Three global carbon isotope shifts in the Silurian of Podolia (Ukraine): stratigraphical implications

Podolia is a classical Silurian area in the southwestern part of the Ukraine. Shallow to open shelf rocks cropping out there are usually, except for a few dolomitic horizons, highly fossiliferous and therefore the biostratigraphy of the region has a long successful history. The Ukrainian meeting of the Subcommission on Silurian Stratigraphy of the International Union of Geological Sciences, held in 1983, became an important milestone. Tsegelnyuk et al. summarized different results and views in a field guide published for the meeting (The Silurian of Podolia. The Guide to Excursion. Naukova Dumka, Kiev, 1983) and introduced a practically new stratigraphical terminology, and in some parts of the section, also new age interpretation of the beds. For our paper carbon isotopes were studied in eight sections on the banks of the Dniester River and its tributaries, covering ca 80% of the Silurian succession (Pridoli excluded). Three positive δC excursions were identified. The first excursion in the Kitaigorod 30 section reaches its peak value of 4.0 in the very bottom of the Demshin Subformation. The second shift reaches the peak value of 4.3 in the middle of the Muksha Subformation. The third shift is the most prominent one identified in the Silurian of Podolia  δC values increase steadily through the Isakovtsy Subformation, reaching 6.9 in the top, and a slightly lower value of 6.6 occurs in the bottom of the Prigorodok Formation. In the middle Silurian of the World three major positive excursions have been identified: in the early and latest Wenlock and in the late Ludlow. The general character of the carbon isotope trend and stratigraphical positions of the excursions established in this paper demonstrate that in Podolia there occurs the same set of global shifts, which can be used for the improvement of regional and global correlation of Wenlock and Ludlow sections of that area. Some refinements are suggested, but some details need additional study.


INTRODUCTION
Podolia is a classical area of the Silurian in the southwestern part of the Ukraine, where various shallow to open shelf rocks crop out magnificently on the banks of the Dniester River and its tributaries.The outcrop area lies approximately from the Studenitsa River in the east up to the Nichlava in the west (some 30 km from Khotin, Fig. 1).Except for a few dolomitic horizons, these rocks are usually highly fossiliferous and therefore the biostratigraphy of the area has a long and successful history.English-language overviews of the study history, biostratigraphy, outcrop lithologies, fossil distribution charts, and palaeogeography are available in Tsegelnyuk et al. (1983), Koren et al. (1989), and Baarli et al. (2003), together with the most important references to the main primary data sources (mostly in Russian).
The above papers contain all of the basic data needed for a clear understanding of our study.Having in mind the main idea of our paper, we would like to underline only a few most important milestones from the 135-year-long history of stratigraphical research in the Silurian of Podolia.Kozlowsky (1929) summarized earlier results in a palaeontological monograph and established four subdivisions in the "etage de Scala" belonging to the upper Silurian.The stratigraphical classification by Nikiforova (1954) introduced a traditional terminology and interregional correlation which have partly been revised later (Nikiforova & Predtechensky 1972;Kaljo 1987).However, in the broad sense this classification is still in use in some publications.In the 1970s Petro D. Tsegelnyuk initiated a multidisciplinary revision of the sequence, including stratigraphy (especially lithostratigraphy), different groups of fossils, metabentonite beds for correlation purposes, etc.A large group of Ukrainian and some Russian colleagues worked together with him and a huge amount of new information was collected.The results and views, published by Tsegelnyuk et al. (1983)  All 8 outcrops (Fig. 1) were sampled at a regular 1 m interval.This gave us nearly 170 whole-rock samples, but there remained several unsampled intervals (marked as sampling gaps in the outcrop logs), including the lower Restevo and lower Sursha subformations, most of the Konovka Formation, and a few metres at the boundary between the Sokol and Bernovo subformations.These sampling gaps were intentional and will be filled during the succeeding project.Still, the most important parts (ca 80%) of the Silurian sequence of Podolia are well covered by samples and accomplishing the study tasks is sufficiently guaranteed by the collected data.

ANALYTICAL METHODS
The goals of our study, in particular the compilation of a general carbon isotope trend for the Silurian of Podolia, require analyses made at regular sampling intervals.For this purpose we analysed bulk-rock samples, not calcite from brachiopod shells, which are commonly considered to be better for isotope studies but occur too unevenly for such kind of research as our project.Most of the analysed rocks are limestones with a rather low content of dolomite, however, real dolomites occur at some levels.For isotope analysis it is useful to know bands with a calcite-dolomite mixture and therefore a series of XRF analyses (from 48 samples) was performed by K. Orlova and T. Kallaste in the XRF Laboratory, Institute of Geology at Tallinn University of Technology (below abbreviated as IG TUT).The main components and some minor elements (Al 2 O 3 , SiO 2 , Fe 2 O 3 , K 2 O, CaO, MgO, etc., Table 1) were measured by  ranging from the Restevo to Ustje subformations and the second from the Goloskov Subformation to the Prigorodok Formation.Both cycles are characterized in the lower 2/3 of the cycle by low, and in the upper 1/3 by increasing and high contents of dolomite, but by opposite contents and trend of calcite.No doubt the described cyclicity reflects changes in the facies-palaeogeographical situation in the basin and also environmental conditions in a wider context.The results of carbon isotope analyses are presented in Table 3.Following a common tradition, we give besides δ 13 C also δ 18 O data, but we strongly underline that due to the specifics of the method whole-rock oxygen isotope data cannot be used for any palaeoclimatological or other environmental interpretations.More detailed explanation is available in our earlier papers (Kaljo et al. 1997;Martma et al. 2005).In the studied outcrop sections the δ 18 O values (Table 3) vary from -3.2 to -6.9‰ (in a few samples up to -7.5‰), being close to the oxygen isotopic values in Silurian brachiopods and marine carbonates (Samtleben et al. 1996;Azmy et al. 1998).On the other hand, these values indicate that corresponding rocks were not seriously affected by later diagenesis.However, to some extent the δ 18 O values are influenced by calcite and dolomite contents in the rock (Table 2) due to their different oxygen fractionation factors (Veizer 1983).From the data presented in Table 3 and Figs    3. The second positive δ 13 C excursion begins in the Bagovitsa 120a section at the boundary of the Sursha and Muksha subformations (just above the boundary with the value of 1.6‰).The peak value (4.3‰) is reached thrice in the middle of the latter unit within an interval of 8-9 m (Fig. 4).The curve in this interval is more like a plateau (most of values over 4‰), not a double-peaked excursion as in Vidukle and elsewhere (Martma et al. 2005;Cramer et al. 2006).The δ 13 C values decrease to ca 1‰ at the top of the Muksha Subformation and become even lower in the Ustje Subformation.In the Muksha 33 outcrop the falling limb of the excursion was observed at the base of the section (Fig. 4).4. The next long interval, including the Ustje Subformation of the Bagovitsa Formation, parts of the Konovka (mainly not sampled), Tsviklevtsy, and the lower Rykhta (Grinchuk) formations are characterized by δ 13 C values close to 0 or lower.The negative values of the trend are connected mostly with the Ustje dolomitic rocks (often between -1.5 and -2.0‰).Higher in the section, in the lower part of the Tsviklevtsy Formation (Sokol Subformation) negative values (-0.5 to -1.0‰) are less variable but rise close to zero (-0.3 to + 0.1‰) in its upper part (Bernovo Subformation).A couple of anomalous more negative analyses (-0.8 and -0.6‰) were registered in the very top of the Bernovo Subformation.The values continue to increase also in the Rykhta Formation, still rather modestly in the lower part (Grinchuk Subformation), where δ 13 C values remain between -0.4 and + 0.5‰ (Fig. 5).

The third positive δ 13 C excursion begins in the
Zhvanets 39 section at the very bottom of the Isakovtsy dolomites (belonging to the upper part of the Rykhta Formation) with an initial value of 1.3‰ and continues into the Prigorodok Formation.This is the most prominent shift identified in the Silurian of Podolia so far: δ 13 C increasing steadily through the Isakovtsy Subformation reaches 5.9‰ in the top and the peak value of 6.6‰ in the bottom of the Prigorodok Formation (Fig. 5).Upwards in the sequence the falling limb of the excursion follows at a rate of ca 1‰ per 1 m (see Table 3, e.g.sample Zhvanets 39-17 shows 1.7‰ 4 m higher).In addition to Zhvanets 39, the main section studied, this excursion was identified also in two nearby outcrops (at Ataki and Braga villages, Fig. 5).In the latter section a maximum value of 6.9‰ was reached already in the topmost Isakovtsy Subformation and after a short plateau-like trend (5.9-6.2‰) in the lowermost Prigorodok Formation the values decrease rather quickly.Small differences between the characters of the trends observed in Zhvanets and Braga need some additional checking, but a curve compiled using mean values for the above three sections (Fig. 6) demonstrates rather smooth and regularly rising and falling limbs of the excursion.Our sampling programme did not allow tracing a continuation of the δ 13 C curve into the Pridoli.This gap in studies will be improved during the next stage of the project.

DISCUSSION AND STRATIGRAPHICAL IMPLICATIONS
The locations of the above excursions were identified in terms of local lithostratigraphy, which clearly fixes the position of an excursion in the sections.Our next task is to date these excursions based on the international stratigraphical standard classification.Figure 2 demonstrates different views on the time correlation of the Podolian Silurian with sequences elsewhere.One reason for the appearance of discrepancies was the correlation of shallow-water Podolian sections containing shelly faunas with deep-water graptolitic sections in the northwestern Ukraine (Volynia) and southwestern Belarus.Because of the lack of graptolites in Podolia this correlation was based mainly on the distribution of 29 metabentonite beds (Tsegelnyuk 1980;Tsegelnyuk et al. 1983).Volcanic ash beds are potentially a good correlation tool but, as shown by Kiipli & Kallaste (2002), a special integrated study of the geochemical properties of bentonites and detailed biostratigraphical control of samples is needed.
In the 1970s-1980s these aspects were not sufficiently studied and later results (Kiipli et al. 2000) gave rise to certain doubts, which are reflected in Figs 2 and 6.The aim of this project was not to discuss in detail reasons for former contradictory results but to produce new data that can generate independent evidence for geological correlation.Carbon isotope correlation has offered a series of successful examples (e.g.Saltzman 2001;Brenchley et al. 2003;Munnecke et al. 2003), and this encourages us to present here our results as follows.In order to make reading of the discussion easier, we add here a composite δ 13 C trend of Podolia correlated with an East Baltic curve (Fig. 6).

The early Sheinwoodian excursion
The first positive δ 13 C excursion identified by us in the Restevo and Demshin subformations (Figs 4 and 6) and earlier by Azmy et al. (1998) is clearly the same global carbon isotopic shift that is most widely known from many parts of the world.We prefer to name it the early Sheinwoodian excursion in order to be as exact as possible and to avoid mixing different kinds of terminology (e.g.Ireviken).This excursion was first described in the lower Wenlock localities of Gotland

Goloskov
Ustje Goloskov the same and also the stratigraphical position of the respective beds fits well with a general understanding of timing of this early Sheinwoodian excursion.Kaljo & Martma (2006) discussed the correlation of the shift with graptolite biozones and showed that increase in the δ 13 C values begins within the Cyrtograptus murchisoni Biozone, the peak occurs in the Monograptus riccartonensis or Streptograptus antennularius Biozone, and the falling limb ends in the M. flexilis Biozone.This definition, however, contains no information about the isotope trend through the Cyrtograptus centrifugus Biozone, missing in many sections.These refinements may help to specify the position of the Restevo and Demshin subformations in the correlation chart, but an uncertainty, caused by distribution problems of Cyrtograptus centrifugus, remains.Tsegelnyuk (1976) reported two important graptolites, Cyrtograptus murchisoni and C. murchisoni bohemicus, from the Restevo Subformation (2-2.2 m from the bottom) at Kitaigorod Village, which places the beginning of the excursion in the Kitaigorod section unambiguously into the C. murchisoni Biozone.Another question is how close it is to the lower boundary of the Wenlock, defined by the appearance of C. centrifugus.Both species are known also in the Baltic, and recently Loydell & Jeppsson (2006) identified C. bohemicus on Gotland from the unit c of the Upper Visby Formation at Ireviken.Still, this level is clearly higher than occurrences in the East Baltic, and especially in Podolia.We suppose that the missing C. centrifugus interval of the δ 13 C excursion in the Restevo Subformation is a short one and therefore the shift begins very shortly above the lower boundary of the Wenlock.Loydell & Frýda (2007) have recently shown that in the Banwy River section, Wales, the early Sheinwoodian excursion commences in the upper C. murchisoni Biozone; the C. centrifugus Biozone is characterized by generally lower δ 13 C values.These new data fix rather clearly relationships of graptolite biozones and beginning of the early Sheinwoodian δ 13 C excursion.Still, debates about the position of the Llandovery-Wenlock boundary is not concluded yet and the suggested (Aldridge et al. 1993) Datum 2 of the Ireviken Event as a correlative level for the boundary GSSP needs to be checked for a possible gap (Loydell et al. 2003) and perhaps some additional criteria should also be added (Männik 2007).Meantime we use the official IUGS terminology.

The late Homerian excursion with or without an extension into the early Ludlow
The second positive δ 13 C excursion identified in the Muksha Subformation (Figs 4 and 6) is more complicated for firm correlation due to global and local stratigraphic uncertainties.In global sense, this shift is not so widely known as the previous one and different views have been expressed about the shape of the curve and its dating.Recent papers by Cramer et al. (2006) and Kaljo & Martma (2006) demonstrate practically identical double-peaked excursions, the former in Nevada and Tennessee in North America and the latter in Estonia, Latvia, and Lithuania in the East Baltic.Both groups of authors conclude that the excursion begins within the Gothograptus nassa-Pristiograptus dubius-P.parvus interval above the last occurrence datum of Monograptus flemingii and ends in the Colonograptus deubeli (Tennessee) or C. ludensis Biozone (Baltic).Besides this conclusion, Cramer et al. (2006) noted that Berry & Murphy (1975) found several Ludlow graptolites (Neodiversograptus nilssoni, Saetograptus colonus, Bohemograptus bohemicus) in the Simpson Park I section, Nevada, in the lower part of the lime mudstonewackestone bed of the Roberts Mountain Formation.If the graptolite sampling data are correctly interpreted, then the above graptolites date the beds with the second peak of the Mulde excursion (terminology of Cramer et al. 2006) as the very beginning of the Ludlow.They discussed different possibilities, but did not find any satisfactory explanation.This finding is in strong contradiction also with Baltic data, especially with those from graptolitic sequences (Priekule and Ventspils in Latvia, Kaljo et al. 1997Kaljo et al. , 1998;;Vidukle in Lithuania, Martma et al. 2005), where the N. nilssoni Biozone lies above the double-peaked late Homerian excursion bed.The δ 13 C trend in the early Ludlow of the Baltic shows dominantly negative values, and the same was documented by Corfield et al. (1992) for Britain.
The above discussion about dating the second peak in the Simpson Park, Nevada, is essential in the context of our study in Podolia because, as outlined in the introduction and shown in Fig. 2, Tsegelnyuk et al. (1983) correlated the Bagovitsa Formation with the lowermost Ludlow.Considering the carbon isotope data from the East Baltic (Kaljo et al. 1997(Kaljo et al. , 1998;;Martma et al. 2005), Gotland (Samtleben et al. 2000), and Poland (Porębska et al. 2004), this correlation is clearly unlikely for the relevant part of the Muksha Subformation.At the same time, negative values revealed in the Ustje Subformation fit well with those from the lower Ludlow of the Baltic area sensu lato and elsewhere (op.cit.above).On the basis of those and also some biostratigraphical data (not discussed here in detail, but see Koren et al. 1989;Gritsenko et al. 1999 and references therein), we prefer to correlate the Muksha Subformation with the uppermost Wenlock.Besides isotope data, most trustworthy information is provided by ostracodes (Abushik 1983; a few noted in Fig. 6) and also some other shelly fauna groups (Gritsenko et al. 1999), but their interpretation is a much more complicated task.We leave the Ustje Subformation at the bottom of the Ludlow, and this means that the series boundary lies within the Bagovitsa Formation.The situation is very similar to that with the Rootsiküla Regional Stage of Estonia, whose upper part was recently transferred, based on the conodont argumentation, to the Ludlow Paadla Regional Stage (Viira & Einasto 2003).It might be added that the dolomitic rocks of the upper part of the Rootsiküla Stage have often been correlated with the Ustje Subformation (Abushik 1983;Kaljo 1987).
However, some doubts remain in Podolia and we very much hope that the real stratigraphical value of graptolite occurrences in the Simpson Park outcrop, discussed in Cramer et al. (2006), will be checked carefully soon.Noble et al. (2005) demonstrate several interesting Wenlock carbon isotope curves from the Canadian Arctic, which show a double-peaked δ 13 C org excursion in the upper Homerian fitting well with Baltic ones.Yet, the δ 13 C carb curve they present is very variable, partly deeply negative (-2‰), and it is hard to see any similarities with the Baltic late Homerian trend.Having in mind the above discussion about the Simpson Park section, we note that Noble et al. (2005) report a low δ 13 C org positive shift (to ca -30‰) within the Lobograptus progenitor Biozone in the Twilight Creek section, Bathurst Island.The shift occurs ca 12 m above the lower boundary of the Ludlow and is not visible in the δ 13 C carb curve.We think that the above C org shift may suggest one more possible explanation of the Simpson Park case.

The middle Ludfordian excursion
The excursion is well recognizable thanks to very high values reported from different localities in beds close to the top of the Ludlow.The highest δ 13 C values are known from Australia (12-13‰, Andrew et al. 1994), Scania of Sweden (11‰, Wigforss-Lange 1999), Gotland, Lithuania, and the Prague Basin (8‰, Samtleben et al. 1996Samtleben et al. , 2000;;Martma et al. 2005;Lehnert et al. 2007b); in several localities the values remain below 6‰ (Saltzman 2001;Lehnert et al. 2007b).There has been some discussion about the dating of this excursion (see for an overview Martma et al. 2005;Kaljo & Martma 2006), but researchers more or less agree that the excursion begins within or above the Polygnathoides siluricus conodont and Neocucullograptus kozlowskii graptolite biozones.The end of the excursion is less clear, being somewhere below the first appearance datum of the conodont Ozarkodina crispa (Kaljo & Martma 2006;Lehnert et al. 2007b) or within this zone (Samtleben et al. 2000) depending on how to interpret the position of the Sundre Beds.The peak shift is usually correlated with the N. kozlowskii Biozone (e.g.Azmy et al. 1998;Saltzman 2001) or higher, embracing also the topmost Ludfordian (Jeppsson et al. 2005).The latter correlation was opposed by Kaljo & Martma (2006).On the basis of the Kosov section, Lehnert et al. (2007b) showed that peak values occur above the last occurrences of N. kozlowskii, but this could be more an artefact caused by a facies change.In the Prague Basin the main peak was recorded in the beds with the Ananaspis fecunda-Cyrtia postera community.Monograptus latilobus has been found above these beds, but well below the upper boundary of the Ludlow.The M. latilobus graptolite Biozone occurs in the drill core sections of eastern Poland at the very beginning of the upper Ludfordian (Urbanek 1997).
The Isakovtsy Subformation and Prigorodok Formation, accommodating the third δ 13 C excursion in Podolia (Figs 5 and 6), have been classified and placed in the stratigraphical scheme rather differently.According to Nikiforova & Predtechensky (1972), these form one unit, the "Isakovtsy Beds", at the bottom of the Skalian Stage (Fig. 2).Tsegelnyuk et al. (1983) introduced two separate units -the Isakovtsy Subformation in the Malinovtsy Series and the Prigorodok Formation in the Rukshin Series and a gap between them (Fig. 2).The age of the Prigorodok Formation is not clearly stated but according to the correlation charts in their paper, it seems to be at the very beginning of the Pridoli or just below it.Kaljo (1987) placed the Prigorodok Formation again into the Skalian Stage, but correlated the Isakovtsy and Prigorodok formations with the upper Ludlow.Koren et al. (1989) dropped both into the topmost Malinovtsy Series (or Stage?) and upper Ludfordian.Drygant (1994) summarized his scheme in sense of time in the same way -he considered Isakovtsy and Prigorodok as separate formations within the Skalian, but belonging in the topmost Ludlow at the level of the Ozarkodina crispa conodont Biozone.
This mess of different views can be clarified using carbon isotope stratigraphy.We refer here to our recent papers (Martma et al. 2005;Kaljo & Martma 2006), where our understanding about the dating of the mid-Ludfordian excursion is demonstrated.We have analysed mainly East Baltic and Gotland δ 13 C data (Samtleben et al. 1996(Samtleben et al. , 2000)), but also biostratigraphy.On the basis of these views we conclude that in Podolia analogically the excursion in the Isakovtsy and Prigorodok formations should be of mid-to early late Ludfordian age, but the upper part of the latter unit reaches the top of the Ludlow.This conclusion is consistent with views by Drygant (1994) but we do not know limits of his O. crispa Biozone.
The gap between the Isakovtsy and Prigorodok formations is not clearly seen in the shape of the carbon isotope trend (Figs 5 and 6), but it may exist like in Kosov quarry in the Prague Basin (Lehnert et al. 2007b).The gaps connected with sea level lowering during a carbon isotope event are widely known phenomena and have been used as an indicator of a short-lived glacial event in the late Ludlow (Lehnert et al. 2007a), but the idea that these support a glaciation seems to be premature.

CONCLUSIONS
1. Three positive δ 13 C excursions were identified, based on the analyses of 149 whole-rock samples.The first excursion in the Kitaigorod 30 section, beginning in the Restevo Subformation, reaches its peak value of 4.0‰ in the very bottom of the Demshin Subformation.The second one reaches the peak value of 4.3‰ in the middle of the Muksha Subformation.The third shift is the most prominent one identified in the Silurian of Podolia: steady increase in δ 13 C values through the Isakovtsy Subformation, reaching 5.9-6.9‰ in the top of the unit.
In Zhvanets the peak value of 6.6‰ is reached in the bottom of the Prigorodok Formation.2. In the "middle" Silurian three major positive global excursions have been identified as follows: in the early Sheinwoodian (early Wenlock), late Homerian (late Wenlock), and middle Ludfordian (late Ludlow).
The general character of the carbon isotope trend and stratigraphical positions of excursions show that the same set of global shifts was established by the present study in Podolia.3. Different aspects of these δ 13 C excursions confirm their utility for the improvement of regional and global correlation of Wenlock and Ludlow sections of Podolia.However, full application of results in regional stratigraphy depends on several details that need additional study.

Table 1 .
Geochemical data based on XRF analyses of the Silurian rocks of Podolia

Table 3 .
Carbon and oxygen isotope data from the Silurian rocks of Podolia