental compositions that correspond to different hydrous aluminosilicates. They concluded that these phyllosilicate films developed during early diagenesis due to an interaction between the clays that coated the organisms and the chemical characteristics of the various parts of the organisms themselves. This could have been a result of either primary compositional differences or a sequence of reactions, with the most reactive clays binding with the most labile tissues (Orr et al. 1998). Significantly, such anatomy-specific variations in composition indicate that little mobilization of major elements occurred during metamorphism, supporting the notion that bulk-rock composition is a reliable indicator of primary sediment composition for these rocks.
The model presented by (Petrovich 2001) requires free Fe2+ ions in solution, probably produced by the metabolic processes of iron-reducing bacteria. It does not require that the initial sediment be enriched in Fe-rich clays. In fact, the predicted result of the mechanism suggested by Petrovich is that Fe-rich sheet silicates would only be abundant in direct association with fossil material where Fe-rich clays grew, in part, due to the local concentration of iron that was adsorbed onto the organic material during early diagenesis. Interestingly, whereas chlorite is only a minor component of most Burgess Shale strata, (Whittington 1980) and (Conway Morris 1990) documented abundant chlorite as a major component of replaced trilobite cuticle.
The models of (Orr et al. 1998) and (Petrovich 2001) do not require special compositions or an abundance of rare mineral species within the detrital component. Accordingly, these taphonomic models are more consistent with the rather unremarkable bulk-rock composition and metamorphic mineral assemblages that characterize the Burgess Shale. The reactive clay based model of (Butterfield 1990, 1995) is inconsistent with the metamorphic characteristics of the Burgess Shale.
Implications for post-fossilization preservation
The reason for post-diagenetic fossil preservation and lack of distinct, macroscopic deformation has been attributed to localized stress variations associated with the Cathedral Escarpment (McIlreath and Aitken 1977). In this model, the limestone of the Cathedral Formation would have behaved more rigidly than either the Burgess Shale Formation or the Stephen Shale Formation during vertical compression. Thus the Cathedral Formation would have acted as a buttress that supported a wedge of Burgess Shale Formation immediately in front of the escarpment. Within this zone, directed pressure was reduced, thereby reducing development of the early horizontal fabric. Samples from this zone are likely to preserve fossils, as well as record the peak metamorphic temperature that was attained during vertical compression. Accordingly, the average Weaver index values from the quarry site (mean = 24) indicate that during vertical compression the rocks of the Burgess Shale reached temperatures correlative with the greenschist facies. This is consistent with the upper estimate of 250–280°C based on a combination of illite crystallinity and vitrinite reflectivity analyses that was noted in (Butterfield 1996) as a personal communication from I. Harding.
Outside of the pressure shadow of the Cathedral Escarpment the Burgess Shale Formation was exposed to considerable directed pressure from approximately 10 km of Paleozoic overburden. This value is based on both greenschist-facies temperatures associated with S1 development and the reconstructed Paleozoic stratigraphic section. Under these metamorphic conditions the rocks recrystallized more extensively and developed a more intense bedding-parallel cleavage. This fabric could have increased permeability during metamorphism. Both increased deformation and permeability would likely have resulted in increased crystallinity of illite–muscovite. The higher Weaver index values yielded by many rocks outside of the fossil quarries are likely related to this phenomenon of strain-induced recrystallization.
Whereas this variation in deformation associated with Paleozoic burial metamorphism could certainly account for the variation in high crystallinity values that occur across Fossil Ridge, it cannot explain the set of samples with unusually low Weaver index values. These lower grade samples are generally associated with a crenulation cleavage, and so these rocks must have experienced a younger deformational event, this one associated with Mesozoic orogenesis. During the rise of the Rockies, the Burgess Shale was tectonically exhumed from approximately 10 km below the sea floor. During this deformation event the previously metamorphosed shales underwent local retrograde metamorphism in the subgreenschist facies.
By its nature, retrograde metamorphism can only occur in permeable zones where the rocks are rehydrated. Again the zones of highest permeability would have occurred in zones of increased deformation. The zone protected by the Cathedral Escarpment during burial metamorphism was once again protected during regional metamorphism. In part, this was due to the fact that these rocks were more competent than the surrounding argillites because intense foliation development had not occurred during prior metamorphism.
Further cause for the preservation of the Burgess Shale fossils, and their host metamudstone, during regional retrograde metamorphism is evident from physical modeling studies. (Dixon et al. 1997) demonstrated that, in general, when a prograde carbonate margin is deformed due to outboard compression, a thrust ramp forms in front of the facies boundary (Fig. 7). Intense deformation within the basinal facies is transferred to the post-platformal strata along this thrust ramp, leaving a wedge of weakly deformed basinal strata immediately adjacent to the platform front. Without strain-enhanced recrystallization or a deformation-associated permeability to increase fluid–rock interaction, the illite crystallinity of rocks from the fossil quarries was unaffected by the retrograde Mesozoic event.
Outside of the undeformed wedge the argillites were potentially subjected to subgreenschist-facies retrograde metamorphism in a domainal fashion. In those domains which were affected by this late event, the rocks recrystallized at lower temperature, allowing for the development of interlayered illite–smectite and a crenulation cleavage. In those domains left unaffected by regional retrograde metamorphism, the characteristics of peak metamorphism were preserved. Certainly, this general model is applicable to the specific environment of the Burgess Shale which is located at a shale–carbonate facies transition.
There is clearly a correlation between the location of softbodied-fossil-bearing strata and proximity to the Cathedral Escarpment (Collins et al. 1983). Data presented here also demonstrate that there is a correlation between deformation–recrystallization and position relative to the Cathedral Escarpment. It has yet to be unequivocally demonstrated, however, that the present distribution of soft-bodied fossils is due to the localized sheltering of the fossils from destruction during regional deformation. Although this conclusion is often presumed, and is certainly consistent with the deformation pattern, it has not yet been proven that the distribution is not a function of primary depositional environment. Detailed mapping of fossil bed distribution relative to deformation criteria is required to interpret the pattern of fossil distribution with confidence.
Implications for interpretation of paleoenvironment
The history of organic carbon in the Burgess Shale has implications beyond questions of initial fossilization processes. It also has implications for the interpretation of the environment of deposition, particularly the interpretation of paleoredox conditions. Buried organic matter in marine shales will decompose through microbial sulphate reduction. This metabolic process, which takes place diagenetically within the sediment under anoxic conditions, consumes organic carbon compounds and produces hydrogen sulphide (e.g., Raiswell and Berner 1986). The reduced sulphur will then react with available reactive iron to form pyrite. In normal marine shales where iron availability is not limited, there is a linear trend in the organic carbon to pyritic sulphur ratio (Corg/Spy), with a slope of 2.8 and an intercept at the origin (Berner 1984). This relationship does not hold for black shales deposited under anoxic conditions where pyrite can form both during deposition and during burial. In this case the correlation (Corg/Spy) is weaker (lower slope) and has a positive intercept on the carbon axis (Berner 1984; Raiswell and Berner 1985).
The Corg/Spy method was brought to bear on the issue of paleo-oxygenation in the Burgess Basin by (Kelafant 1987). In his thesis, he reported an average reduced sulphur content of 0.35% and organic carbon values of 0.42%. Based on these low values, particularly the low reduced sulphur levels, Kelafant inferred an oxygenated environment of deposition. The pattern for Corg/Spy, however, is altered in shales that are thermally mature or metamorphosed and in which organic carbon was lost due to decarboxylation and microbial methanogenesis (Raiswell and Berner 1987). Given that the Burgess Shale must have lost considerable organic carbon during metamorphism, neither the overall content of organic carbon nor analytical methods that involve organic carbon can be used to interpret depositional and diagenetic conditions of the Burgess Shale.
The Burgess Shale experienced a burial metamorphic event due to the accumulation of Paleozoic overburden. Metamorphic conditions during this event corresponded to the greenschist facies, as determined from illite crystallinity of samples from the fossil quarries.
The region experienced a retrograde metamorphic event during Mesozoic orogenesis. Metamorphic conditions during this event corresponded to the subgreenschist facies, as determined from illite crystallinity of crenulated metamudstones from Fossil Ridge, outside of the fossil quarries.
Bulk-rock composition and petrographic analysis of soft-bodied-fossil-bearing strata indicate that the Burgess Shale essentially has, in general, an average shale composition, although it tends to be relatively rich in potassium and poor in silica. These findings suggest an illite-rich primary sediment that contained no significant proportion of Fe- or Na-rich clays. This is contrary to the conditions necessary for the organic-preservation model of fossilization as presented by (Butterfield 1990, 1995). The taphonomic models of (Orr et al. 1998) and (Petrovich 2001), however, are both consistent with the present metamorphic character of the Burgess Shale.
The rocks and fossils of the Burgess Shale were preserved from deformation that accompanied both metamorphic events. This was due to their proximity to the Cathedral Escarpment. This paleosubmarine cliff buttressed an adjacent wedge of fossil-bearing shale during vertical compression associated with burial, and deflected orogenic structures and deformation during the rise of the southern Canadian Rocky Mountains.
The current organic carbon content of the Burgess Shale is likely to be significantly less than that during fossilization due to decarboxylation during metamorphism. Initial organic carbon contents were probably more than five times the current values. Accordingly, any arguments that center on the current content of organic carbon, such as the interpretation of organic preservation of the soft tissues or the paleoredox conditions of deposition, are essentially invalid. Answers to such questions must be obtained by methods that are unaffected by metamorphic processes, or at least methods that potentially would allow one to interpret premetamorphic conditions with a degree of certainty.
The XRD study was conducted at the University of Calgary as part of the Burgess Shale High-School Research Project. During 1995–1997, approximately 75 students from Banff Community High School, Canmore Collegiate, and Golden Secondary School were involved in the project. I would like to thank each of the students who was involved in this outreach program and their teachers Magy Butterfield, Sandy Lloyd, Delores Janzen, and Gary Bjarnasson, and my coleader, Professor Charles Henderson. The Department of Geology at the University of Calgary sponsored the initial research, including both research costs and released time. This study benefitted greatly from the technical assistance of Deb Glatiotis and Mike Glatiotis (University of Calgary) and Jennie Wong (Geological Survey of Canada). An editorial review from Des Collins improved the quality of the final paper. Geochemical sample “Tuzoia” was provided by the Royal Tyrrell Museum of Paleontology (sample RTMP 99.165.57).
- Received June 19, 2002.
- Accepted November 15, 2002.
- Published on the NRC Research Press Web site at http://cjes.nrc.ca on January 17, 2003.
- © 2003 NRC Canada