Volume 14, Issue 1 p. 86-101
Regular Article
Free Access

Reconstructing the environmental conditions around the Silurian Ireviken Event using the carbon isotope composition of bulk and palynomorph organic matter

Thijs R. A. Vandenbroucke

Corresponding Author

Thijs R. A. Vandenbroucke

Université Lille 1 - Sciences et Technologies, UMR 8217 du CNRS: Géosystèmes, Avenue Paul Langevin – bât. SN5, 59655 Villeneuve d'Ascq cedex, France

Research Unit Palaeontology, Department of Geology, WE13-S8, Ghent University, 9000 Ghent, Belgium

T. R. A. Vandenbroucke, Université Lille 1 - Sciences et Technologies, UMR 8217 du CNRS: Géosystèmes, Avenue Paul Langevin – bât. SN5, 59655 Villeneuve d'Ascq cedex, France. [email protected]Search for more papers by this author
Axel Munnecke

Axel Munnecke

Universität Erlangen-Nürnberg, GeoZentrum Nordbayern, Fachgruppe Paläoumwelt, Loewenichstr. 28, D-91054 Erlangen, Germany

Search for more papers by this author
Melanie J. Leng

Melanie J. Leng

Department of Geology, University of Leicester, Leicester, LE1 7RH UK

NERC Isotope Geosciences Laboratory, British Geological Survey, Nottingham, NG12 5GG UK

Search for more papers by this author
Torsten Bickert

Torsten Bickert

MARUM, Center for Marine Environmental Sciences, Universität Bremen, D-28359 Bremen, Germany

Search for more papers by this author
Olle Hints

Olle Hints

Institute of Geology, Tallinn University of Technology, Ehitajate tee 5, 19086 Tallinn, Estonia

Search for more papers by this author
David Gelsthorpe

David Gelsthorpe

The Manchester Museum, University of Manchester, Oxford Road, Manchester, M13 9PL UK

Search for more papers by this author
Georg Maier

Georg Maier

Universität Erlangen-Nürnberg, GeoZentrum Nordbayern, Fachgruppe Paläoumwelt, Loewenichstr. 28, D-91054 Erlangen, Germany

Search for more papers by this author
Thomas Servais

Thomas Servais

Université Lille 1 - Sciences et Technologies, UMR 8217 du CNRS: Géosystèmes, Avenue Paul Langevin – bât. SN5, 59655 Villeneuve d'Ascq cedex, France

Search for more papers by this author
First published: 31 January 2013
Citations: 22

Abstract

The carbon isotope composition (δ13C) of bulk organic matter and two palynomorph groups (scolecodonts and chitinozoans) from the Llandovery-Wenlock strata of Gotland (E Sweden) are compared to gain knowledge about carbon cycling in the Silurian (sub)tropical shelf environment. The δ13C values of the palynomorphs are mostly lower than the δ13C values of the bulk organic matter, and the δ13C values of the benthic scolecodonts are lower than those of the planktonic chitinozoans. While the difference between bulk and palynomorph δ13C may be in part a function of trophic state, the lower values of the scolecodonts relative to those of chitinozoans, which are assumed to live in the well-mixed water column, might imply an infaunal mode of life for the polychaetes that carried the scolecodonts. Lower δ13C for the scolecodonts in the middle of the section may represent variations in primary marine productivity (supported by acritarch abundance data), oxidation of organic matter in the bottom waters, or genera effects. In general, however, trends between the three data sets are parallel, indicating similarities in the low frequency, environmentally forced controls. The δ13C data show a decreasing trend from the base of the section, up to a horizon well below the base of the Upper Visby Formation. At this level, and therefore probably several 10 kyr before the δ13C increase in the carbonates, the δ13C organic values increase by ~1‰. This perhaps is an expression of a changed composition of the bulk organic matter associated with the extinction events prior to the Llandovery-Wenlock boundary.

Key Points

  • In Lusklint1, d13Cwhole rock is consistently higher than d13Cchit and d13Cscol
  • There is a 1 permil drop in d13Corg 2 m below the base the Upper Visby Formation
  • Palynomorph d13C suggest increased primary productivity prior to Ireviken event

1 Introduction

Based on records of the carbon isotope composition of marine carbonates (δ13Ccarb), the Late Ordovician and the Silurian represent one of the most dynamic intervals in the Phanerozoic. It is characterized by several global, short-lived, positive δ13Ccarb excursions with amplitudes exceeding +5‰, and maxima of over +10‰ [Munnecke et al., 2003, 2010; Bergström et al., 2009; Cramer et al., 2011]. New radiometric ages constrain the duration of some of these excursions to notably less than 1 Myr [Cramer et al., 2012]. The positive excursions are considered to be records of rapid and large-scale changes in the ancient Earth's global carbon cycle, oceanography, and climate system; however, currently there is no consensus interpretation that integrates cause-and-effect scenarios with the totality of the stratigraphic data [Jeppsson et al., 1995; Bickert et al., 1997; Cramer and Saltzman, 2005, 2007a; Noble et al., 2012]. Bulk organic matter associated with the carbonates often has a carbon isotope composition (δ13Corg) that is covariant with that of δ13Ccarb [e.g., Cramer and Saltzman, 2007b; Young et al., 2008; Noble et al., 2012]. However, the interpretation of δ13Corg can be complex because of the various factors that influence the values, including varying sources of organic matter (e.g., bacteria, phytoplankton, zooplankton), varying production rates of the organic material, growth rates and geometry of phytoplankton cells, and the potential presence of carbon concentrating mechanisms, properties of the ambient sea water (temperature, pH, pCO2, light, nutrients), preservation and diagenesis, the inorganic C pool, and the analytical (e.g., acid digestion) method [e.g., Hinga et al., 1994; Popp et al., 1997, 1998; Laws et al., 2002; Brodie et al., 2011a, 2011b; Könitzer et al., 2012]. Despite these difficulties, Early Paleozoic global δ13Corg events are widely interpreted in terms of global climate variation. Here we focus on the variety of sources that could constitute the bulk organic carbon that is measured for δ13Corg in rocks of Llandovery-Wenlock age. Lécuyer and Paris [1997] tested the contributions of several palynomorph types to the δ13C composition of Paleozoic organic matter. They indicated variations of up to 4‰ among marine palynomorphs (in a range of Silurian-Devonian marine sediment samples), and up to 2.5‰ within a single sediment sample. Where there is evidence for a significant contribution of terrestrial derived organic matter, they showed a greater range in δ13Corg. Other studies have also shown the potential for isotope bias caused by the presence of noncontemporaneous carbon (i.e., reworked geological carbon) and of allochthonous components (e.g., land and fresh water derived plant remains in marine deposits), and the origin of amorphous organic matter is still debated [Paris et al., 2008]. It is clear that it is important to know about the composition of the different organic components that contribute to the “bulk organic matter” of marine sediments that form sedimentary rocks if δ13Corg is to be used to make inferences about environmental change. Other studies have realized the potential flaw in the bulk organic carbon approach. For example, isotope variation of up to 4‰ in the Late Ordovician Guttenberg Carbon Isotope Excursion (GICE) from Iowa was interpreted in terms of wide-spread environmental change and perturbations of the carbon cycle, linked to organic carbon burial and pCO2 (ascribed to the GICE in other areas), as well as a partial local shift from normal-marine to Gloeocapsomorpha prisca-dominated organic matter [Pancost et al., 1999]. The GICE excursion in Iowa exhibited values more than 4‰ more positive than in other places in the U.S. (e.g., Pennsylvania), and the authors (also see recommendations by Popp et al. [1997]) suggested that one way to decipher local from global effects on δ13C is by compound-specific (i.e., molecules or groups of molecules) or component-specific (i.e., in our case, on isolated specimens of a particular organic-walled fossil group) isotope analyses.

A component-specific study was undertaken by Snelling et al. [2011]. They analyzed δ13C of graptolite (Ordovician-Silurian macro-zooplankton) periderm in comparison to δ13Corg of the enclosing whole rock, and noted that the difference between both signals is inconsistent over the Ordovician-Silurian, but also that the δ13C of the Aeronian (Early Silurian) graptolites from Wales and Scotland are “in general” more positive than the values of whole rock samples. This was also described by LaPorte et al. [2009] in the U.S., where they suggested that graptolites occupied a higher trophic level than algae, which were assumed to make up the bulk of the organic matter in the samples. Lo Duca and Pratt [2002] noted that algae taxa co-occurring with the graptolites were 1–2‰ more negative than associated graptolites. Other interpretations for this offset between selected fossil and bulk rock δ13Corg data include differences in time represented by the graptolite periderm and by the whole rock sample (the latter integrating a much longer time period), and selective diagenetic modification between the different organic matter types. Snelling et al. [2011], based on their Aeronian data set, concluded that for chemostratigraphic purposes whole rock δ13C will likely produce more reproducible results than graptolite δ13C, because of the small scale inhomogeneities in the graptolite signals. However, for those sections prone to nonmarine input, or time intervals of greater riverine discharge, they suggest that graptolite carbon may be a better proxy for marine bulk organic matter, despite heterogeneities.

Here we present δ13Corg data from the Llandovery-Wenlock boundary interval in Gotland and δ13C data from handpicked, organic-walled microfossils, chitinozoans (egg cases of marine zooplankton) and scolecodonts (jaws of benthic polychaete annelids). Chitinozoans, and likely their parent-animals, were part of the epizooplankton, lived in the shallow “mixed layer” of the Ordovician-Silurian oceans, and thus occupied a habitat comparable to that of the graptolites [Vandenbroucke et al., 2009, 2010a, 2010b]. Chitinozoan animals are now widely accepted as metazoans (i.e., nonautothrophes), based on solid morphological analyses of chitinozoan clusters [Paris and Nõlvak, 1999]. Chitinozoan animals and graptolites have been hypothesized to occupy a higher trophic level than primary producers such as acritarchs, mainly based on inferred modern analogues for acritarchs and parallel biodiversity trends between these groups [Servais et al., 2008]. Scolecodont-bearing worms, in contrast, are known to have been benthic animals living close to the water-sediment interface [Eriksson et al., 2004]. They are (at least) secondary consumers, based on their advanced feeding/predatory apparatus. The aim is to better understand the contribution of the various fossil groups to the bulk organic signal, and hence the variation within the isotope signals.

2 Geological Setting and Sampling

The Silurian strata, which outcrop on the island of Gotland, are the erosional remains of a low-latitude carbonate platform along the margins of the Baltic Basin. The platform rocks show little diagenetic alteration or tectonic disturbance. Witness to this excellent state of preservation are: (1) the very low Conodont Alteration Index [Jeppsson, 1983], (2) the exceptional preservation of a variety of calcareous microfossils and nannofossils such as “calcispheres” [Munnecke et al., 2000; Munnecke and Servais, 2008], and (3) the abundance of excellently preserved organic-walled microfossils (“palynomorphs”) such as full relief chitinozoans, almost unique in the world [Laufeld, 1974; Munnecke and Servais, 1996]. On the Correia [1967] scale they represent stage N1, suggesting they have never been heated above 100 °C [Laufeld, 1974]. Snelling et al. [2011] detected no significant difference between δ13C of organic graptolite periderm from low- and high-anchizone-grade localities, which suggests that any low-grade metamorphism suffered by the Gotland succession would have had negligible influence on δ13C.

The waters on the Gotland shelf were well-mixed, and the depositional depth for our samples was not deeper than a few tens of meters. This is based on δ18O values from brachiopods from shallow platform environments that are systematically only about 0.5 to 1‰ lower than values from coeval deeper shelf environments [Samtleben et al., 1996], indicating ~2 to 4 °C higher water temperatures. However, no difference was observed with respect to the δ13Ccarb values. In addition, fossil, photosynthetic, encrusting algae are recovered from the Lower and Upper Visby formations, which also yield the data of this study [Nitecki and Spjeldnaes, 1993; Calner et al., 2004]. Moreover, the lithological and faunal changes from the Lower to the Upper Visby formations near the top of the section through into the overlying Högklint Formation (not sampled), reflect a continuous development of a gradually shallower depositional depth. The reconstructed sea level for this interval is controversial; whereas Calner et al. [2004] and Cramer and Saltzman [2005] suggest sea-level rise and highstand (based on sequence stratigraphic arguments, and the development of keep-up reefs in the Högklint Fm), Brand et al. [2006] and Lehnert et al. [2010] proposed a sea-level drop and (glacial) low stand based on oxygen isotope data. Understanding the depositional environment has its importance. These well-mixed, shallow waters a priori prohibit the use of a declining δ13C gradient through the water column [Kroopnick et al., 1972] to explain any δ13C differences between fossils of organisms living at different water depths. Such mechanisms would have required a much deeper environment than that of the Silurian Gotland platform.

A 15.5 m long trench in the Lusklint 1 section on Gotland was cleaned and sampled with an average resolution of 10 cm (Figure 1). The section spans the regularly alternating limestone and marls of the Lower Visby Formation (marls dominating) and the more irregularly alternating limestones and marls of the Upper Visby Formation (limestones dominating). The Phaulactis layer (a ~20 cm thick layer with abundant large rugose corals) marks the boundary between the Lower and Upper Visby formations [Samtleben et al., 1996], and distinctive bentonite beds allowed us to position the samples in the stratigraphic succession [Jeppsson et al., 2005]. The sampled Lusklint 1 section incorporates most of the famous Ireviken Event (Figures 1 and 2). The extinction event affected many groups of organisms, but perhaps most impacted were the conodonts, with over 80% of species loss and as many as eight discrete conodont extinction datums recognized [Jeppsson, 1987; Jeppsson et al., 2005]. The first four of these precede the Ireviken oxygen and carbon (carbonate) isotope excursion by a few meters in the rock record [Munnecke et al., 2003], representing a time interval of several tens of kiloyears [Cramer et al., 2010, 2012].

Details are in the caption following the image
(A) Gotland chrono- and lithostratigraphy, and generalized δ13Ccarb curve, after Calner et al. [2005] and Cramer et al. [2011]. Numerical ages from Cohen et al. [2012]. (B) Geological map of the northern part of the Island of Gotland, after Calner et al. [2005], including the location of the Lusklint 1 section. (C) Sketch profile of the Lusklint cliff, detailed lithostratigraphy and compilation of Gotland brachiopod δ13Ccarb data, after Munnecke et al. [2003].
Details are in the caption following the image
δ13Corg from bulk organic matter (“whole rock”), separated according to the lithology of the sample (“limestones” versus “marls”) alongside δ13C results from the isolated palynomorph groups “chitinozoans” and “scolecodonts” in the Lusklint 1 section. Other events are from the literature: onset of δ13Ccarb excursion [Munnecke et al., 2003; Cramer et al., 2010], start of conodont extinction [Jeppsson, 1987; Munnecke et al., 2003], main extinction of acritarch species [Gelsthorpe, 2004]. The bentonite beds are reference levels in the section and allow precise dating [e.g., Cramer et al., 2012, for the Ireviken bentonite].

3 Methods

Samples for bulk organic matter analysis were cleaned mechanically. The weathered crust was removed by a saw; the remaining part was powdered with an agate ball mill. Two grams of each powder were reacted with 40 mL of 10% HCl to remove CaCO3. The residue was treated with distilled water to wash out remaining chloride ions, dried at 25 °C, and pestled again to achieve a homogenous powder.

A series of rock subsamples, 50 to 200 g each, were selected for extraction of organic-walled palynomorphs at the Institute of Geology of the Tallinn University of Technology (Estonia), using standard palynological techniques. This involved mechanical cleaning of the rock surface, crushing of the rock samples into pieces of ~1 cm3, followed by acid digestion using 10% HCl at room temperature (~25 °C) until all carbonates were dissolved (usually, this is followed by an HF treatment, although this was not necessary in this case, due to the virtual absence of silicates in the samples). The organic residue is washed with distilled water until neutral, and sieved at 45 µm. For each of these samples, a few tens to a hundred specimens of scolecodonts (depending on size) were handpicked from the organic residue (fraction > 45 µm) using glass micropipettes, washed repeatedly in distilled water, and dried. The bulk of the specimens represented the families Paulinitidae, Polychaetaspidae and Mochtyellidae. Jawed polychaete faunas across the Llandovery-Wenlock boundary in Gotland are discussed in detail by Eriksson [2006]. The same organic residue was then handpicked for chitinozoans at the University of Ghent (Belgium). These are smaller (typically around 100–200 µm, 10–3 to 10–4 mg, 60 wt % carbon), thus 1000 to 1500 specimens per sample were isolated using glass micropipettes and washed, thereby avoiding contamination by other organic matter in the residues. The chitinozoan faunas predominantly contain a number of different species of the genera Conochitina, Ancyrochitina, and Angochitina, in various relative concentrations. A detailed appraisal of the chitinozoan assemblages within the Upper and Lower Visby formations can be found in Laufeld [1974]. The isotope measurements were carried out in the same way as for the bulk organic material (see above).

The acid processing for both groups of palynomorphs (chitinozoans and scolecodonts) was undertaken in Tallinn to avoid the possibility of interlaboratory variation [Brodie et al., 2011a, 2011b]. Such effects, however, can a priori not be excluded when comparing the results from bulk organic matter (acid treatment in Erlangen) with those from the palynomorphs. For the transfer of liquid residue between Tallinn and Ghent, a few drops of formaldehyde were added to the residue to exclude the potential risk of the growth of, and contamination by, modern organisms. We confirmed that the formaldehyde did not have an influence on the δ13C measurements of the palynomorphs, by processing an additional sample in Ghent (L2-26c, from the nearby Lickershamn section), once without adding formaldehyde, and a second time with added formaldehyde: the difference in δ13C between these two tests only was of 0.15‰, i.e., much lower than significant offsets discussed in the results section (L2-26c without formaldehyde: 0.08 mg of chitinozoans: –26.68‰; L2-26c with formaldehyde: 0.05 mg of chitinozoans: –26.83‰).

Isotope analysis of the organic materials (decalcified rocks, chitinozoans, and scolecodonts) were performed with an elemental analyzer (Carlo-Erba1110) interfaced with a ThermoFinnigan Delta Plus mass spectrometer in the isotope laboratory at GeoCenter Erlangen (Germany). All carbon isotope values are reported relative to the Vienna-PDB standard. Accuracy and reproducibility of the analyses were checked by replicate analyses of an international standard (USGS 40). Reproducibility was better than ±0.08‰ (1σ). In addition, replicate analyses of four δ13Corg samples were performed (Table 1), indicating reproducibility better than 0.1‰ for samples under similar conditions (LU 44, 155), and better than 0.33‰ for samples of highly different weights (LU 46, 121). Variation below 0.3‰ will not be discussed here.

Table 1. Replicate Analyses of Bulk Organic Matter
Date Identifier Weight [mg] Ampl. 44 mV δ13C ‰ Vienna-PDB
19 August 2008 Lu 44 13.76 4745 –27.91
19 August 2008 Lu 44 13.66 5559 –27.93
6 August 2008 Lu 46 1.83 581 –28.04
6 August 2008 Lu 46 10.43 3119 –27.94
6 August 2008 Lu 121 0.55 177 –27.53
6 August 2008 Lu 121 7.97 2124 –27.86
15 August 2008 Lu 155 9.97 2003 –27.41
18 August 2008 Lu 155 10.22 5394 –27.36

4 Results

The δ13C data are presented in Figure 2 and Table 2. The δ13C from whole rock (bulk) organic matter (δ13Corg) are separated according to their original sample lithology, “limestone” or “marl” [cf. Munnecke and Samtleben, 1996]. No systematic difference is observed in δ13Corg data between the limestones and the marls (Figure 2). The δ13C from chitinozoans and scolecodonts are abbreviated to δ13Cchit and δ13Cscol, respectively.

Table 2. δ13C From the Lusklint Section
Sample Height in Section (m) δ13Corg Marl δ13Corg Limestone Scolecodont Weight (mg) δ13Cscol Chitinozoan Weight (mg) δ13Cchit Difference Between δ13C of Palynomorphs
Lu 1 0.00 –27.63
Lu 3 0.20 –27.40
Lu 4 0.30 –27.66
Lu 5 0.40 –27.71
Lu 6 0.50 –27.62 0.03 –28.80 0.05 –28.39 0.41
Lu 7 0.60 –27.65
Lu 8 0.70 –27.36
Lu 9 0.80 –27.39
Lu 10 0.90 –27.91 0.05 –28.61 0.04 –28.44 0.17
Lu 11 1.00 –27.63
Lu 12 1.10 –27.52
Lu 13 1.20 –27.22
Lu 14 1.30 –27.70
Lu 15 1.40 –27.77
Lu 16 1.50 –28.21
Lu 17 1.60 –27.69
Lu 18 1.70 –27.71
Lu 19 1.80 –27.91
Lu 20 1.90 –27.95 0.03 –29.17 0.04 –28.84 0.33
Lu 21 2.00 –27.98
Lu 22 2.10 –27.91
Lu 23 2.20 –27.86
Lu 24 2.30 –28.07
Lu 25 2.40 –27.50
Lu 26 2.50 –27.84
Lu 27 2.60 –28.17
Lu 28 2.70 –27.71
Lu 29 2.80 –27.54
Lu 30 2.90 0.05 no signal 0.03 –28.15
Lu 31 3.00 –27.45
Lu 32 3.10 –27.70
Lu 33 3.20 –27.57
Lu 34 3.30 –27.55
Lu 35 3.40 –28.09
Lu 36 3.50 –28.37
Lu 37 3.60 –27.96
Lu 39 3.80 –28.10
Lu 40 3.90 –28.13 0.04 –29.72 0.05 –28.87 0.85
Lu 41 4.00 –28.05
Lu 42 4.10 –27.96
Lu 43 4.25 –28.17
Lu 44 4.40 –27.91
Lu 45 4.50 –27.96
Lu 46 4.60 –27.99
Lu 47 4.70 –27.90
Lu 48 4.80 –27.53
Lu 49 4.90 –27.55
Lu 50 5.00 –27.59
Lu 51 5.10 –28.01
Lu 52 5.20 –27.90
Lu 53 5.30
Lu 54 5.40 –28.04
Lu 55 5.50 –28.32
Lu 56 5.60 –27.86
Lu 57 5.70 –27.99
Lu 58 5.80 –28.08
Lu 59 5.90 –28.27
Lu 60 6.00 –28.41
Lu 61 6.10 –28.48
Lu 62 6.20 –28.70
Lu 63 6.30 –28.33
Lu 64 6.40 –28.00
Lu 65 6.50 –27.70
Lu 66 6.60 –27.19 0.03 –29.60 0.07 –28.55 1.05
Lu 68 6.70 –28.24
Lu 69 6.80 –28.69
Lu 70 6.90 –28.47
Lu 71 7.00 –28.16 0.04 –29.97 0.05 –28.98 0.99
Lu 72 7.10 –28.19
Lu 73 7.20 –28.33
Lu 74 7.30 –28.24
Lu 75 7.40 –28.40
Lu 76 7.50 –28.52
Lu 77 7.60 –28.17
Lu 78 7.70 –28.25
Lu 79 7.80 –27.84
Lu 80 7.90 –27.99
Lu 81 8.00 –27.85
Lu 82 8.10 –27.92
Lu 83 8.20 –28.10
Lu 84 8.30 –27.67 0.04 –29.67 0.05 –28.83 0.84
Lu 85 8.40 –28.57
Lu 86 8.50 –27.92
Lu 87 8.60 –28.10
Lu 88 8.70 –27.67
Lu 89 8.80 –28.39
Lu 90 8.90 –28.27 0.02 –29.79 0.02 –29.00 0.78
Lu 91 9.00 –28.81
Lu 92 9.10 –28.37
Lu 93 9.20 –28.19
Lu 94 9.30 –28.15
Lu 95 9.40 –28.35
Lu 96 9.50 –28.47
Lu 97 9.60 –28.05
Lu 99 9.90 –28.09
Lu 100 10.00 –29.03 0.11 no signal 0.05 –28.46
Lu 101 10.10 –28.08
Lu 103 10.30 –28.40
Lu 104 10.40 –27.24
Lu 105 10.50 –27.58
Lu 106 10.60 –27.87
Lu 107 10.70 –27.65
Lu 108 10.80 –27.72
Lu 109 10.90 –27.97
Lu 110 11.00 –27.93
Lu 111 11.10 –27.80
Lu 112 11.20 –27.33 0.05 –28.23 0.06 –28.72 0.49
Lu 113 11.30
Lu 114 11.40 –27.54
Lu 115 11.50
Lu 116 11.60 –27.41
Lu 117 11.70 –27.81
Lu 118 11.80 –27.87
Lu 119 11.90 –27.95
Lu 120 12.00 –27.68
Lu 121 12.10 –27.53 0.05 –29.24 0.07 –28.80 0.43
Lu 122 12.20 –27.61
Lu 123 12.30 –27.42
Lu 124 12.40 –27.37
Lu 125 12.50 –25.45
Lu 126 12.60 –27.41
Lu 127 12.70 –27.26
Lu 128 12.80 –27.37
Lu 129 12.90 –27.56
Lu 130 13.00 –28.60
Lu 131 13.10 –27.31
Lu 132 13.20 –27.50
Lu 133 13.30 –27.42
Lu 134 13.40 –27.50
Lu 135 13.50 –27.45 0.02 -28.32 0.06 -27.98 0.34
Lu 136 13.60 –27.25
Lu 137 13.70 –27.92
Lu 138 13.80 –27.33
Lu 139 13.90 –27.89
Lu 140 14.00 –27.45
Lu 141 14.10 –28.10 0.05 –28.06 0.05 –27.70 0.37
Lu 142 14.20 –27.50
Lu 143 14.30 –27.40
Lu 144 14.40 –27.32
Lu 145 14.50 –27.31
Lu 146 14.60 –27.69
Lu 147 14.70 –27.41
Lu 148 14.80 –27.70
Lu 149 14.90 –28.11
Lu 150 15.00 –27.31
Lu 151 15.10 –27.25
Lu 152 15.20 –27.34
Lu 153 15.30 –27.48
Lu 154 15.40 –27.58
Lu 155 15.50 –27.38 0.10 no signal 0.04 –27.66
Average: 0.59

The δ13Corg data show a decreasing trend from the base of the trench section up to a height of ~10 m (from ~ –27.6‰ to –28.3‰). At ~10.5 m, i.e., ~2 m below the Phaulactis layer, δ13Corg increases abruptly by ~1‰. Above 10.5 m δ13Corg are relatively constant at about –27.5‰. The data therefore do show increasing δ13Corg during the Ireviken Event, but deviate in detail from the δ13Ccarb excursion (Figure 1) as published by Munnecke et al. [2003] and Cramer et al. [2010]. The increase in δ13Corg starts earlier (about 2 m lower), and does not display a continuous increase; rather, it represents a marked and abrupt 1‰ rise (or “step”) in the curve. In addition to the section-wide tendencies, the δ13Corg displays more rapid variations, that at times is of important amplitude: for instance, there is a sudden and short-lived ~1.5‰ increase of δ13Corg just below the Lusklint Bentonite (Figure 2).

There is an apparent systematic difference in δ13Corg, δ13Cchit, and δ13Cscol, with the fossil data being generally lower. Chitinozoans are ~0.5 to 1.5‰ lower than δ13Corg, and δ13Cscol are ~0.5 to 1‰ lower than δ13Cchit, with an average difference of 0.6‰ (single exception in sample LU 121 at +11.2 m). The difference in δ13C between chitinozoans and scolecodonts is not constant, however, and differs between samples. Nevertheless, there seems to be a trend in these data: in the lower section (between 0 m and +3 m) the difference is significantly less than average (well below 0.5‰), in the middle section (+3 m to ~+10 m) it rises higher than average (between 0.8 and 1‰), and in the upper section roughly from the Phaulactis layer upward (+10–12 m and above), it drops again below average values (< 0.5‰).

5 Discussion

Chitinozoan egg-cases, frequently found in Ordovician to Devonian strata, have a different preservation potential than their weak-bodied parent organisms, currently unknown from the fossil record [Paris and Nõlvak, 1999]. This is probably due to differences in organic compounds between the egg-cases and organisms, which might imply a certain degree of isotope fractionation. However, because this has not yet been quantified in the Paleozoic, we will treat δ13C of chitinozoan egg-cases as identical, or very close to that of the chitinozoan parent animals. Similarly, fossilizing scolecodonts must have had a significantly different original organic composition than their soft-bodied polychaete hosts, which are extremely rare in the fossil record [Eriksson et al., 2004], entailing a fractionation that we cannot quantify here.

The δ13Cchit data are consistently higher (by 0.5 to 1.5‰) than δ13Cscol. This is in contrast to what might be expected from their hypothetical position in the Silurian food chain sketched in the introduction, assuming that “you are what you eat plus a few per mil” applies [DeNiro and Epstein, 1976, 1978], and that chitinozoans did not biologically fractionate carbon isotopes in a particularly unusual way. Instead, δ13C signatures of the scolecodonts, lighter than those of both the whole rock and chitinozoans, may in part be a function of their habitat, on or within the sea bed of the Silurian Gotland shelf. We suggest that (part of) these jaw-bearing polychaetes were burrowing in the sea bottom, consuming 12C-enriched organic matter, or in contact with bacterially respired CO2, explaining the light δ13Cscol values of their fossilized jaws. Alternatively, the dissolved carbon flux on the seafloor may be derived from organic-bearing sediments depending on the redox state at the sediment-water interface.

δ13Cchit and δ13Cscol contribute to, but are both lower than δ13Corg, so an isotopically heavier component must also exist in the system. An important group of palynomorphs, the acritarchs, have not been analyzed, because of their very small size. Nevertheless, acritarch census from the (~8 Myr younger) Lau Event on Gotland shows that acritarchs and other primary producers make up the bulk of the number of fossilized palynomorphs specimens in the pre-event interval, with up to several thousand specimens per gram of sediment [Stricanne et al., 2006]. However, we hypothesize that acritarchs, believed to be primary producers [Servais et al., 2008], produced lighter rather than heavier δ13C values relative to the other organic fossil components (Figure 3A). Still, other Early Paleozoic “consumers” with a δ13C lower than “producers” have been reported before: graptolites and scolecodonts from the Prague Basin are isotopically lighter than leiosphere-phytoplankton from the same samples [Lécuyer and Paris, 1997]. Another potential, and perhaps more likely source of heavier C isotopes, is present in the organic fraction residues; a relatively important weight-percentage fraction of the organic carbon can consist of nonpalynomorph, amorphous organic matter, of unknown origin.

Details are in the caption following the image
(A) An explanation of the data as a “classic” Silurian food chain. This scenario predicts that δ13C of acritarchs (primary producers), if analysed, would at least be 1 to 2‰ lower than those of chitinozoans (low-level consumers?), based on DeNiro and Epstein [1976, 1978]. The consequence of such a hypothesis is that an important part of the unidentified (amorph) organic matter has to have a consistently much more positive δ13C than the bulk organic material (e.g., of metazoan origin from higher trophic levels). When this material is buried together with the isotopically much lighter components (acritarchs, chitinozoans, scolecodonts) it produces the bulk organic signal (Figure 2). The relatively light scolecodont δ13C values can at least be partly explained if they had an infaunal benthic mode of life. (B) The increasing offset between chitinozoans and scolecodonts in this scenario is largely caused by the lower values of the scolecodonts (which is consistent with the raw data), in turn caused by an increased input of microphytoplankton to the sediment during periods of increased primary productivity. When this period of high productivity ended, potentially linked to the start of the Ireviken extinction event, the offset between chitinozoans and scolecodonts reduced, and the δ13Corg shifted positively by about 1‰.

There is an intriguing increase in the difference between δ13Cchit and δ13Cscol (~0.8–1‰) in the middle part of the Lusklint trench, roughly between +3 and +10 m, which seems to result from the lower values of the scolecodonts (while chitinozoan values remain relatively more constant). So far, we have discussed the offset between chitinozoan and scolecodont δ13C as function of different biological fractionations within the fossil groups, their different organic compound composition and subsequent preservation, and their different habitat. Faunal turnovers within the scolecodont group could provide an explanation for the observed pattern. Their diversity curves at generic level for the Visby beds [Eriksson 2006, Figure 3, page 93] do display important changes in the genus composition of the scolecodont fauna, but, in detail, these do not exactly reflect the pattern we observe. In addition to biological factors, we are therefore looking for a changing external factor that will mainly (or only) influence the environment in which the scolecodont-bearing polychaetes lived, i.e., the sediments in the sea bottom.

δ13C fluctuations can also be driven by variations in primary productivity in these shelf waters, not affecting individual chitinozoans (assuming they do not start consuming more acritarchs simply because they are available). Instead, the increased input of microphytoplankton to the sediment could affect the carbon pool and eventually the diet of the polychaetes (Figure 3B). A consecutive drop in primary productivity would then explain the reduced difference of δ13C between chitinozoans and scolecodonts at ~+10 m in the section.

Fossil acritarch census data (number of specimens per gram of rock) do not provide direct information on primary productivity, but can be used as a proxy for productivity in combination with other data. Such census data are available from a segment of the Lusklint 1 section (Figure 4), as part of a study on the extinction and origination pattern of acritarchs through the Lusklint 1 section [Gelsthorpe, 2004]. In general, the high acritarch counts in the lower part of the Gelsthorpe [2004] section (between ~ +7.5 and 10.5 m in Figure 2) confirm our interpretation of the isotope data in terms of increased productivity, and low acritarch counts above the Ireviken bentonite (between ~ +10.5 and 15 m in Figure 2) are in accordance with our suggested decrease in primary productivity. Only the low acritarch counts in the first sampled meter of the Gelsthorpe [2004] log (between ~ +6.5 and 7.5 m in Figure 2) are difficult to explain.

Details are in the caption following the image
Acritarch census data, previously only available in a (partly unpublished) Ph.D. thesis, are reproduced here [Gelsthorpe, 2002]. For details on sampling, see Gelsthorpe [2004]; for details on methodology, see Gelsthorpe [2002]. Unfortunately, the trench section sampled by Gelsthorpe [2002, 2004] does not fully overlap with the one sampled in this paper. The original measured section (including the position and nomenclature of marker beds used) of Gelsthorpe [2002] is given on the far right of the figure, for completeness, and for easy comparison with the acritarch turnover data [Gelsthorpe, 2004]. This correlates well with the measured section used to collect the isotope samples, although there are some unavoidable, small differences. The latter measurements are given on the far left of the figure (and match those of Figure 2). All measured heights in the column given in the text, refer to the section measured for isotope samples (far left of this figure, or Figure 2). The Lusklint bentonite has been used as reference level for correlation between both stratigraphic frameworks.

In terms of diversity, acritarchs did not suffer much during the Ireviken Event, where overall the number of originations was higher than the number of extinctions. Acritarch species that went extinct, essentially did so above datum 6 of the conodont extinction event, i.e., well above the positive shift in our δ13Corg data (just above datum 2). Acritarch originations occur throughout the section [Gelsthorpe, 2004]. The δ13Corg shift thus seems to be unrelated to acritarch turnover events at species level.

The interpreted productivity changes would also have implications for the amount and preservation of primary producers in relation to other organic components making up the whole rock carbon (Figure 3), and thus for δ13Corg. Our data do not allow us to discriminate whether this mechanism has triggered the shift to lower δ13Corg values through the Lower Visby Formation, or only provided positive feedback for a trend that already started below the sampled section (unexposed on Gotland). In addition, there is a striking correlation between the interpreted productivity drop based on the δ13Cchit–δ13Cscol comparison, the stepped 1‰ increase in δ13Corg at 10.5 m, and the drop in number of acritarchs per gram of sediment [Gelsthorpe, 2004]. Moreover, a drop in primary productivity, evidenced by decreasing relative and absolute acritarch/sphaeromorph/prasinophyte abundances, has been reported at the onset of the Lau event on Gotland, one of the main other “events” in the Silurian [Stricanne et al., 2006], and has been suggested as a potential general trigger for the Llandovery-Wenlock δ13Ccarb excursions [Cramer and Saltzman, 2007a]. This included the δ13Ccarb excursion at the Ireviken Event, although here we note a stratigraphic offset of a couple of meters between the δ13Corg event / suggested drop in biodiversity (between datum 2 and 3) and the onset of the δ13Ccarb excursion (at the Phaulactis layer, i.e., at datum 4).

Changes in the offset between δ13Cchit and δ13Cscol, and a 1‰ shift in the δ13Corg at the very beginning of the Ireviken extinction event, suggest a link between the extinction of certain groups or species and the way carbon is distributed in this biotope and its sediments. Overall trends in δ13Cchit and δ13Cscol follow those in δ13Corg. The necessity of using isolated palynomorphs (Figure 5), a time-consuming and expensive protocol, as compared to bulk organic matter (or “whole rock”) seems low if the objective of the study is to observe general trends in δ13Corg signatures. However, this conclusion should not be generalized for other sections, especially not those more prone to reworking of fossils and sediments.

Details are in the caption following the image
Palynomorphs discussed in the text. All specimens are from sample Lu100. Acritarchs: (A) Visbysphaera sp., (B) Diexallophasis remota, and (C) Visbysphaera sp. Chitinozoans: (D) Ancyrochitina sp., (E) Conochitina visbyensis, and (F) Angochitina longicollis. Scolecodonts: (G–I) Oenonites spp. Single-scale bar: 100 µm; double-scale bar: 50 µm; triple-scale bar: 20 µm.

Sudden, short-lived variation is even more difficult to explain, and may mainly represent changing local conditions. The Storbrut and Ireviken bentonites seem to be associated with a rapid change from more positive values before the events, to more negative values afterwards, and there might be a causal link between the bentonites and δ13Corg anomalies. However, a plausible mechanism has not yet been defined. Similarly, some of the extinction events seem to be preceded by minima in δ13Corg (e.g., datums 1, 3, 4/5, and 6); however, since these minima mostly are well within the overall short-term variability of the δ13Corg record, their significance, if any, remains speculative.

6 Conclusions

  1. In the Lusklint section (Llandovery-Wenlock of E. Sweden) δ13Corg is consistently higher than δ13Cchit and δ13Cscol.
  2. There is a rising trend in the difference in δ13Cchit and δ13Cscol through the lower part of the section, decreasing again at the start of the Ireviken Event, below the Phaulactis layer.
  3. The difference between δ13Cchit and δ13Cscol could, to a certain degree, reflect biological fractionation or biochemical “vital” effects, but can also, at least partly, be explained by their respective epiplanktonic and (infaunal?) benthic mode of life, the latter group perhaps taking up more bacterially respired CO2 from within the sediment.
  4. This δ13C difference between the fossil groups is not constant, but displays a trend through the section. Therefore, it must be amplified by environmental change; here we suggest that increased contrast might be due to increased primary productivity, dropping back to background just before, or at the start of, the Ireviken Event. This is largely confirmed by the available acritarch abundance data. This increased productivity may also have triggered or enhanced the declining δ13Corg trend through the Lower Visby Formation, while the return to normal or reduced production could have influenced the 1‰ δ13Corg abrupt “step” or excursion, a few meters below the onset of the “Ireviken Event” δ13Ccarb excursion. The changes in the carbon cycle, expressed through the combined δ13C data at the very beginning of the Ireviken extinction event suggest a relatively close tie between the both perturbations, although the exact cause-and-effect relations remain unclear.
  5. All three components measured for δ13C show similar curves suggesting that the carbon isotope variations seen in the Gotland section are to a large degree influenced by carbon cycle fluctuations beyond the Baltic shelf environment, and therefore at least regional rather than local process related.

Acknowledgments

We thank Chris Brody (associate editor), Christophe Lécuyer (Lyon1 University) and one anonymous reviewer for their constructive remarks and comments that have greatly improved this paper. We thank Bradley Cramer (University of Iowa), Seth Finnegan (California Institute of Technology), Richard Aldridge (University of Leicester) and Kevin Lepot (Lille1 University) for sharing their ideas on the presented patterns, and Michael Joachimski and Daniele Lutz for the isotope analyses. We thank the Research foundation, Flanders (FWO-Vlaanderen, Belgium) and CNRS (France) for funding to TRAV, the “Agence Nationale de la Recherche” (ANR research project TERRES) for funding to TS, the German Research Foundation (Mu 2352-1) for financial support to AM and GM, Sabine Van Cauwenberghe and Nathalie Vanderputten for the chitinozoan lab work (FWO research grant 3G027105 to Jacques Verniers, Ghent University), and finally Florentin Paris for inspiration and sharing technical know-how. This is contribution to IGCP 591.