2. Geological settings

The isotopic profile presented in this paper derives from several outcrops along the Kulyumbe River, located ~100 km E of the Enisej River (~68.0° N; 88.7–89° E), at the northwestern margin of the Siberian Platform (Rozova, Reference Rozova1968, Reference Rozova1984; Sokolov, Reference Sokolov1982; Rozanov et al. Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992; Pavlov & Gallet, Reference Pavlov and Gallet1998, Reference Pavlov and Gallet2001) (Fig. 1). The Middle Cambrian–Lower Ordovician sediments are exposed along ~10 km of the Kulyumbe River, including in ascending order: the Ust'-Brus, Labaz, Orakta, Kulyumbe, Ujgur and Iltyk formations (Fig. 2). These deposits are described in several published studies (e.g. Rozova, Reference Rozova1963, Reference Rozova1964, Reference Rozova1968, Reference Rozova1984; Rozova & Yadrenkina, Reference Rozova and Yadrenkina1967; Datsenko et al. Reference Datsenko, Zhuravleva, Lazarenko, Popov and Chernysheva1968).

Figure 2. δ¹³C and δ¹⁸O profiles, ⁸⁷Sr/⁸⁶Sr data, magnetostratigraphy and changes in associations of trilobites through the upper Middle Cambrian–lowermost Ordovician strata of the Kulyumbe section. 7320, etc. as in Figure 1. Samples with δ¹⁸O<−11‰ may possibly be altered during diagenesis and are shown as open circles (δ¹³C curve) and rectangles (δ¹⁸O curve). ⁸⁷Sr/⁸⁶Sr values refer to ‘best-preserved’ samples with Mn/Sr<0.5 and Mg/Ca<0.04 only (see text).* – Rozova (Reference Rozova1968, Reference Rozova1984) placed the Cambrian–Ordovician boundary at the base of Nyajan Horizon (names of horizons given by Rozova Reference Rozova1968, Reference Rozova1984). Single-sample polarity zones are shown by half bars in the magnetostratigraphic column.

The Ust'-Brus Formation, sampled through the upper 250 m, is followed by 630 m of the Labaz Formation. They are characterized by grey, greenish-grey and reddish clayey and silty dolomitic lime mud/wackestones with algal boundstones, lenses of intraclast flat-pebble conglomerates, local cross-bedding and ripple marks.

The Orakta Formation comprises 450 m of grey and greenish-grey clayey and silty dolomitic lime mud/wackestones with lenses of bio-grainstones and packstones, algal boundstones, intraclast flat-pebble conglomerates, local cross-bedding, ripple marks and slump structures. Layers of breccia and megabreccia are developed in the lower part of the formation.

The Kulyumbe Formation is built of 645 m of grey, greenish-grey and reddish clayey and silty dolomitic and dolomitic lime mud/wackestones with grey algal boundstones, lenses of bio- and intraclast grainstones and packstones, intraclast flat-pebble conglomerates, local cross-bedding and ripple marks; slump structures occur in the lower and upper parts of the formation.

The overlying 210 m of the Ujgur Formation is represented by grey, greenish-grey and reddish clayey and silty dolomitic and dolomitic lime mud/wackestones with oolitic beds, algal boundstones and lenses of intraclast flat-pebble conglomerates. Slump structures and quartz-feldspar sandstone beds are reported from the lower part of the formation.

The lower 125 m of the Iltyk Formation (Nyajan Horizon) was sampled in the uppermost part of the studied section. It is composed of grey clayey and silty dolomitic and dolomitic lime mud/wackestones, grey and rare reddish dolo- mud/wackestones, as well as grey algal boundstones.

Folding and dislocations occur in the Kulyumbe section, where beds monoclinally dip at 15–25° E–SE. They are locally intruded by sills reflecting Carboniferous–Triassic tectonic activity at the margin of the Siberian Platform (Pavlov & Gallet, Reference Pavlov and Gallet1998, Reference Pavlov and Gallet2001). These intrusions are not included in our stratigraphic column, but the positions of local marker horizons, such as the Madui and Mansi sills, are indicated by arrows (Figs 1, 2). The Madui sill is ~20 m thick in the river outcrop and located 15–20 m above the base of the Kulyumbe Formation (within the lowermost part of the Entsian Horizon according to Rozova, Reference Rozova1984). The Mansi sill is ~60 m thick, and its base is situated ~50 m above the base of the Ujgur Formation (base of the Mansian Horizon, after Rozova, Reference Rozova1984).

4. Results of the geochemical analyses

The Kulyumbe profile exhibits δ¹³C values which plateau around zero through the upper Ust'-Brus and Labaz formations, followed by a decreasing trend that includes minor oscillations with a peak-to-peak amplitude of 1–2‰ in the Orakta and lowermost Kulyumbe formations (Fig. 2). In the Yurakhian Horizon of the Kulyumbe Formation, the δ¹³C values rise as high as +4.6‰ and then drop, showing the characteristic features of the worldwide SPICE excursion (Fig. 2). The excursion in the Kulyumbe section is defined by >70 samples and extends through ~200 m of carbonate rock. Previously, the excursion was recognized in the same section on the basis of five samples only (S. Ebneth & J. Veizer, unpub. data; Table 3 of Supplementary Material). A smaller positive δ¹³C peak with amplitude of ~1.5‰ is observed immediately above the SPICE excursion, in the lower Khetian Horizon of the Kulyumbe Formation (‘post-SPICE’ in Fig. 2). The δ¹³C values then decrease in an oscillatory fashion towards the lower boundary of the Ujgur Formation. At the boundary between the Kulyumbe and Ujgur formations, δ¹³C values reach a minimum of ~−1‰, before rising in the Ujgur Formation. They peak at ~1‰ in the upper Mansian Horizon and drop to ~0‰ in the upper Loparian Horizon (Fig. 2). In the uppermost Loparian and in the Nyajan Horizon of the Iltyk Formation, the values oscillate near 0‰ and drop to ~−1‰ in the upper Nyajan Horizon.

Trace element and oxygen isotope characteristics may help to assess the state of preservation of carbonate δ¹³C and ⁸⁷Sr/⁸⁶Sr values. Most of the samples show δ¹⁸O values between −6‰ and −11‰. However, carbonates of the upper Entsian Horizon of the Kulyumbe Formation as well as several samples from other parts of the section yielded anomalously low δ¹³C and δ¹⁸O values. In sedimentary carbonates, very low δ¹⁸O values can result from a diagenetic resetting of δ¹⁸O (Kaufman & Knoll, Reference Kaufman and Knoll1995). Although δ¹³C is much less prone to diagenesis than δ¹⁸O, anomalously low oxygen isotope ratios are commonly associated with alteration of the carbon isotopic record as well. Therefore, samples with δ¹⁸O<−11‰ are considered herein with caution, because they are possibly altered, and are shown as open circles (δ¹³C curve) and rectangles (δ¹⁸O curve) in Figure 2. All samples analysed show Mn/Sr<10, except for sample K473 (Mn/Sr = 12) from the Yurakhian Horizon (Fig. 3). Although this sample might be excluded as probably altered (e.g. Kaufman & Knoll, Reference Kaufman and Knoll1995), its δ¹³C and δ¹⁸O values of respectively +3.16‰ and −7.15‰ are not anomalous.

Figure 3. Mg/Ca (a) and Mn/Sr (b) against δ¹⁸O plots showing association of increasing dolomite content with higher δ¹⁸O and elevated Mn/Sr.

Carbon isotope values from the earlier Ottawa–Bochum dataset (Table 3 of Supplementary Material) are consistent with the higher resolution dataset herein, and with published data from Siberia (Brasier & Sukhov, Reference Brasier and Sukhov1998). Oxygen isotope values, however, are more internally consistent in the Ottawa–Bochum dataset, showing a clear trend towards more ¹⁸O-depleted values up-section (Fig. 2; Tables 1 and 3 of Supplementary Material). This difference would appear to reflect mineralogy, in that the smaller Ottawa–Bochum study used selected calcitic samples only, while the more variable δ¹⁸O values of the upper Kulyumbe section studied herein seem likely to reflect variable degrees of dolomitization in that part of the high-resolution profile (Fig. 2). This interpretation is supported by the fact that Mg/Ca, Mn/Sr ratios and δ¹⁸O (Fig. 3) show positive correlation (positive outliers in δ¹⁸O are all partially dolomitized).

Sr isotope ratios range from 0.708732 to 0.709171 through the interval of the Middle Cambrian Mayan Stage to the Cambrian–Ordovician boundary (Table 3 of Supplementary Material). In general, ratios rise through the Middle to Upper Cambrian. The generally negative correlation in older samples between Mn/Sr (and Mg/Ca, in Atdabanian samples only) and ⁸⁷Sr/⁸⁶Sr indicates that uppermost values, unusually, may best approximate contemporaneous seawater ⁸⁷Sr/⁸⁶Sr through the lower half of the section. By contrast, the positive correlation between Mn/Sr, Mg/Ca and ⁸⁷Sr/⁸⁶Sr ratios, as well as the generally high consistency of isotope values through the SPICE interval, indicate a well-preserved ⁸⁷Sr/⁸⁶Sr ratio of 0.709095 for that level. Data scatter is reduced when only the ⁸⁷Sr/⁸⁶Sr ratios of samples below suitable diagenetic cut-off points are considered. If only samples with Mn/Sr<0.5 and Mg/Ca<0.04 are included, as being considered to be least altered (Fig. 2), then ⁸⁷Sr/⁸⁶Sr changed from 0.70885–0.70900 in the Middle Cambrian Labaz Formation to 0.70903–0.70905 in the Orakta Formation, and then to higher values of 0.70909–0.70911 in the Upper Cambrian Kulyumbe Formation.

5. The Middle–Upper Cambrian boundary in the Kulyumbe section

Traditionally, the Middle–Upper Cambrian boundary on the Siberian Platform has been placed at the top of the Lejopyge laevigata–Aldanaspis truncata Biozone of the Mayan Stage (Rozanov et al. Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992, Geyer & Shergold, Reference Geyer and Shergold2000; Fig. 2). In the Kulyumbe section this level is usually traced to the boundary between the Labaz and Orakta formations (Rozova, Reference Rozova1963, Reference Rozova1964, Reference Rozova1968, Reference Rozova1984) and coincides with the base of the Pedinocephalina–Toxotis Biozone (Krasnov et al. Reference Krasnov, Savitsky, Tesakov and Khomentovsky1983; Fig. 2). The ratification by the International Union of Geological Sciences of the Middle–Upper Cambrian boundary between the Glyptagnostus stolidotus and G. reticulatus biozones in the Paibi section on the South China Platform (Peng et al. Reference Peng, Babcock, Robison, Lin, Rees and Saltzman2004; Fig. 4) requires that the traditional position of this boundary on the Siberian Platform be changed accordingly.

Figure 4. Global correlation of the Upper Cambrian–lowermost Ordovician part of the Kulyumbe section. Open circles show samples possibly altered by diagenesis as indicated by the authors, and/or by δ¹⁸O<−11‰ data, where available. Occurrence of Cordylodus proavus (Cp) and C. angulatus (Ca) in the Kulyumbe section according to Tesakov et al. (Reference Tesakov, Kanygin, Yadrenkina, Simonov, Sychev, Abaimova, Divina and Moskalenko2003).

Khos-Nelege, situated on the northeastern Siberian Platform (Fig. 1), is the reference section through the Middle–Upper Cambrian boundary in Siberia (Lazarenko & Pegel', Reference Lazarenko, Pegel', Peng, Babcock and Zhu2001). In this section, the boundary is now placed at the base of the Glyptagnostus reticulatus–Eugonocare (Olenaspella) evansi Biozone (Lazarenko & Pegel', Reference Lazarenko, Pegel', Peng, Babcock and Zhu2001; Fig. 5). The Khos–Nelege section is associated with deeper-water basin/slope facies (Fig. 1) and cosmopolitan trilobites (for the purposes of this paper, the term ‘trilobites’ also includes agnostids). By contrast, the Middle–Upper Cambrian transitional strata of the Kulyumbe section were deposited on the carbonate platform, in inner shelf environments (Fig. 1), where almost all the trilobites are endemic (Rozova, Reference Rozova1963, Reference Rozova1964, Reference Rozova1968, Reference Rozova1984; Rozanov et al. Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992; Sukhov, Reference Sukhov1997; Varlamov, Pak & Rozova, Reference Varlamov, Pak and Rozova2006). Therefore, biostratigraphic correlation between these two sections is difficult, and alternative methods, such as chemostratigraphy, are needed.

Figure 5. Stratigraphy of the Cambrian part of the Ogon'or Formation at Khos-Nelege (see location KN in Fig. 1), after Krasnov et al. (Reference Krasnov, Savitsky, Tesakov and Khomentovsky1983) and Rozanov et al. (Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992).

No major δ¹³C oscillations are revealed from the upper Ust'-Brus, Labaz and Orakta formations, correlated herein with the upper part of the Mayan Stage (Fig. 2). This is in agreement with data from Brasier & Sukhov (Reference Brasier and Sukhov1998) showing a plateau during this interval from the Siberian Platform and with the values oscillating near zero in Laurentia, reported by Saltzman (Reference Saltzman2005), and China (Zhu et al. Reference Zhu, Zhang, Li and Yang2004; Peng et al. Reference Peng, Babcock, Zuo, Lin, Zhu, Yang, Robison, Qi, Bagnoli and Chen2006). Mild oscillations in the Ust'-Brus and Labaz formations (Fig. 2 herein) are similar to those shown by Peng et al. (Reference Peng, Babcock, Zuo, Lin, Zhu, Yang, Robison, Qi, Bagnoli and Chen2006, figs 6, 7) from the Lejopyge laevigata Zone and upper Drumian Stage of China. The upper Stage 4–lowermost Drumian Stage interval in China (Zhu et al. Reference Zhu, Zhang, Li and Yang2004, figs 2, 3, 6; Peng et al. Reference Peng, Babcock, Zuo, Lin, Zhu, Yang, Robison, Qi, Bagnoli and Chen2006, fig. 7; Zhu, Babcock & Peng, Reference Zhu, Babcock and Peng2006), as well as the traditional Middle Cambrian of Laurentia (Montañez et al. Reference Montañez, Osleger, Banner, Mack and Musgrove2000) show pronounced oscillations, probably present in the lower Mayan and Amgan stages of the Middle Cambrian of Siberian Platform (Brasier & Sukhov, Reference Brasier and Sukhov1998; A. Kouchinsky, unpub. obs.). According to biostratigraphic correlation (Geyer & Shergold, Reference Geyer and Shergold2000; Babcock et al. Reference Babcock, Peng, Geyer and Shergold2005; Peng et al. Reference Peng, Babcock, Zuo, Lin, Zhu, Yang, Robison, Qi, Bagnoli and Chen2006), these excursions belong, however, to an interval below the profile sampled at Kulyumbe.

In addition to Siberia, the SPICE excursion has been documented from Laurentia (Saltzman, Runnegar & Lohmann, Reference Saltzman, Runnegar and Lohmann1998; Saltzman et al. Reference Saltzman, Davidson, Holden, Runnegar and Lohmann1995, Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000, Reference Saltzman, Cowan, Runkel, Runnegar, Stewart and Palmer2004; Glumac & Walker, Reference Glumac and Walker1998; Buggisch, in press), Australia (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000), Kazakhstan (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000), China (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000; Peng et al. Reference Peng, Babcock, Robison, Lin, Rees and Saltzman2004; Zhu et al. Reference Zhu, Zhang, Li and Yang2004), Antarctica (Buggisch, Reference Buggisch2006) and Scandinavia (Ahlberg et al. Reference Ahlberg, Axheimer, Eriksson, Schmitz, Terfelt and Jago2006) (Fig. 4). There is some uncertainty, however, in the position of the basal G. reticulatus Biozone with respect to the rising trend of the SPICE excursion (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000). Although the Middle–Upper Cambrian boundary is considered to correspond to the onset of the SPICE excursion, it is situated almost in the middle of the rising trend in its stratotype at Paibi (see Peng et al. Reference Peng, Babcock, Robison, Lin, Rees and Saltzman2004, fig. 6; Fig. 4 herein). In Kazakhstan, Australia and Laurentia, the G. stolidotus–G. reticulatus boundary is associated with carbon isotope values of ~1‰ at the beginning of the rising trend (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000 and Fig. 4 herein).

The two biozones constraining the Middle–Upper Cambrian boundary are now regarded as phylozones, because G. stolidotus is believed to be ancestral to G. reticulatus (Peng et al. Reference Peng, Babcock, Robison, Lin, Rees and Saltzman2004). Non-evolutionary changes in trilobite complexes, such as migrations, have been suggested, however, to be common especially at the beginning of the G. reticulatus Biozone (Rozanov et al. Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992). Palmer (Reference Palmer1965, Reference Palmer1979) also noted that diachronous boundaries of biostratigraphic units occurred in the Middle and Upper Cambrian, but he later suggested that the difference was not significant for biostratigraphic correlation (Palmer, Reference Palmer1984).

A combined carbon isotopic curve through the Middle–Upper Cambrian strata at Paibi and Wa'ergang sections in South China is provided by Saltzman et al. (Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000) (Fig. 4 herein; a vertical arrow in this combined profile indicates a junction between δ¹³C profiles of the Paibi section below and the Wa'ergang section above). Sampling intervals are rather large (several metres to more than 10 m), but a more detailed profile comprising 50 samples through the transitional 12 m at the G. stolidotus–G. reticulatus boundary in the Paibi section has been published (Peng et al. Reference Peng, Babcock, Robison, Lin, Rees and Saltzman2004). The combined data from South China suggest that the Middle–Upper Cambrian boundary in the stratotype is situated at the beginning of the rising trend of the SPICE excursion. Carbon isotopic values through the transitional strata in Paibi oscillate between 1‰ and 2.5‰, with the boundary lying at 1.5–2.0‰ (Peng et al. Reference Peng, Babcock, Robison, Lin, Rees and Saltzman2004, fig. 6) and at ~3‰ in Wa'ergang (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000). Values of ~3‰ at the level of first appearance of G. reticulatus are also reported from the Wangcun section in South China (Zhu et al. Reference Zhu, Zhang, Li and Yang2004).

Based on the position of the SPICE excursion in the Kulyumbe section, the Middle–Upper Cambrian boundary is situated between the base of the Yurakhian Horizon of the Kulyumbe Formation, where carbon isotope values begin to rise, and a level showing values around ~2‰ at the top of section 6 (Figs 1, 2). The position of the Middle–Upper Cambrian boundary, traced to the boundary between the Labaz and Orakta formations according to the local trilobite biostratigraphy, is thus ~700 m lower than that indicated by chemostratigraphy. Therefore, biostratigraphic correlation of the basal Orakta Formation with the G. reticulatus Biozone at Khos-Nelege is unlikely correct.

A worldwide mass extinction of trilobites at the base of the Glyptagnostus reticulatus Biozone marks the beginning of the SPICE excursion (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000). Eighty species from the Glyptagnostus stolidotus zone do not proceed into the G. reticulatus Biozone in Laurentia. The diversity of trilobites increases later, during the rising trend of SPICE. In the Kulyumbe section the diversity of trilobites declines from the Labaz Formation (upper Middle Cambrian) towards the Cambrian–Ordovician boundary, with the most dramatic decline in number of species taking place in the upper Orakta Formation (Fig. 2; Table 5 of Supplementary Material). Trilobite complexes show significant renewal through each boundary between biozones and horizons in the section, including the Middle–Upper Cambrian boundary transition identified herein by chemostratigraphy at the base of the Yurakhian Horizon. The trilobite fauna of the Yurakhian Horizon is completely renewed with respect to that of the Entsian Horizon (Fig. 2), but in contrast with data from Saltzman et al. (Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000), there is no evidence for increase in diversity during the rising trend of the SPICE excursion in the Kulyumbe section. Although new species and genera of trilobites continue to appear throughout the Kulyumbe Formation (Datsenko et al. Reference Datsenko, Zhuravleva, Lazarenko, Popov and Chernysheva1968; Rozova, Reference Rozova1968, Reference Rozova1984), the diversity generally declines (Fig. 2; Table 5 of Supplementary Material). The trilobites from the Kulyumbe section are endemic forms (Rozova, Reference Rozova1963, Reference Rozova1964, Reference Rozova1968, Reference Rozova1984; Rozanov et al. Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992; Sukhov, Reference Sukhov1997; Varlamov, Pak & Rozova, Reference Varlamov, Pak and Rozova2006), and therefore their evolutionary pattern and diversity changes differ from those of forms with global distribution. Hence, the observed decline in diversity at Kulyumbe could be related to local palaeogeography and environments.

6. The Cambrian–Ordovician boundary in the Kulyumbe section

There are two distinct proposals for how to correlate the Kulyumbe section with the base of the Ordovician System. The first one places the Cambrian–Ordovician boundary on the Siberian Platform at the base of the Mansian Horizon (Krasnov et al. Reference Krasnov, Savitsky, Tesakov and Khomentovsky1983; Fig. 2). In the second one (Rozova, Reference Rozova1968, Reference Rozova1984), the boundary is positioned at the base of the Nyajan Horizon. The latter option was supported by the Cambrian–Ordovician boundary International Working Group in 1986 (Fig. 2).

A lower boundary of the Ordovician System was finally ratified by the Cambrian–Ordovician boundary International Working Group at the base of the Tremadocian Stage at Green Point in western Newfoundland, where it was defined by the first appearance of the conodont Iapetognathus fluctivagus (Cooper & Nowlan, Reference Cooper, Nowlan, Kraft and Fatka1999). Thus, the occurrence of Cordylodus proavus can now be used to assign the Loparian and Mansian horizons to the Upper Cambrian (Dubinina, Reference Dubinina2000; Tesakov et al. Reference Tesakov, Kanygin, Yadrenkina, Simonov, Sychev, Abaimova, Divina and Moskalenko2003). The Ordovician C. angulatus was found in the Nyajan Horizon (Tesakov et al. Reference Tesakov, Kanygin, Yadrenkina, Simonov, Sychev, Abaimova, Divina and Moskalenko2003), but taxonomic revision of Siberian Cordylodus species is still needed (Tatiana Tolmacheva, pers. comm. 2006).

Although the carbon isotopic record from the Cambrian–Ordovician boundary stratotype at Green Point seems to be distorted by diagenesis (Cooper, Nowlan & Williams, Reference Cooper, Nowlan and Williams2001), a combination of data from other sections, such as Dayangcha in China (Ripperdan, Magaritz & Kirschvink, Reference Ripperdan, Magaritz and Kirschvink1993), Black Mountain in Australia (Ripperdan et al. Reference Ripperdan, Magaritz, Nicoll and Shergold1992; Miller et al. Reference Miller, Ethington, Evans, Holmer, Loch, Popov, Repetski, Ripperdan and Taylor2006) and Lawson Cove in Utah (Ripperdan & Miller, Reference Ripperdan, Miller, Cooper, Droser and Finney1995; Miller et al. Reference Miller, Ethington, Evans, Holmer, Loch, Popov, Repetski, Ripperdan and Taylor2006) enabled a generalized (global) curve to be compiled (Cooper, Nowlan & Williams, Reference Cooper, Nowlan and Williams2001, fig. 4; Fig. 4 herein). The Cambrian–Ordovician boundary is marked by the highest δ¹³C value of the transitional interval, in the Iapetognathus Biozone, before the value decreases through the C. angulatus Biozone (Cooper, Nowlan & Williams, Reference Cooper, Nowlan and Williams2001). There are also two main positive cycles below this excursion, between the base of the Cordylodus proavus and the base of the C. lindstromi biozones (Ripperdan & Miller, Reference Ripperdan, Miller, Cooper, Droser and Finney1995; Cooper, Nowlan & Williams, Reference Cooper, Nowlan and Williams2001, fig. 4; Miller et al. Reference Miller, Ethington, Evans, Holmer, Loch, Popov, Repetski, Ripperdan and Taylor2006).

A Lower–Middle Ordovician carbon isotopic curve of high resolution and biostratigraphically constrained by Laurentian conodont biozones was published by Buggisch, Keller & Lehnert (Reference Buggisch, Keller and Lehnert2003) from an isolated terrane of probable Laurentian provenance in Argentina. However, there are no species of Iapetognatus and Cordylodus known from that section, probably because of unfavourable facies. The SPICE excursion is also missing from that profile, because of a stratigraphic hiatus. The lower and upper boundaries of the Tremadocian Stage are marked by falling values from local maxima at the Cambrian–Ordovician boundary and in the upper Tremadocian (see Buggisch, Keller & Lehnert, Reference Buggisch, Keller and Lehnert2003, fig. 6). A single positive excursion of 2‰ amplitude (from −1‰ to 1‰) in the R. manitouensis Biozone is unknown beyond Argentina owing to lack of data from this stratigraphic interval (Buggisch, Keller & Lehnert, Reference Buggisch, Keller and Lehnert2003). Otherwise, the Upper Cambrian–Tremadocian portion of this curve can be satisfactorily correlated with other transitional profiles (Buggisch, Keller & Lehnert, Reference Buggisch, Keller and Lehnert2003; Fig. 4 herein).

There are apparently three positive excursions and four troughs above SPICE in the Kulyumbe section (Fig. 2). The lower maximum is situated in the lower Khetian Horizon of the Kulyumbe Formation (‘post-SPICE’ excursion in Fig. 2). It is similar in position and magnitude to a reported positive δ¹³C feature with an amplitude up to 2‰ observed immediately above the SPICE excursion in Laurentia, above the last occurrence of Irvingella (Saltzman, Reference Saltzman2001), and above the Ivshinagnostus ivshini Biozone in Kazakhstan (Saltzman et al. Reference Saltzman, Ripperdan, Brasier, Lohmann, Robinson, Chang, Peng, Ergaliev and Runnegar2000). The subsequent decrease in values coupled with the SPICE excursion is recognized as the HERB event, a Sunwaptan-aged negative δ¹³C anomaly identified in Australia and Laurentia (Ripperdan et al. Reference Ripperdan, Magaritz, Nicoll and Shergold1992; Ripperdan, Reference Ripperdan2002; Miller et al. Reference Miller, Ethington, Evans, Holmer, Loch, Popov, Repetski, Ripperdan and Taylor2006). It is represented by minimum values of −3‰ to −4‰ in Australia (Ripperdan et al. Reference Ripperdan, Magaritz, Nicoll and Shergold1992; Miller et al. Reference Miller, Ethington, Evans, Holmer, Loch, Popov, Repetski, Ripperdan and Taylor2006; Fig. 4) or a less pronounced trough in Laurentia (Buggisch, Keller & Lehnert, Reference Buggisch, Keller and Lehnert2003; Saltzman et al. Reference Saltzman, Cowan, Runkel, Runnegar, Stewart and Palmer2004; Miller et al. Reference Miller, Ethington, Evans, Holmer, Loch, Popov, Repetski, Ripperdan and Taylor2006; Fig. 4). Buggisch, Keller & Lehnert (Reference Buggisch, Keller and Lehnert2003) suggested, however, that negative anomalies, such as HERB, could have been produced or enhanced by diagenesis at sequence boundaries rather than representing the true magnitudes of primary oceanic carbon reservoir changes. Although more pronounced, the HERB minimum can be tentatively correlated to an isotopic minimum of −1‰ at the boundary between the Kulyumbe and Ujgur formations (HERB? in Figs 2, 4). The location of these features in the Khetian Horizon is consistent with its terminal Upper Cambrian affinity.

At least two more peaks appear in the Kulyumbe section, within the Ujgur and Iltyk formations. Through a combination of chemostratigraphy and biostratigraphy, a lowermost possible and a most probable position of the Cambrian–Ordovician boundary in Kulyumbe may be indicated. The upper portion of the Iltyk Formation is referred to as the Ugorian Horizon (not shown here) and tentatively attributed to the ‘Arenig Stage’ of the Lower Ordovician (Rozanov et al. Reference Rozanov, Repina, Apollonov, Shabanov, Zhuravlev, Pegel', Fedorov, Astashkin, Zhuravleva, Egorova, Chugaeva, Dubinina, Ermak, Esakova, Sundukov, Sukhov and Zhemchuzhnikov1992). According to biostratigraphy, the lower Tremadocian boundary is situated below the Ugorian Horizon. On the other hand, based on the available chemostratigraphic information, the Cambrian–Ordovician boundary is unlikely to be situated at the base of the Mansian Horizon because of a well-defined isotopic trough there (HERB? in Figs 2, 4). Two intervals in the Kulyumbe profile may therefore be suggested for potential placement of the Cambrian–Ordovician boundary: the upper Mansian and the upper Loparian–middle Nyajan horizons. The upper Mansian isotopic maximum represents the first local maximum after the post-SPICE and HERB? excursions, and its position is therefore regarded to be too low for the Cambrian–Ordovician boundary (Figs 2, 4). A more probable location for this boundary lies within the upper Loparian–middle Nyajan interval, but the positive excursion there may be composite, since values oscillate throughout these beds, and the isotopic signature may be insufficiently resolved owing to larger sampling intervals (Figs 2, 4). Nevertheless, a fall in carbon isotope values in the upper Nyajan Horizon and its biostratigraphic correlation with the lowermost Ordovician (Dubinina, Reference Dubinina2000; Tesakov et al. Reference Tesakov, Kanygin, Yadrenkina, Simonov, Sychev, Abaimova, Divina and Moskalenko2003) is consistent with our proposition that the most probable position of the Cambrian–Ordovician boundary at Kulyumbe is in the middle part of the Nyajan Horizon, in outcrop 7322 (Figs 1, 2, 4).

7. Strontium isotopic interpretation

Our data imply a subtle rise in seawater ⁸⁷Sr/⁸⁶Sr from the Middle Cambrian to the middle of the Late Cambrian. This rise might continue to all-time high values for the Phanerozoic, possibly as high as 0.70920, as recorded in the slightly younger upper Pterocephaliid biomere in the USA (Saltzman et al. Reference Saltzman, Davidson, Holden, Runnegar and Lohmann1995). Published micrite and conodont ⁸⁷Sr/⁸⁶Sr data show a subsequent decrease across the Cambrian–Ordovician boundary worldwide, from 0.70910 to 0.70900 (Saltzman et al. Reference Saltzman, Davidson, Holden, Runnegar and Lohmann1995; Ebneth et al. Reference Ebneth, Shields, Veizer, Miller and Shergold2001). Strontium isotope ratios were as high during the Middle–Late Cambrian as they are today, and possibly even higher (Saltzman et al. Reference Saltzman, Davidson, Holden, Runnegar and Lohmann1995; Montañez et al. Reference Montañez, Osleger, Banner, Mack and Musgrove2000), which indicates a unique importance during that time of chemical weathering on ocean composition relative to that of submarine hydrothermal exchange with juvenile ocean crust (Ebneth et al. Reference Ebneth, Shields, Veizer, Miller and Shergold2001). The unique situation during late Cambrian times is underscored by the fact that the ⁸⁷Sr/⁸⁶Sr ratio of carbonate rocks undergoing weathering and that of the depleted mantle would have been considerably lower than today, thus pushing seawater ⁸⁷Sr/⁸⁶Sr to lower, rather than higher, values. The high ⁸⁷Sr/⁸⁶Sr of the late Cambrian ocean has been attributed to elevated chemical weathering rates of crustal silicate minerals uplifted during the Pan-African (Montañez et al. Reference Montañez, Osleger, Banner, Mack and Musgrove2000) and Transgondwanan orogenies (Squire et al. Reference Squire, Campbell, Allen and Wilson2006).

8. Palaeomagnetism

The detailed magnetostratigraphic log of the Kulyumbe section presented herein (Fig. 2) combines published magnetostratigraphic columns for the Middle Cambrian–Lower Ordovician interval (Pavlov & Gallet, Reference Pavlov and Gallet1998, Reference Pavlov and Gallet2001), corrected, refined and complemented with additional data from the Orakta Formation (Table 1 of Supplementary Material).

Additional sampling in 1998 from the lower part of the Entsian Horizon revealed three intervals of normal polarity instead of the two reported by Pavlov & Gallet (Reference Pavlov and Gallet1998) (see Pavlov & Gallet, Reference Pavlov and Gallet2005, fig. 4; Fig. 2 herein). There are also two new very narrow reversed-polarity zones confirmed only by single samples from the lower Loparian Horizon (sample K294) and the middle Labaz Formation (sample 410), as well as several narrow polarity intervals in the middle Labaz Formation, mostly based on single samples (among samples 531–846). New and corrected polarity intervals are indicated by arrows in Figure 2 (see also comments in Table 1 of Supplementary Material). Polarity zones represented by a single sample only are shown by half bars in the magnetostratigraphic column in Figure 2 and Table 1 of the Supplementary Material. They await confirmation after more detailed sampling. Most of the sampling gaps larger than 10 m are shown in the new version of magnetostratigraphic column by hiatuses (Fig. 2).

Data from the Orakta Formation were obtained from samples collected in 1998 and yielded a very weak and noisy palaeomagnetic signal. Isolation of characteristic palaeomagnetic components from most of these samples was either impossible or very difficult. Nevertheless, some samples show a palaeomagnetic two-component record (fig. 6a) suggesting a direction of the characteristic magnetization (fig. 6c). Some samples show three components of magnetization: less stable, recent (low-temperature) and traps-related (middle-temperature) ones, and a more stable, characteristic (high-temperature) component. The mean palaeomagnetic direction of the isolated characteristic component is close to that which is well defined from the Labaz and Kulyumbe formations. Some samples from the Orakta Formation do not allow determination of the direction of the characteristic component (fig. 6b), but their behaviour during analyses indicates the polarity of the ancient magnetization (marked ‘great circle’ in Table 1 of Supplementary Material). Given the overall low quality of the palaeomagnetic record in the Orakta Formation, a corresponding polarity sequence needs to be confirmed by data from other sections and should therefore be considered as preliminary (indicated as ‘low quality signal’ in Fig. 2).

Figure 6. (a) Zijderveld diagrams; open (solid) circles – projections on vertical (horizontal) plane. (b) Demagnetization behavior of natural remnant magnetization which indicates the normal polarity of the characteristic component; solid circles – projections on lower hemisphere. (c) Distribution of isolated directions of the characteristic component (stratigraphic system of coordinates); solid (open) circles – projections on lower (upper) hemisphere.

Thus, a succession of 19–20 magnetic intervals from the Orakta and lower Kulyumbe formations can probably be added to 54–80 intervals previously reported by Pavlov & Gallet (Reference Pavlov and Gallet2005) from the Middle Cambrian of Siberia. This number is again a minimum estimate, as many magnetic intervals have a small stratigraphic thickness of less than 1 m and are based on single samples. Our combined results confirm observations (Pavlov & Gallet, Reference Pavlov and Gallet2001; Gallet, Pavlov & Courtillot, Reference Gallet, Pavlov and Courtillot2003) that during the Middle Cambrian the geomagnetic reversal frequency was as high as 7–10 reversals per Ma, assuming a total duration of about 10 Ma and up to 100 magnetic intervals in the Middle Cambrian. By contrast, the sequence attributed herein to the Upper Cambrian on the grounds of chemostratigraphic correlation contains only 10–11 magnetic intervals, whereas there are 2–3 intervals in the lowermost Ordovician (lower Tremadocian) part (Fig. 2). Hence, magnetostratigraphic data from the Upper Cambrian strata of the Kulyumbe section obtained by Pavlov & Gallet (Reference Pavlov and Gallet1998) and adjusted herein show a one-order-of-magnitude lower number and frequency of magnetic reversals than in the Middle Cambrian.

Upper Cambrian–Lower Ordovician magnetostratigraphic results from several sections, such as the Black Mountain in Australia (Ripperdan et al. Reference Ripperdan, Magaritz, Nicoll and Shergold1992), Dayangcha (Ripperdan, Magaritz & Kirschvink, Reference Ripperdan, Magaritz and Kirschvink1993) and Tangchan (Yang et al. Reference Yang, Otofuji, Sun and Huang2002) in China, Green Point in western Newfoundland, Bartyrbay in Kazakhstan, and from Texas (Ripperdan & Kirschvink, Reference Ripperdan, Kirschvink, Webby and Laurie1992) do not, however, provide a standardized magnetostratigraphic scale through the Cambrian–Ordovician interval (Cooper, Nowlan & Williams, Reference Cooper, Nowlan and Williams2001). A tentative correlation of magnetostratigraphic zones of the Upper Cambrian–Lower Ordovician of China and Australia with the Kulyumbe section was suggested by Pavlov & Gallet (Reference Pavlov and Gallet2005). It has been also reported that in general the geomagnetic field had a predominantly reversed polarity through the Cambrian–Ordovician boundary interval, but with episodes of normal polarity in the late Proconodontus, late Eoconodontus biozones of the Upper Cambrian and in the Early Ordovician (Cooper, Nowlan & Williams, Reference Cooper, Nowlan and Williams2001). As for the Middle Cambrian and most of the Upper Cambrian, palaeomagnetic zonation is known only from the Siberian Platform (Pavlov & Gallet, Reference Pavlov and Gallet1998, Reference Pavlov and Gallet2001, Reference Pavlov and Gallet2005; this study).