Abstract
The δ18O values of phosphatic microfossils recovered from NW Ireland are used to determine the timing and magnitude of cooling associated with the onset of the Carboniferous glaciation. Microfossil fish δ18Oapatite demonstrates a +2.4‰ (V-SMOW) shift, which, once corrected for δ18Oseawater changes owing to evolving ice volumes, equates to an approximate 4.5 °C reduction in equatorial sea surface temperature between the earliest Asbian and the mid-Brigantian (late Visean). Both conodont and microfossil fish δ18Oapatite indicate stabilization of an ‘icehouse’ climate during the Brigantian and into the Serpukhovian. Substantial late Visean cooling identified herein is in good agreement with global glacioeustatic records.
Supplementary material: Further information on the palaeogeography, lithostratigraphy and palaeoenvironments of the sections examined as well as the sample composition and analytical methods of oxygen isotope analyses is available at http://www.geolsoc.org.uk/SUP18511.
The Permo-Carboniferous glaciation (late Palaeozoic ice age) is recognized as one of the most significant in Earth’s history and has also been identified as an important analogue for modern climate (Montañez et al. 2007). Despite this, considerable disagreement remains, particularly concerning the timing of the main phase of glaciation. Earlier studies proposed a relatively simplified protracted glaciation throughout much of the Carboniferous and Permian (e.g. Veevers & Powell 1987). More recently, discrete glacial and temperate episodes have been recognized superimposed on larger-scale climate trends (e.g. Fielding et al. 2008).
Glaciogenic deposits and erosional surfaces provide, essentially, the only direct evidence for past glaciations. However, significant difficulties are associated with their use in this respect, as they are often undatable, extensively reworked or even completely eroded during subsequent glacial advances (Frank et al. 2008). For this reason, distal proxies have become popular for reconstructing ancient ‘icehouse’ phases. By definition, however, proxies provide only a substitute record and all of the controlling factors of the ‘system’ must be well understood before palaeoclimatic interpretations are attempted.
The proliferation in the use of stable isotopes, particularly those of oxygen, to construct high-resolution records of palaeoenvironmental shifts has demonstrated the potential for well-understood proxy systems. The bulk of published work has focused on utilizing carbonates as isotopic media; however, as ancient carbonates are susceptible to diagenetic alteration, screening of older samples is vital. Although methods have been developed to assist in selecting only well-preserved carbonate samples for analysis (e.g. Mii et al. 2001), the potential for carbonate alteration has led to an expansion in the use of biogenic phosphate as an alternative isotopic reservoir. The relative high fidelity of the isotopic signal preserved by biogenic phosphate has been demonstrated through a number of histological and geochemical studies (e.g. Trotter et al. 2007; Joachimski et al. 2009). Isotopic analyses of Palaeozoic phosphates are predominantly undertaken on the bio-apatite of ichthyolith (fish microfossil) and conodont elements. The latter represent the only mineralized parts of an extinct group of small, pelagic or nektobenthic, primitive, jawless marine vertebrates (Aldridge et al. 1993). Importantly, conodont elements have an outer layer of hyaline tissue, which, like ichthyolith enameloid, is densely crystalline and ideal for preserving primary δ18O signatures.
During the Early Carboniferous, Ireland was located just south of the palaeoequator and marine sediments accumulated across a series of shallow platforms and basins during a transgressive phase spanning the suspected onset of the late Palaeozoic ice age. Conodonts and ichthyoliths recovered from five related, late Visean to Serpukhovian, thermally mature (conodont colour alteration index (CAI) of c. 3.5) sedimentary sequences were isotopically analysed. The predominantly shallow marine succession examined in NW Ireland represents a unique opportunity to examine the onset of the late Palaeozoic ice age, owing to (1) the relatively expanded temporal record preserved, (2) commonly occurring ichthyolith and conodont elements and (3) good biostratigraphical control.
Results
The results of oxygen isotope analyses on conodont (n = 12) and ichthyolith (n = 36) apatite samples are presented in Figure 1. δ18Oichthyolith values show an increasing trend from the earliest Asbian to the Brigantian, ranging from 15.1 to 19.1‰ (V-SMOW). Conodont and ichthyolith δ18O values are relatively stable throughout the Brigantian, with conodonts demonstrating a slight decrease from the mid-Brigantian into the Serpukhovian (from 21.1 to 20.2‰ V-SMOW). δ18Oichthyolith values were found to be relatively consistently depleted in 18O by an average of 2.0 ± 0.6‰ (1σ, n = 5) in comparison with conodonts recovered from either the same sample or laterally equivalent stratigraphic levels (Fig. 1). Where multiple samples were analysed from the same level, no statistically significant, consistent δ18O differences were observed that would indicate palaeoecological effects.
δ18O data of conodont and ichthyolith apatite. Columns from the left illustrate western European Stages, British Substages and conodont biozones respectively. Question marks at the Asbian–Brigantian boundary represent a current disagreement in the local biostratigraphy (Sevastopulo & Wyse Jackson 2009, and references therein).
Discrepancy between conodont and ichthyolith δ18O
A ‘vital effect’ is discounted as an explanation for the offset between the two phosphatic media analysed, as no evidence exists for the fundamental physiological difference between fish and conodonts that would be required to explain the essentially stable discrepancy across the range of taxa analysed. Additionally, an offset of a greater magnitude has been identified between similar fish and conodont taxa in coeval but thermally more mature strata elsewhere in Ireland (Barham 2010).
Neither can palaeoecology explain the 2‰ difference between conodonts and ichthyoliths, as significant overlap is expected in the habitats of the conodonts and fish analysed. The average δ18O offset would hypothetically translate into either a habitat palaeotemperature difference of 8–9 °C (Pucéat et al. 2010) or a significant difference in salinity. It is difficult to conceive such large, consistent, primary δ18Oseawater differences either vertically within the tropical water column or laterally across the predominantly shallow platform studied.
Alternatively, it has been suggested that conodont elements, owing to their dense apatite structure (like enamel), more accurately preserve primary environmental signals, whereas the more intrinsically porous Palaeozoic ichthyoliths are prone to isotopic depletion via interaction with relatively 18O-depleted diagenetic fluids (Barham 2010; Žigaite et al. 2010). However, the mechanism of a relatively stable diagenetic O-exchange induced isotopic offset is difficult to conceive and inorganic O-exchange with the PO43− group occurs very slowly (Blake et al. 1997). The deposition of 18O-depleted apatite within the primary pore spaces of ichthyoliths could also account for the observed isotopic offset, although no significant or consistent microscopic evidence for this was observed. δ18Oichthyolith values were interpreted to be less faithful than those of conodonts owing to the more plausible palaeotemperatures derived from δ18Oconodont values. Using the equation of Pucéat et al. (2010), temperatures calculated on δ18Oichthyolith in part exceeded 50 °C, whereas δ18Oconodont values translate into more viable palaeotemperatures of 30–35 °C, considering the palaeogeographical setting of Ireland and uncertainties in estimating the absolute oxygen isotopic composition of Mississippian seawater (Pucéat et al. 2010). In addition, δ18Oconodont values are correlatable: average late Visean δ18Oconodont values of 20.7‰ from NW Ireland are compatible with coeval average values of 21‰ reported by Buggisch et al. (2008) from northern Spain. Furthermore, conodont elements consistently appeared lustrous whereas ichthyoliths, in contrast, commonly displayed coarsened, partially recrystallized basal tissue indicating a different preservation of both phosphatic microfossils.
Although δ18Oichthyolith values appear to have been affected by diagenesis, the offset from (primary) δ18Oconodont values is relatively uniform, which is consistent with the comparative work of Barham (2010) and Žigaite et al. (2010). We tentatively suggest that to compensate for this diagenetic effect δ18Oichthyolith values may be corrected through comparison with temporally and spatially coeval conodonts. δ18Oichthyolith values in this study may therefore be translated by the average offset of +2.0‰ to best estimate their original values and reconstruct a corrected-ichthyolith–conodont δ18O dataset.
Despite probable diagenetic overprinting of ichthyolith oxygen isotope values, the positive trend in the Asbian may be interpreted as primary and representative of palaeoclimatic conditions for the following reasons.
The steady increase in δ18Oichthyolith is statistically unlikely if the data represent a ‘random walk’.
Thermal gradients are insufficient to explain the magnitude of the observed positive trend. Hypothetically, the δ18O increase through the Asbian could be the sole result of complete overprinting at a relatively low temperature (40–50 °C) and a plausible thermal gradient of c. 30 °C km−1. However, complete overprinting of δ18Oichthyolith values is improbable at these relatively low temperatures without enzyme mediation (Blake et al. 1997). It might also be expected that complete overprinting would result in lower δ18O values reflecting the peak temperatures experienced during diagenesis (c. 200 °C according to regional CAI values). Furthermore, partial thermal gradient overprinting would require a hypothetical gradient much steeper than that developed in the region (Corcoran & Clayton 2001). Finally, δ18Oichthyolith values remain stable through Brigantian strata despite similar thermal gradients.
δ18Oichthyolith values from coeval but geographically separate strata are compatible.
As discussed below, the isotopic evolution identified is palaeoenvironmentally plausible.
Palaeotemperature and palaeoclimate interpretations
Regardless of whether absolute δ18Oichthyolith values can be corrected for an offset, the overall increasing δ18O trend can be interpreted independently in terms of palaeoenvironmental change, as the trend retains a primary signal. The positive oxygen isotope shift should, therefore, represent changes in δ18Oseawater associated with the evolution of the depositional environment and the expansion of continental ice masses, as well as cooling of sea surface temperature (SST). Changes in local salinity may have affected δ18Oseawater (the interval of highest δ18Oapatite values broadly coincides with the most restricted conditions). However, the majority of the positive δ18O trend is correlated with platform carbonates, with no substantial salinity changes being evident from the sedimentology or faunal composition. Furthermore, δ18Oapatite values are relatively stable from the Brigantian onwards despite a large range of lithologies (and environments) being sampled in an often rapidly fluctuating sequence, and average δ18Oconodont values from the most restricted depositional interval analysed are essentially identical to those reported from Spain.
Determining the magnitude of SST cooling requires knowledge of the change in δ18Oseawater as a consequence of the growth of isotopically light ice masses (Fig. 2). It is proposed that the climate of the early to mid-Visean may have been transitional to icehouse (Shi & Waterhouse 2010); that is, broadly akin to the present (δ18Oseawater = 0‰ V-SMOW) or less glaciated (δ18Oseawater = −0.5‰ V-SMOW). By Serpukhovian times, ice sheets are interpreted to have been comparable in extent with the Last Glacial Maximum (LGM; González-Bonorino & Eyles 1995), with δ18Oseawater as high as 1.2‰ (Fairbanks 1989). As this dataset does not continue far into the Serpukhovian, a δ18Oseawater value of 1‰ (V-SMOW) is assumed for the peak glacial conditions identified here. This implies that the formation of ice sheets in Gondwana may have accounted for c. 1.5‰ of the total 2.5‰ increase documented in δ18Oapatite. The remaining change of 1‰ is interpreted to reflect c. 4–5 °C cooling, comparable with tropical SST change during the LGM (Beck et al. 1997). It is arguable that this estimate represents a cooling minimum, as the marine lithologies sampled may not capture glacial maxima. This may also explain why no obvious ice sheet coupled isotopic fluctuation is recorded.
Illustration of how SST interpretations may vary depending on assumptions of initial (Init.) and temporal evolution (Δ) of the isotopic composition of seawater. It should be noted that the absolute temperature scale is based on the (probably) erroneous assumption that δ18Oseawater has not changed through time. The trendlines plotted were produced by combining the conodont and ichthyolith datasets after a +2‰ translation of δ18Oichthyolith values.
Comparison of results with other studies
Figure 3 compares the results of this work with a number of significant studies on the late Palaeozoic ice age. The Visean–Serpukhovian boundary has been identified as marking the start of significant climate deterioration using sedimentological data (fig. 1 of Fielding et al. 2008) and oxygen isotope data (Mii et al. 2001; Buggisch et al. 2008). However, late Visean cooling has recently been suggested based on δ18O values of brachiopod shells from Spain (Armendáriz et al. 2008). More significantly, sedimentary cyclicity and eustatic fluctuations with Milankovitch periodicities have been reported from Asbian and Brigantian strata: (1) in the NW Irish sections examined (Schwarzacher 1989; Barham 2010); (2) elsewhere in Ireland and Britain (Wright & Vanstone 2001); (3) in equivalent Canadian sequences (Giles 2009); (4) in the stratigraphically significant Arrow Canyon Section in Nevada, USA (Bishop et al. 2009); (5) elsewhere globally (Rygel et al. 2008, and references therein). The compatibility of this collective body of work strongly indicates an initiation of cooling before the Visean–Serpukhovian transition (Fig. 3).
Comparison of climate proxies used to reconstruct the onset of the late Palaeozoic ice age. Background shading indicates periods suggested to be glaciated by that particular study. Black arrows mark the proposed onset of significant cooling at the start of Asbian time. All data are plotted relative to the time scale of Davydov et al. (2004).
The reasons for the disagreement between supposedly globally representative oxygen isotope records are not entirely clear at present. Despite the usefulness illustrated by isolated studies of low-magnesium brachiopod calcite, the dataset compiled by Frank et al. (2008) demonstrates (1) the significant spread of δ18Ocarbonate data relative to δ18Oapatite, and (2) the incoherence of the trendline, which shows little or no correlation with any other indicators of the late Palaeozoic ice age, produced by combining the results of studies from different palaeogeographical areas.
The palaeogeographical spread of samples is a particularly pertinent consideration, as the majority of oxygen isotope studies are based on sequences deposited in epeiric seas, which because of their somewhat restricted nature are more susceptible to local palaeoenvironmental-induced heterogeneities in δ18Oseawater (Brand et al. 2009). It may be inaccurate to compare the absolute values of coeval oxygen isotope analyses from different continental seas as they probably had different δ18Oseawater properties. The sedimentary strata examined in this study were all deposited from the same regional watermass and where stratigraphic overlap exists, δ18Oapatite values are found to be very consistent (Fig. 1). In any case, only the δ18O trend in this study is given particular credence.
Discrepancies between studies utilizing different fossil hardparts may be explained by currently unrecognized physiological effects or differential diagenetic susceptibilities. More generally, inadequate or different bio- and lithostratigraphic constraints and incomplete stratigraphic resolution of the extremely dynamic climate system may explain conflicting results. In the case of the comprehensive δ18Oapatite dataset of Buggisch et al. (2008), an inability to correlate late Visean to early Serpukhovian isotopic curves may be explained by (1) the variable diagenetic conditions experienced by the Cantabrian sections (CAI 1.5 to 7), and/or (2) difficulties associated with precisely plotting data from several condensed Cantabrian sections. This paper is based on over 2.5 times as many data, within a stratigraphy that is c. 20 times more expanded than that of Buggisch et al. (2008), making for a more resolved picture of climate dynamics during this interval. However, the discrepancies between the various studies clearly indicate the need for more research during this critical climatic interval.
Implications for the onset of the late Palaeozoic ice age
The correlation between increasing δ18Oapatite values and the global lithostratigraphic and sequence-stratigraphic records of glacioeustatic fluctuations represents the first integration of proxies supporting a late Visean onset for a significant phase of the late Palaeozoic ice age. It is proposed that rapid climate deterioration began in earliest Asbian time and by Brigantian time the magnitude of glaciation was comparable with the LGM. An ‘icehouse’ climate state was maintained from mid-Brigantian time and, although limited to a single datum, no rapid degradation of climate apparently occurred at the Visean–Serpukhovian boundary.
Acknowledgments
The authors would like to thank D. Lutz (University of Erlangen–Nürnberg) for instruction in δ18O analyses. G. Sevastopulo (Trinity College Dublin) provided invaluable discussions. The comments of two anonymous reviewers have resulted in a significantly improved paper. This research was funded with a Thomas Crawford Hayes scholarship from NUI, Galway and an Irish Research Council for Science and Engineering Technology EMBARK fellowship.
- © The Geological Society of London
References
The onset of the Permo-Carboniferous glaciation: reconciling global stratigraphic evidence with biogenic apatite δ18O records in the late Visean
