The Mazon Creek Lagerstätte: a diverse late Paleozoic ecosystem entombed within siderite concretions

The late Carboniferous (Pennsylvanian) Mazon Creek Lagerstätte, Illinois, USA, is exceptional for the diversity and abundance of preserved fauna and flora found within siderite concretions. It hosts the highest diversity of Carboniferous terrestrial plants (Horowitz 1979) and exceptionally preserved 3D animals representing over 300 species from 11 phyla and 23 classes (Shabica & Hay 1997; see Table 1 and Box 1). Unlike most Lagerstätten, it includes fossil organisms that inhabited a number of closely connected habitats including terrestrial swamps, nearshore and fully marine environments. This means that the Mazon Creek provides a unique ‘window’ into multiple Carboniferous animal and plant communities.

Exceptionally preserved fossils from the Mazon river area have been known since the mid-19th century (see Nitecki 1979), but it was only after intensive strip mining for coal began in the 1940s that the importance of the Mazon Creek Lagerstätte (Mazon Creek) was truly realized. The fossiliferous concretions of the Mazon Creek occur within the Francis Creek Shale Member of the Carbondale Formation, removed as overburden to one of the largest and most profitable coal seams in the northern USA (Wright 1979). Records indicate that over 83 shaft mines and 15 strip mines operated between the 1900s and 1980s across a 250 km² geographical area in northern Illinois (Shabica & Hay 1997). Several companies, including the ‘Peabody Coal Company’ (which operated the largest and most famous strip mine, the 14 km² ‘Pit 11’) extracted coal for over 50 years, resulting in the accumulation of immense spoil heaps that were picked over for fossil-bearing siderite (iron carbonate, FeCO₃) concretions by collectors and researchers for decades. This resulted in huge collections of fossil material being amassed; it is not uncommon for collectors to own tens of thousands of fossiliferous concretions, with many hundreds of thousands of Mazon Creek fossils being donated to museums worldwide or sold privately.

Owing to the flat topography of Illinois, very few natural exposures exist; the banks of the Mazon River reveal some outcrop (Fig. 1e); however, high river discharge, vegetation overgrowth and land ownership make collecting problematic. Unfortunately for palaeontological research, the majority of the coal mines were closed and back-filled during the 1990s. Because of this, collecting localities are greatly diminished, especially the largest, Pit 11, which was deliberately flooded to create a fishing reserve and cooling lake for a nuclear power station. Nevertheless, study of the Mazon Creek is currently undergoing a revival (e.g. Sallan & Coates 2014; Clements et al. 2016; Cotroneo et al. 2016; Gabbott et al. 2016; Locatelli et al. 2016; McCoy et al. 2016; Murdock et al. 2016), predominantly utilizing the huge collections of fossiliferous material at several North American institutions such as the Field Museum of Natural History, Chicago, USA and the Royal Ontario Museum, Toronto, Canada.

The Mazon Creek is named after a tributary of the Illinois River SW of Chicago (Illinois, USA), that flows close to the town of Morris, Grundy County, Illinois, USA (Fig. 1). The fossiliferous concretions are found across an area of c. 150 km² spanning several counties; when found in situ, the concretions occur within the lower 3–8 m of the Francis Creek Shale Member.

Baird (1979) stated that the Francis Creek Shale Member was deposited at 290 Ma based on Eysinga (1975), whereas later work and reviews (i.e. Baird et al. 1985a; Shabica & Hay 1997; Schellenberg 2002) used an age of 296 Ma based on Harland et al. (1982). However, nearly all subsequent literature (i.e. Baird et al. 1985b, 1986; Sallan & Coates 2014; Clements et al. 2016; Cotroneo et al. 2016; McCoy et al. 2016, etc.) cited the Francis Creek Shale Member as having an age of 306–311 Ma. The latter age is based on studies utilizing palynological and palaeobotanical data (Pfefferkorn 1979; Wagner 1984; Peppers 1996), which indicate an age that equates to the upper part of the Moscovian stage, the top of which has been dated at 307.0 ± 0.1 Ma (Cohen et al. 2013).

Late Carboniferous global-scale geological processes were dominated by the collision of Laurussia and Gondwana and the onset of the Late Paleozoic Ice Age. Several discrete glaciations in both the southern and northern hemispheres and intermittent warming periods (Montañez & Poulsen 2013) caused sea-level oscillations during the late Carboniferous (Heckel 1986; Veevers & Powell 1987).

The Francis Creek Shale Member containing the Mazon Creek Lagerstätte was deposited on the NE margin of the Illinois Basin, which during the late Carboniferous was situated c. 4–10° south of the palaeoequator (Fig. 1) (Ziegler et al. 1979; Baird et al. 1985a, b; Rowley et al. 1985) in a vast epeiric sea that covered most of Illinois and the adjoining states (Wanless 1975; Wright 1979; Cecil et al. 2003). The unit underlying the Francis Creek Shale Member, the Colchester Coal (no. 2) Member, represents an extensive lycopod-dominated swampy forest found close to the palaeo-coastline (Pfefferkorn 1979; Phillips et al. 1985; Winston 1986) that grew in a warm, humid and wet climate (Schopf 1979a; Phillips & DiMichele 1981; Phillips et al. 1985; DiMichele 2014). Ultimately, a marine transgression inundated and drowned much of the forest adjacent to the palaeo-coast (Baird et al. 1985b). As this transgression occurred, the continuing Alleghanian orogeny to the NE disrupted atmospheric conditions and led to increased precipitation (Phillips et al. 1985; Cecil et al. 2003), which resulted in powerful erosive rivers that eventually deposited their load into newly formed shallow seas (Potter & Pryor 1961). In northern Illinois, the washout resulting from this upland flooding formed the Francis Creek Shale Member, an expansive and vertically thick member (25 m maximum thickness). The NE-flowing river delta systems created a large ‘wedge-shaped’ fan of terrigenous sediment (Baird et al. 1985b) that pinches out to the south and west (Baird 1997a). The Francis Creek Shale Member is overlain by stratigraphic units representing the continuing sea-level rise throughout the late Carboniferous (Baird 1979; Baird et al. 1985b).

The Mazon Creek Lagerstätte is characterized as an obrution deposit. Geological investigations have stated that the Francis Creek Shale Member represents a river delta system (e.g. Baird et al. 1985a, b, 1986). Subsequently, the Mazon Creek Lagerstätte has been hypothesized as resulting from a large-scale cataclysmic event such as bursting levees during storm events resulting in freshwater and sediment inundating and burying organisms rapidly (see Baird 1990; Shabica & Hay 1997; Schellenberg 2002). Other models include episodic incursion of turbid freshwater during periods of flash flooding (Baird et al. 1986), or storm surges (Johnson & Richardson 1966). Although these ideas are consistently cited in subsequent reviews, there are substantial issues with all these hypotheses.

These palaeoenvironmental interpretations are based on the entire 25 m thickness of the Francis Creek Shale Member, but the fossiliferous Mazon Creek concretions are found only within the basal 3–8 m of the Francis Creek Shale Member. Sedimentary structures in this subsection indicate rapid deposition in a calm marine environment (Fig. 2) (Baird 1979). No sedimentary structures indicating turbulent deposition are found in the fossiliferous layers, making a high-energy mortality site unlikely (e.g. Baird & Sroka 1990). Rather, the fossiliferous section of the Francis Creek Shale Member represents voluminous sedimentation of muds and silts fed by one or more distributary channels into quiet water (Baird et al. 1985b) inundating and burying the already submerged SW-sloping peat deposits. This rapidly deposited sediment smothered organism carcasses and engulfed the already drowned lycopod trees in life position, as seen at the base of Pit 11 (Shabica 1979).

Further evidence of this calmer embayment palaeoenvironment comes from highly regular cyclic laminations found throughout the fossil-bearing section of the Francis Creek Shale Member. Interpreted as flood–ebb tidal rhythmites (Kuecher 1983; Baird et al. 1985b; Kuecher et al. 1990; Feldman et al. 1993), these ‘pinstripe’ laminations record daily and twice a month neap–spring tidal cycles (Baird et al. 1985b; Kuecher et al. 1990) in both marine and freshwater areas (Fig. 2b) (Kuecher 1983; Kuecher et al. 1990), indicating that the entire Mazon Creek area was tidally influenced. It should be noted that the ebb tide band is far thicker than the flood band, suggesting a high sediment discharge rate (Baird et al. 1986). The presence of tidal laminae and rapid sedimentation rates, and the absence of high-energy sedimentary structures within the fossiliferous zone indicate that the organisms of the Mazon Creek inhabited an area around a river delta system that discharged large amounts of sediment and freshwater into a shallow, non-turbulent, brackish marine basin (Archer & Feldman 1994).

Much of the literature concerning the Mazon Creek considers the biota in terms of two separate faunas (e.g. Johnson & Richardson 1966; Baird et al. 1985a, b, 1986; Baird 1997b; Baird & Anderson 1997; Schellenberg 2002; Selden & Nudds 2012; Sallan & Coates 2014). Historically these biotas are based on communities found in distinct geographical localities: terrestrial wash-in and freshwater organisms make up the ‘Braidwood fauna’, whereas a predominantly stenohaline marine animal association is known as the ‘Essex fauna’ (Fig. 2a) (Johnson & Richardson 1966; the names come from the settlements near which the respective fuanas are supposedly most prevalent). As with the palaeoenvironmental reconstructions, the evidence does not support this.

The ‘Essex’ marine fauna is dominated by cnidarians (locally known as ‘blobs’), arthropods (Schram 1974, 1981; Schram et al. 1997), polychaetes (Thompson 1979), lobopodians (Haug et al. 2012; Murdock et al. 2016), chitons (Yochelson & Richardson 1979), holothurians (Sroka 1988), fishes (agnathans, acanthodians, dipnoans, sarcopterygii and actinopterygii; Bardack 1979), chondrichthyans (Zangerl 1979), lamprey (Bardack & Zangerl 1968), hagfish (Bardack 1991), cephalopods (Kluessendorf & Doyle 2000; Mapes et al. 2007) and the famous Tullimonstrum (Richardson 1966; see Fig. 3). The Essex fauna (Box 1), although clearly marine, is not typical of a Pennsylvanian offshore shelf community (Baird 1997b); it is devoid of calcareous algae, sponges, corals, articulated brachiopods, crinoids and trilobites (Baird et al. 1985b, 1986); marine organisms sensitive to salinity and high sedimentary input, which would have struggled to cope close to a palaeo-river mouth (Baird 1997b).

The ‘Braidwood’ biota is dominated by flora including horsetails, lycopods and pteridosperm tree fern material (Darrah 1969; Peppers & Pfefferkorn 1970). The vast diversity of plants has been attributed to the washout of plant material from the entire drainage basin of the rivers (DiMichele 2014) and this has resulted in a grossly inflated estimation of the diversity of Pennsylvanian wetland landscapes (DiMichele 2014; Moore et al. 2014). Animal fossils are significantly rarer, including reptiles and litter-dwelling invertebrates such as myriapods (Mundel 1979; Hannibal 2000), arachnids (Dunlop 1995), scorpions (Tetlie & Dunlop 2008) and insects, including some of the earliest known cockroach-like animals (Carpenter 1943; Burnham 1983). These plants and organisms represent allochthonous wash-out from their upstream terrestrial ecosystems. In fact, only freshwater bivalves and amphibians are considered autochthonous representatives of a freshwater biota (Milner 1982). However, poor preservation of the bivalves makes identification difficult; furthermore, their occurrence with known marine organisms (i.e. polychaetes) demonstrates that they are not good freshwater indicators (Baird et al. 1985b; Schultze 2009). Moreover, amphibians are not good salinity indicators either; some Devonian amphibians are known to be marine (George & Blieck 2011) and Carboniferous forms had the capability to disperse through coastal marine environments and may have spawned there (Godfrey 1992; Parker & Webb 2008).

Currently, there is no clear evidence, fossil or sedimentological, to support that a distinct truly freshwater ecosystem extended into the bay. Investigations have not identified any distinct geographical demarcation between freshwater and marine organisms (Baird 1997a; Schultze 2009) nor have any sedimentary structures that differentiate these environments been identified (Schultze 2009); in fact, tidal laminae have been described throughout ‘Braidwood’ areas (Kuecher et al. 1990; Baird 1997c). Previous work on the ‘one-directional mixing’ between the Braidwood and the Essex faunas is based not on animal fossils, but on fossilized terrestrial plant remains decreasing in size, abundance and diversity distally from the palaeo-coastline (Baird et al. 1985b, 1986). Most of the major groups found in the Mazon Creek could tolerate varying degrees of salinity; these groups include palaeoniscoid fish (Schultze & Bardack 1987), horseshoe crabs (Fisher 1979; Anderson 1994), syncarid shrimps, polychaetes, eurypterids (Kjellesvig-Waering 1963), ostracods and shark egg cases (Palaoxyris). Stenohaline organisms have been found within the ‘Braidwood’ fauna (Schultze 2009), further diminishing the argument for a distinct freshwater biota. Therefore, rather than a distinct freshwater biota, it is more likely that the Mazon Creek represents a brackish marine environment.

Areas that have previously been identified as freshwater (i.e. Braidwood) are actually part of the bay closer to the palaeo-coast (Schopf 1979b). This area was most probably influenced by the freshwater discharge of nearby rivers, with the discharge, sedimentary load and tidal influence acting as biogeographical filters, explaining why benthic organisms in the northern Mazon Creek have ‘patchy’ distributions occurring in dense concentrations locally (Baird et al. 1985b; Baird & Anderson 1997). Where large river systems enter the sea, the freshwater often exists as a hypopycnal plume overlying marine water. In this situation, marine faunas, both benthic and nektonic, can rapidly colonize and inhabit areas with normal marine salinity that occur beneath freshwater (Moura et al. 2016). These areas tend to exhibit lower biodiversity than ‘true’ marine environments (e.g. Carriker 1967) and this reduced diversity has been noted in northern Mazon Creek fossil assemblages (Baird 1997b). Pit 11, famous for its ‘true’ marine fossils, is bioturbated by benthic invertebrates such as polychaetes and holothurians, but only in the northern section of the pit. Only nektonic organisms, including fish, shrimp and Tullimonstrum (Box 2), are found in the southern end, which exhibits very little bioturbation and is deemed to be closer to the palaeo-coast (Baird et al. 1985b). Similar patterns of distributions have been observed in modern marginal marine environments (Moura et al. 2016). The brackish and sediment-rich water of the Mazon Creek would also explain why typical Carboniferous marine communities of corals, articulate brachiopods, bryozoans, trilobites and nuculid bivalves are found only in areas distal to the palaeo-coast, where the Francis Creek Shale Member pinches out, and are not preserved in concretions (see fig. 1a (iv) of Baird et al. 1986).

Although fossil-bearing sideritic concretions are known from several late Carboniferous Lagerstätten, such as Sosnowiec, Poland (Krawczynski et al. 1997; Pacyna & Zdebska 2002), Montceau-les-Mines, France (Perrier & Charbonnier 2014), and Coseley (Wilson & Almond 2001; Garwood et al. 2009), Crock Hey (Prokop et al. 2006; Garwood & Dunlop 2011) and Bickershaw, UK (Anderson et al. 1997), the Mazon Creek is distinctive, not only for the high diversity of organisms, but also because the fidelity of highly labile soft tissues preserved is unparalleled in the other Carboniferous concretionary fossil deposits.

The controls on concretion formation (Box 3) and the taphonomic pathways before, during and after concretion growth are under renewed investigation. Rapid and constant terrigenous sedimentary input from the upland peaty forests created the ideal scenario to bury organic matter and carcasses quickly, preventing scavenging and retarding aerobic decay. It is important to note that the kill mechanism that caused the mortality of Mazon Creek organisms is unclear, although pulses of freshwater, water column anoxia or water poisoning could be conceivable stresses on marine life. The burial of organisms obliquely to bedding planes suggests that some organisms were, however, smothered and buried in situ. The high fluxes of terrigenous sediment supply also had the secondary effect of supersaturating porewaters with sufficient interstitial iron for siderite formation. In normal marine conditions, pyrite is the primary reduced-iron precipitate as it is thermodynamically and kinetically favourable for iron to form sulphide minerals at the expense of carbonate precipitation (Cotroneo et al. 2016). However, in environments lacking sulphate, such as freshwater, or marine systems where sulphate is exhausted, iron carbonate (siderite) will preferentially precipitate (see Cotroneo et al. 2016, and references therein).

Although the presence of marine communities and the occurrence of tidally influenced sedimentary structures indicate that a persistent head of freshwater did not exist above the sediment (Kuecher et al. 1990), the Francis Creek Shale Member is dominated by illite–chlorites, which reflect deposition in low-salinity oxygen-rich water (Hughes 1970). Rapid deposition, tidal pumping and limited bioturbation probably limited the penetration of seawater into the sediment. This scenario would restrict sulphate supply (sulphate poisons the precipitation of siderite) and minimize the diffusion of oxygen, creating a reducing environment ideal for siderite formation. Furthermore, waters from upland forests would have decreased Eh and pH, further enhancing Fe⁺ activity, increasing the likelihood of siderite formation (Woodland & Stenstrom 1979), although oxygen isotopes suggest that the concretions formed in marine waters (Cotroneo et al. 2016). Although oxygen levels in the Mazon Creek are poorly constrained, it is likely that the muds were anoxic: studies on modern estuaries demonstrate that high bacterial and detrital content of the sediment can generate varying, but significant, levels of anoxia (Jonas 1997).

Decay of soft tissues was probably retarded by inhibition of both oxidant diffusion and bacterial activity by rapid authigenic precipitation of ‘proto-concretions’ around organic-rich remains (Fig. 2c). Oxygen, carbon and sulphur isotopic signals indicate that the closed environmental conditions were depleted of sulphate, swiftly changing from the precipitation of pyrite to siderite and forcing methanogenesis to become the dominant metabolic pathway for anaerobic bacteria (Cotroneo et al. 2016). Subsequently, methane fermentation is posited to have cemented the proto-concretion by supplying the carbonate required for diagenetic siderite to form (Cotroneo et al. 2016). Siderite tends to form pervasively within the sediment rather than concentrically, filling pore space and resisting compression, protecting the fossil from compaction or obliteration (Raiswell & Fisher 2000). Post-lithification, minerals including kaolinite, galena and sphalerite fill cracks and overgrow fossils, although it is not known if this happens during late diagenesis or as a result of weathering. The high carbonate content, isotopic stability and fossilized soft tissues establish that the concretions formed rapidly (Woodland & Stenstrom 1979; Baird 1990; Cotroneo et al. 2016). It has been suggested that tidal pumping may have been important to the rapid growth of the concretions (Allison & Pye 1994).

What initiated the formation of a concretion is not fully understood. The close relationship between concretion size and shape and the fossil within suggests that organic remains acted as passive nuclei (Baird 1997c). Large amounts of macerated, compressed and carbonized plant matter and infrequent fragments of insect or crustacean cuticle are found in the Francis Creek Shale Member along bedding planes (Baird et al. 1986), suggesting that the sediment was saturated with organic remains. It has been suggested that the decay of larger ‘juicer’ organic remains may have created geochemical microenvironments that triggered siderite precipitation (Baird et al. 1986; Baird 1997c), but the preservation of small animals and plant fragments brings this into question. Moreover, many concretions in the Mazon Creek are ‘dud’ (barren) and seemingly devoid of fossil material.

Such barren concretions may have nucleated around detrital organic remains or small carcasses, and the lack of organic carbon may have limited the bacterial respiration required for the lithification of the proto-concretion (Baird et al. 1986; Cotroneo et al. 2016). Potential retardation of the formation of a proto-concretion and prolonged aerobic bacterial activity prior to entombment in siderite would allow organic remains to decay before preservation. The formation of barren concretions is likely to have been exacerbated by the slower deposition and higher bioturbation commonly seen distally from the sedimentary source. This hypothesis is supported by analysis of fossiliferous versus barren concretions; in the areas closer to the palaeo-coastline 40% of concretions are barren (n = 56 974), whereas in the marine section 63% are barren (n = 229 979) (Baird & Anderson 1997).

Distance from the palaeo-coast also considerably affected the preservational fidelity of organisms (see Fig. 2d). Concretions found close to the palaeo-coast mouth are rich in soft-tissue fossils (Baird et al. 1985b, 1986; Baird 1997b) whereas areas up to 50 km west of this are characterized by concretions that preserve only small bivalve shells or are barren (Baird 1997a). Towards the towns of Galesburg and Peoria, 170 km SW of the Mazon River, the Francis Creek Shale Member thins, and comprises a highly bioturbated and non-concretionary marine siltstone containing a more typical Carboniferous marine shelf fossil fauna, including trilobites, bryozoans, brachiopods and bivalves (Smith et al. 1970; Shabica 1979; Baird et al. 1985b; Baird 1997b; Baird & Anderson 1997). Between the concretion zone and this ‘normal’ non-concretionary deposit, the Francis Creek Shale Member contains pyritic/calcite concretions almost completely devoid of any fossils, except for a few poorly preserved marine organisms (see fig. 1a of Baird 1979). These areas are typified by high levels of bioturbation that obliterated the tidal laminae that probably slowed the formation of concretions either by aerating or allowing sulphate-rich marine waters into the sediment, or both (Baird et al. 1986). Coupled with the slower rates of sedimentation around the outer margins of the delta fan, this may explain the taphonomic variation (termed a ‘taphonomic discontinuity’ by Baird et al. 1986) observed within the Francis Creek Shale Member.

Recent isotope analysis of Mazon Creek concretions has yielded some interesting preliminary results. Concretions found in the Braidwood area tend to have very little sulphide mineral content compared with Essex concretions, suggesting different diagenetic pathways (Cotroneo et al. 2016). This may be due to proximity to the palaeo-coast; however, further geochemical investigation with a rigorous sampling strategy with more than one Braidwood site is required to accurately constrain this.

The Mazon Creek represents a marine bay where allochthonous terrestrial and native marine organisms were buried in an environment favourable for siderite concretion formation. Rapid entombment of the flora and fauna led to exquisite soft-tissue preservation, fossilizing entire organisms rarely seen in the fossil record such as polychaetes, chitons and holothurians. In areas distal to the sedimentary source, burial was slower and increased bioturbation retarded the formation of siderite, preventing exceptional preservation and creating a taphonomic bias. In the most distal areas, where sedimentary input was extremely low, ‘typical’ offshore shelf communities flourished but were not preserved in concretions, nor are any fossils with soft tissues found in these areas.

The Mazon Creek is arguably the most complete record of a late Carboniferous equatorial ecosystem, and renewed research on familiar and in some cases commonly found fossils is providing new insights. For example, recent work on the Tully Monster (Tullimonstrum) has shown the benefits of using modern analytical techniques for detailed anatomical re-evaluation of taxa that have been largely neglected since their original descriptions. The current revival of Mazon Creek research has highlighted several gaps in our current understanding of how this marvellous fossil deposit formed. Investigation of modern estuarine environments has elucidated some of the controls on the formation of siderite concretions, which can now be applied to the Mazon Creek. This work, in particular, will lead to a better understanding of the unique mode of preservation seen in this Lagerstätte when compared with other Carboniferous sideritic concretionary deposits. Additional high-resolution geochemical investigations through the interval and across the geographical area have yet to be attempted; these would complement and validate the isotope studies that have recently been reported. In-depth geochemical investigations on the shale host rock may also better constrain the depositional environment and influence of freshwater on the Mazon Creek. Investigation of analogous deltaic settings would also benefit Mazon Creek research.

Although these large complex research questions will require considerable integrated research to answer, there is scope for citizen science to have a significant impact on the current understanding of the Mazon Creek. There are thousands upon thousands of unopened and unsorted concretions at many major North American museums and new material is still being collected from the last few exposures by collectors and amateur organizations. The sheer amount of fossil material available and the fidelity of soft-tissue preservation makes the Mazon Creek an ideal Lagerstätte for large-scale taxon-specific investigations, many of which are under way. These studies, along with descriptions of the many undescribed taxa, mean that the Mazon Creek is a vital Lagerstätte for reconstructing Carboniferous ecosystems.

Staff at The Lapworth Museum of Geology, Oxford University Museum of Natural History, Burpee Museum of Natural History, Field Museum of Natural History (FMNH) and Royal Ontario Museum (ROM) are thanked for assistance with this project. The Leicester University Palaeobiology, Palaeoenvironments & Palaeoclimates Research Group is thanked for inputs and advice. A. Clements, E. Dunne, P. Orr and M. Williams are thanked for proofreading and helpful comments. Thanks go to R. Garwood for his constructive review, which greatly added to this paper.

Scientific editing by Philip Donoghue