ARTICLE IN PRESS Quaternary International 178 (2008) 276–287 Ordovician K-bentonites: Issues in interpreting and correlating ancient tephras Warren D. Huff Department of Geology, University of Cincinnati, 2100 Clifton Avenue, Cincinnati, OH 45221, USA Available online 23 May 2007 Abstract Prominent Upper Ordovician K-bentonites include the Deicke and Millbrig beds in the Late Ordovician of eastern North America and the Kinnekulle K-bentonite in the uppermost Sandbian of Baltoscandia. These beds are thought to represent some of the largest ash fall deposits known in the Phanerozoic record. This report presents a comprehensive study of biotite compositions from the three beds covering a far more extensive geographic range than has previously been described. These data show that the Kinnekulle and Millbrig are multiple event ash beds, some parts of which are indistinguishable from one another. Tectonomagmatic discrimination diagrams combined with Mg number data indicate that the Deicke–Millbrig–Kinnekulle sequence represents the progressive transformation from calc-alkaline to peraluminous magmatic sources, consistent with a model of progressively evolving magmatism during the closure of the Iapetus Ocean. It is concluded that the Millbrig and Kinnekulle beds are coeval and represent primarily simultaneous episodes of explosive volcanism although it cannot be excluded that some portions of both beds were produced by the same event. r 2007 Elsevier Ltd and INQUA. All rights reserved. 1. Introduction Explosive volcanism plays a fundamental role in the exchange of material and energy from the Earth’s interior to the hydrosphere and atmosphere as well as the Earth’s surface, and major explosive events are thought to be capable of perturbing the Earth’s climate on time scales of 1–5 years, generally resulting in net global surface cooling (Rampino et al., 1988). Historically documented eruptions do not, however, represent the full range of intensity and magnitude of all explosive eruptions in geologic history. Deposits in the geologic record provide compelling evidence for eruptions that were orders of magnitude larger than ones witnessed by mankind. Contemporary explosive eruptions may have lithofacies associations of near-vent subaerial phenomena that typically include pyroclastic surge deposits, thin welded tuff beds, various lava flow morphologies, abundant erosional unconformities, and fluviatile and laharic facies. While volcanotectonic subsidence might aid in the preservation of such deposits, they generally are not well known in the geologic Tel.: +1 513 556 3731; fax: +1 513 556 6931. E-mail address: warren.huff@uc.edu record. Ash layers in marine sediments, on the other hand, are the best geologic record of explosive volcanism, and many of the Phanerozoic volcanic ash layers represent enormous explosive eruptions that dispersed fine ash and aerosols through the atmosphere over tens of thousands of km2. While evidence of very large magnitude events can be found throughout the Phanerozoic, this report will focus on Ordovician events recorded in both North America and Europe. The Ordovician record of explosive volcanism consists of examples of both near-vent pyroclastic flows and ignimbrites and distal sequences of altered fallout tephras known as K-bentonites. Questions frequently arise as to whether a particular clay-rich bed might be an altered volcanic ash fall in the form of a bentonite or K-bentonite. These beds are often datable using fission track and U/Pb dating of zircons, K/Ar, and Ar/Ar of amphibole, biotite and/or sanidine. Due to their unique composition, they provide an indispensable tool when correlating sections. The criteria for recognizing such beds are varied, but fall into two broad categories, field criteria and laboratory criteria. Ideally, one would want both, but often that is not possible. However, there are key features to look for in each case that can aid in reliable identification. 1040-6182/$ - see front matter r 2007 Elsevier Ltd and INQUA. All rights reserved. doi:10.1016/j.quaint.2007.04.007 ARTICLE IN PRESS W.D. Huff / Quaternary International 178 (2008) 276–287 1.1. Field criteria K-bentonites can be different colors when wet (blue, green, red, yellow), but are characteristically yellow when weathered. Due to their clay-rich nature, they will feel slippery and waxy when wet. Some K-bentonites contain euhedral to anhedral volcanogenic biotite, quartz, feldspar, amphibole, zircon and apatite, and less commonly, clinopyroxene, magnetite and garnet. The typical appearance of a K-bentonite bed in outcrop (Fig. 1) is that of a fine-grained, clay-rich band ranging between 1 mm and 2 m 277 in thickness that has been deformed by static loads from the enclosing siliciclastic or carbonate sequence. Accelerated weathering of K-bentonites causes them to be recessed into the outcrop face. For thicker K-bentonites, there is sometimes a zone of nodular or bedded chert in the adjacent strata at both the base and the top of the bed. 1.2. Laboratory criteria Most bentonites and K-bentonites are smectite- or illite/ smectite-rich, although some may contain a considerable Fig. 1. (a) Core showing a K-bentonite (arrow) in a carbonate section, (b) Roadcut at Gladeville, TN, with Deicke (D) and Millbrig (M) separated by about 4 m of Eggleston Fm. limestone, (c) Deicke 3.2 m below the Millbrig in the Decorah Fm. at Minke Hollow, MO, and (d) Deicke in the Eggleston Fm. at Carthage, TN. ARTICLE IN PRESS 278 W.D. Huff / Quaternary International 178 (2008) 276–287 amount of kaolinite, and those that have undergone lowgrade metamorphism may be dominated by R3 I/S and/or sericite plus chlorite/smectite (corrensite) and/or chlorite. Initial steps should begin with separation and XRD analysis of the clay fraction as a preliminary means of identification. Bentonites often contain volcanic phenocrysts and, for younger beds, even remnants of unaltered volcanic glass, and their presence provides confirming evidence of a volcanogenic origin. Wet sieving followed by conventional heavy mineral separation techniques is usually sufficient to isolate these different components. Study of the non-clay fraction using high-quality optical microscopy or SEM is satisfactory to determine the type and characteristics of crystals in the sample. Thin section study is also useful when more quantitative data are desired. 2. Ordovician K-bentonites K-bentonites have been recognized since the early decades of the 1900s as very useful stratigraphic tools (Nelson, 1921, 1922; Kay, 1935). Since they were deposited over large areas in a short time period, they provide precise event-stratigraphic correlation besides the biogeographical, paleoecological and sedimentological studies on both local and regional scales. They also have tectonomagmatic and paleogeographic significance because they preserve mineralogical and geochemical evidence of their origin. As with every Phanerozoic system, Ordovician successions contain a number of K-bentonites representing episodes of explosive volcanism, most commonly associated with large caldera-forming tectonic events. Fig. 2 shows the general stratigraphic and geographic distribution of beds. Numerous beds have been reported from North and South America, Asia and Europe. The Ordovician successions of North America are known to contain nearly 100 K-bentonite beds, one or more of which are distributed over 1.5  106 km2 (Kolata et al., 1996) and around 150 beds are recorded in Baltoscandia (Bergström et al., 1995). The first report of an Ordovician K-bentonite in North America was made by Ulrich (1888), who described a thick bed of clay in the upper part of what is now known as the Tyrone Limestone, near High Bridge, Kentucky. Subsequent work by Nelson (1921, 1922) showed that the bed was volcanogenic in origin and that it could be correlated into Tennessee and Alabama. After the 1930s, K-bentonite beds of Ordovician age began to be reported from localities throughout eastern North America (Kay, 1935; Weaver, 1953; Brun and Chagnon, 1979; Huff and Kolata, 1990; Kolata et al., 1996). Two prominent K-bentonites occur throughout the eastern Midcontinent, the early Chatfieldian Millbrig K-bentonite or T-4 bed and the late Turinian Deicke K-bentonite or T-3 bed (Wilson, 1949). The Deicke and Millbrig K-bentonites are the two most widespread and prominent of the many ash beds in the Upper Mohawkian/Champlainian ( ¼ uppermost Sandbian) strata of eastern North America. In their type areas in the Upper Mississippi Valley, the Deicke, Millbrig and other Mohawkian K-bentonites are only a few centimeters thick and are noticeably lacking in non-clay minerals (Willman and Kolata, 1978; Kolata et al., 1986). Using geophysical logs of wells, Huff and Kolata (1990) correlated the Deicke and Millbrig from that region with the T-3 (Pencil Cave) and T-4 (Mud Cave) K-bentonites of the Cincinnati Arch and the western Valley and Ridge. The T-3 and T-4 beds are well known (Wilson, 1949; Milici, Fig. 2. Global stratigraphic distribution of Ordovician K-bentonites. ARTICLE IN PRESS W.D. Huff / Quaternary International 178 (2008) 276–287 1969; Milici and Smith, 1969; Drahovzal and Neathery, 1971; Chowns and McKinney, 1980), and are relatively thick and coarse grained. In Mohawkian strata of the southern Appalachian basin, of which the Rocklandian K-bentonites are a part, carbonate rocks predominate in most of the area, with clastic rocks confined to the easternmost outcrop belts of the Valley and Ridge (Rodgers, 1953; Rader, 1982; Chowns and Carter, 1983; Benson, 1986). The Deicke and Millbrig occur in the Tyrone Limestone of the uppermost high Bridge Group in central Kentucky; the Carters Limestone of the uppermost Stones River Group in central and southeast Tennessee and northwest Georgia; the upper Moccasin and lower Eggleston Formations in southwest Virginia, southeast West Virginia, and northeast Tennessee; and the upper Stones River Group (undivided) and Chickamauga Limestone in Alabama (Kolata et al., 1996) (Fig. 3). The Stones River and Nashville Groups are defined and subdivided in the type area of the Chickamauga limestone Group (Milici and Smith, 1969). In Kentucky, the Tyrone Limestone, the youngest formation of the High Bridge Group is 18–36 m thick (Wolcott et al., 1972) and is sublithographic, dense, creamcolored limestone, breaking with conchoidal fracture and with small facets of coarsely crystalline calcite. Several K-bentonites occur in the Tyrone Limestone. The Deicke 279 K-bentonite occurs 4.5–7.5 m below the top of the Tyrone, and is the principal and the most consistently developed bed. It ranges in thickness from 10 to 60 cm (McFarlan, 1943). The second most persistent and widespread K-bentonite horizon is the Millbrig K-bentonite, which occurs at the top of the Tyrone at the contact with the overlying Lexington Limestone and is more coarsely textured than the Deicke. A third, unnamed K-bentonite has been observed in several localities at about 24 m below the top of the Tyrone, and is generally about 3 cm or less thick (Cressman, 1973). In central Tennessee, the Carters Limestone is the uppermost unit of the Stones River Group, and is divided into upper thin-bedded and lower massive-bedded units separated by the Deicke. The Upper member of the Carters Limestone occurs between the Deicke and the overlying Curdsville member of the Hermitage Formation. It is characterized by dense, fine-grained, dove-colored lithology and thin bedding, which contains partings of silt and clay. Throughout a large part of the central basin its average thickness is 4 m. At numerous localities within the eastern part of the central basin the Millbrig occurs near the top of this member (Wilson, 1949). In Alabama, the western facies of the Middle and Upper Ordovician lies west and north of the Helena thrust fault (Drahovzal and Neathery, 1971). The Stones River Group Fig. 3. Stratigraphic distribution of the Deicke and Millbrig K-bentonites in the southern Appalachian basin (after Ryder et al., 1997). Note: Pencil Cave and Mud Cave are drillers’ names previously applied to the Deicke and Millbrig K-bentonites, respectively. ARTICLE IN PRESS 280 W.D. Huff / Quaternary International 178 (2008) 276–287 and the Nashville Group constitute the Chickamauga Limestone Group of the Middle Ordovician. Several K-bentonite horizons have been reported (Drahovzal and Neathery, 1971) which can be correlated with the Tennessee K-bentonites. In this area, the Deicke and Millbrig K-bentonites have been traced from Tennessee (Haynes, 1994). The Deicke K-bentonite is often 1–1.2 m thick in Alabama (Drahovzal and Neathery, 1971) and its color is yellow–gray to light yellow–green. It is characteristically underlain by a chert layer. The upper Millbrig K-bentonite may be as much as 1 m thick, and is characterized by abundant biotite flakes and chert layers both above and below. The clastic rocks along the eastern Valley and Ridge in which the K-bentonites occur are assigned to the Blount Group (Rodgers, 1953), a great clastic wedge that extends along strike from Alabama to Virginia and persists laterally across only two or three thrust sheets in the southern Valley and Ridge. Both the Deicke and Millbrig occur in the Colvin Mountain Sandstone in Alabama, but only the Deicke has been found in the Bays Formation in southeast Tennessee. In northeast Tennessee and southwest Virginia, only the Millbrig is present in the Bays Formation (Haynes, 1994). The Kinnekulle K-bentonite is the thickest and most widespread among the many K-bentonites in the Ordovician of Baltoscandia. It is generally present in stratigraphically continuous Johvian–Keilan sections from southeastern Norway to Scandia and Bornholm and the East Baltic (Fig. 4). It can be traced for about 1000 km across Baltoscandia. The thickness ranges from more than 2 m in the Fylla Mosse drill core in Östergotland, Sweden to o5 cm in southeastern Estonia, central Latvia and western Russia. Bergström and Nilsson (1974) reported more than 150 K-bentonite beds in the Middle Ordovician shale succession of the Koangen drill core in south-central Scania, but the majority of these are thin and apparently not represented in the shallow-water carbonate-dominated successions in much of south-central Baltoscandia. The type locality of the Kinnekulle bed is the Mossen section on the eastern slope of Kinnekulle. This abandoned quarry is situated 2.0 km north of Osterplana church and 0.2 km west of Mossen farm (Thorslund, 1948; Jaanusson, 1964). At Mossen, it has a total thickness of 180 cm, but Fig. 4. Stratigraphic position of the known Ordovician K-bentonites in North American and northern Europe. M, Millbrig, D, Deicke and K, Kinnekulle. ARTICLE IN PRESS W.D. Huff / Quaternary International 178 (2008) 276–287 due to the slumping of the overlying rocks of the Skagen Formation only the uppermost part of the bed is exposed (Bergström et al., 1995). The possibility of a common source for these North American and Baltoscandian ash beds was suggested by Huff et al. (1992). They presented biostratigraphical, geochemical, isotopic and paleogeographic data that indicated that the Millbrig K-bentonite, a widespread isochron and marker bed in North America, and the Kinnekulle bed in Baltoscandia are coeval and probably derived from the same eruptive event. Saltzman et al. (2003) summarized the correlation of the Guttenberg 13C Isotope Excursion (GICE) between eastern North America and northern Europe, with a closely similar stratigraphic position to both the Millbrig and Kinnekulle K-bentonites. This excursion occurs in the same conodont and graptolite zones as both the Millbrig and Kinnekulle beds and shows clearly that these beds are coeval deposits. Haynes et al. (1995) suggested that the proposed transAtlantic correlation of the Millbrig and Kinnekulle beds is, at best, permissive. They argued that the utility of the discriminant function analysis of whole rock compositions, while perhaps appropriate for more localized regional correlations, is not valid for large-scale regional or global correlations. They maintained that uncertainty results from the variable mobility of several major and certain trace elements during diagenesis that might result in regional shifts in bulk composition. Haynes et al. (1995) studied the compositions of primary biotite phenocrysts and concluded that they are more reliable as specific bed indicators than bulk rock composition even though long-distance correlation based on phenocryst compositions can be still suspect for various reasons. They reported a compositional difference between Kinnekulle and Millbrig biotites with respect to their FeO, MgO, Al2O3, MnO and TiO2 content, and further suggested that these variations represent separate magmatic sources. However, Haynes et al. (1995) used data from only one Millbrig site in North America and one Kinnekulle site in Baltoscandia, and they did not evaluate lateral variation as well as within-bed variation in biotite compositions. Furthermore, they culled some of the analyses, leaving a data set representative of both the modal and average composition of biotite in each sample set. Hence, the consanguinity of these ash beds, the largest recorded in the Paleozoic, has remained uncertain. 3. Isotopic ages One method of trying to resolve the question of the equivalence of the Millbrig and Kinnekulle beds might be to compare isotopic ages. Fig. 5 is a graphic compilation of published ages for the Deicke, Millbrig and Kinnekulle beds since 1960 (Table 1). The methods include fission track, K–Ar, Ar–Ar, Rb–Sr and both batch and single grain U–Pb isotopic dating. Stratigraphically, the Deicke lies below the Millbrig, but this cannot be demonstrated on 281 Fig. 5. Published isotopic ages for the Deicke, Millbrig and Kinnekulle K-bentonite beds since 1960. Computed standard deviations are not shown. Methods include fission track, K–Ar, Ar–Ar, Rb–Sr and U–Pb. Data are taken from various literature sources (Adams et al., 1960; Byström-Asklund et al., 1961; Harland et al., 1964; Ghosh, 1972; Ross et al., 1981, 1982; Williams et al., 1982; Baadsgaard and Lerbekmo, 1983; Kunk and Sutter, 1984; Kunk et al., 1985, 1986, 1988; Samson et al., 1989; Tucker et al., 1990; Compston and Williams, 1992; Tucker, 1992; Tucker and McKerrow, 1995; Toulkeridis et al., 1998; Min et al., 2001). the basis of published isotopic age dates alone. Similarly the Kinnekulle and Millbrig beds occur in the same biostratigraphic zones, which makes them broadly coeval at the very least, however, isotopic ages also do not resolve the age relations any further. Part of the problem surely lies in the dating methodologies themselves and their inherent constraints on precision and accuracy (see Renne et al., 2000; Schmitz et al., 2003) notwithstanding substantial improvements in equipment and procedures over the past four decades. But the available data do not show any progressive or systematic change in measured ages over that time span that would suggest convergence upon a ‘‘true’’ value for each bed. Certainly another factor contributing to the apparent age variance is the complex nature of the Millbrig and Kinnekulle beds themselves (Huff et al., 1999). Close examination in the field shows that both beds consist of several internally graded units, suggesting that each bed represents the cumulative deposition of multiple ash falls in environments characterized by low background sedimentation rates. This aspect was examined in some detail by Kolata et al. (1998) and Haynes (1994) for the Millbrig and by Huff et al. (1999) for the Kinnekulle, all of whom showed systematic mineralogical and grain size variation within individual subunits. A careful examination of apatite compositions from the Deicke and Millbrig (Fig. 6) reveals two distinct populations in the Millbrig from the lower and middle portions of the bed. In the Deicke, apatite occurs only in the basal portion. Given higher rates of sediment accumulation, it is conceivable that these units would be preserved as a series of closely spaced but separate beds (Bergström et al., 1997; Huff et al., 1999). ARTICLE IN PRESS 282 W.D. Huff / Quaternary International 178 (2008) 276–287 Table 1 Isotopic ages for the Deicke, Millbrig and Kinnekulle K-bentonites published since 1960 Bed Formation Location Age (Ma) Method Reference Deicke Deicke Deicke Deicke Deicke Deicke Deicke Deicke Deicke Deicke Deicke Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekullle Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Millbrig Tyrone – – Plattin Stones River Lexington – Carters Tyrone – Carters – Dalby Dalby Dalby Dalby Dalby – – – – – – – Dalby Carters Little Oak Decorah – Stones River Shakertown Decorah – Tyrone – Stones River – – Kentucky – – Missouri Alabama Kentucky – Tennessee Kentucky – Tennessee Hullegard Kinnekulle Kinnekulle Kinnekulle Kinnekulle Kinnekulle – – – – – – – Kinnekulle Tennessee Alabama Missouri – Alabama Kentucky Missouri – Kentucky – Alabama – – 447715 462715 438715 450710 454.273 449.872.3 454.570.5 453.873.6 457.773 454.570.5 457.171 455.873.3 45474 444720 44673 455.673.7 45573 45778 456.871 44573 45076 454 45274 456.971.8 44771 453710 453 454712 456712 453.373 448.072.0 453.771.8 453.171.3 453.373 453.171.3 455.779.8 454.172.1 453.171.3 Fission track Fission track Fission track Fission track Ar–Ar Ar–Ar U–Pb Ar–Ar Ar–Ar U–Pb U–Pb Ar–Ar Ar–Ar K–Ar K–Ar Ar–Ar Ar–Ar K–Ar U–Pb Rb–Sr K–Ar K–Ar SHRIMP U–Pb Rb–Sr U–Pb K–Ar Fission track Fission track Ar–Ar Ar–Ar U–Pb U–Pb Ar–Ar U–Pb K–Ar U–Pb U–Pb Ross et al. (1982) Ross et al. (1982) Ross et al. (1982) Ross et al. (1982) Kunk and Sutter (1984) Min et al. (2001) Tucker (1992) Kunk and Sutter (1984) Kunk and Sutter (1984) Tucker and McKerrow (1995) Samson et al. (1989) Min et al. (2001) Min et al. (2001) Byström-Asklund et al. (1961) Williams et al. (1982) Kunk et al. (1985) Kunk and Sutter (1984) Harland et al. (1964) Tucker (1992) Williams et al. (1982) Williams et al. (1982) Kunk et al. (1985) Compston and Williams (1992) Tucker and McKerrow (1995) Baadsgaard and Lerbekmo (1983) Adams et al. (1960) Ghosh (1972) Ross et al. (1981) Ross et al. (1981) Kunk and Sutter (1984) Min et al. (2001) Tucker et al. (1990) Tucker (1992) Kunk and Sutter (1984) Tucker and McKerrow (1995) Toulkeridis et al. (1998) Kunk et al. (1985) Tucker (1992) The Middle Ordovician section at Röstånga in Scania (southern Sweden) contains 18 K-bentonite beds ranging from 1 to 67 cm in thickness, and all occur within the D. foliaceus graptolite biozone. At Kinnekulle, 290 km to the north, this interval includes the type section of the Kinnekulle K-bentonite, which is very widespread and has been correlated throughout northern Europe (Bergström et al., 1995). In most sections, the Kinnekulle K-bentonite can be recognized by distinctive geochemical fingerprints, its prominent thickness, and by its biostratigraphic and lithostratigraphic position. However, at Röstånga, whole rock chemistry is inconclusive at identifying which of the 18 beds is the Kinnekulle K-bentonite. Several beds at Röstånga correlate equally well with the Kinnekulle bed (Bergström et al., 1997) and thus argue strongly for the composite nature of what is called the Kinnekulle K-bentonite. The Deicke, on the other hand, appears to be a single event deposit. The regional aspects of ash accumulation on submarine surfaces has been discussed by Kolata et al. (1998) and Ver Straeten (2004). 4. Biotite geochemistry Any comparative chemostratigraphic study of the Deicke, Millbrig and Kinnekulle beds must clearly take into account both vertical and lateral within-bed variations in phenocryst content and phenocryst composition. Single samples cannot be considered fair representations of bed chemistry. This report presents a comprehensive microprobe study of K-bentonite biotite composition covering a far more extensive geographic range for these three Ordovician beds than has previously been reported. The Deicke and Millbrig beds were sampled at 16 localities in five US states and the Kinnekulle bed was sampled at 15 localities in Norway, Sweden, Denmark and Estonia (Fig. 7). Separated biotite grains were mounted in epoxy, ARTICLE IN PRESS W.D. Huff / Quaternary International 178 (2008) 276–287 polished and analyzed by electron beam microprobe operating at 15 kV and 20 nA with a 1 mm beam size. After removing several hundred analyses due to mineral grain alteration, a total of 666 Kinnekulle biotite analyses, 97 Millbrig biotite analyses and 39 Deicke biotite analyses representing 31 separate geographic localities provide the most comprehensive view to date of the nature and extent of internal compositional homogeneity of these widespread Laurentian and Baltoscandian ash beds. 283 Bivariate plots of some major element oxides are compared in Fig. 8, much in the manner used by Haynes et al. (1995). In each case, the Millbrig and Kinnekulle beds are characterized by a central cluster of data points but with significant scatter and overlap between the two and also with the Deicke. Secondary clusters of data points suggest multiple episodes of biotite crystallization, consistent with field evidence pointing to a multiple event history for both the Millbrig and Kinnekulle. The Deicke continues to appear as a single event deposit. Total iron as FeO is an effective discriminator of Deicke biotite, but not of the other two beds. These show partial overlap using FeO and MgO (Fig. 8a and c), and nearly complete overlap using MnO (Fig. 8b). 5. Tectonomagmatic setting Fig. 6. Compositional variation in phenocrystic apatite grains from the Millbrig and Deicke K-bentonites at Gadsden, Alabama. Igneous biotite composition is broadly indicative of the magma from which it crystallized and thus it serves as a general petrogenetic indicator that can provide insight into its tectonomagmatic setting. The method of AbdelRahman (1994) was used, which employs the ratios between MgO, FeO as total iron, and Al2O3 in biotite to discriminate between anorogenic extensional-related alkaline rocks, calc-alkaline I-type suites and peraluminous suites including S-type rocks. Anorogenic rocks tend to be high in Fe and low in Mg, a trend that reverses itself in calc-alkaline suites. Peraluminous rocks are characterized by an increase in Al compared with both Fe and Mg with the substitution occurring primarily in octahedrally coordinated sites. When these biotite data are plotted on the FeO–Al2O3 discrimination diagram of Abdel-Rahman (1994), the Deicke samples plot clearly in the calc-alkaline field (Fig. 9). Both the Millbrig and Kinnekulle beds show Fig. 7. Location map for analyzed samples. Location details are given in Bergström et al. (1995) and Kolata et al. (1996). ARTICLE IN PRESS 284 W.D. Huff / Quaternary International 178 (2008) 276–287 Fig. 9. Magmatic discrimination diagram based on compositional variation in biotite, after Abdel-Rahman (1994). The symbols are the same as in Fig. 8. Fig. 8. Bivariate plots of biotite major elements. a bimodal distribution. Millbrig samples plot primarily in the calc-alkaline field with a subset overlapping with Deicke samples. The majority of Millbrig samples are more enriched in Fe than the Deicke samples and fall at the extreme edge of the calc-alkaline field. The Kinnekulle samples cluster primarily in the peraluminous field with a subset apparently less Fe-rich than the Deicke samples and plotting to the left of them in the center of the calc-alkaline field. Mg concentrations in these samples show just the opposite trend, suggesting that there is a petrogenetic trend from relatively high-Mg, low-Fe Deicke samples forming in a subduction-related environment to Fe-enrichment in most of the Millbrig to Al enrichment in most of the Kinnekulle samples. This scenario is generally consistent with a model of the tectonic closure of the Iapetus Ocean, which includes the involvement of increasing amounts of Al-rich continental crust in the formation of volcanic magmas. The fact that some Kinnekulle and Millbrig samples have low Fe/Mg ratios suggests those portions may represent an earlier stage in this process, or perhaps even separate source volcanoes. The compelling evidence, however, points to distinct compositional differences between many of the Kinnekulle and Millbrig biotites indicating that most were not sourced from the same eruptive event as proposed by Huff et al. (1992), but rather represent successive events, closely spaced in time, triggered by the progressive closure of the Iapetus Ocean and its associated volcanic arc magmatism. The data also reinforce the concept that the Millbrig and Kinnekulle beds represent multiple eruptive events, some of which are indistinguishable from one another. Fig. 10 shows the bivariate distribution of Al2O3 vs. the Mg number. The Mg number (Mg#) is defined as Mg2+/ (Mg2++Fe2+) and is widely used as an index of fractional crystallization in evolving magmas, with values decreasing from 1.0 as magmas become more highly evolved and thus more Si-rich (Ragland, 1989). Here, the Mg# from biotite data alone is computed and consider it a reasonable proxy for the whole rock value. A plot of Mg# against Al2O3 shows a transition from high-Mg to low-Mg biotites and a corresponding transition from dominantly Deicke through Millbrig to Kinnekulle samples. Some overlap between Deicke and Millbrig and between Millbrig and Kinnekulle samples supports the concept that the three ash beds represent a broad evolutionary pattern of magmatic evolution. As with the distribution pattern in Fig. 8, a subset of Kinnekulle samples has the highest Mg# in the entire collection and thus reinforces the notion that the Kinnekulle represents a multiple event deposit with some ARTICLE IN PRESS W.D. Huff / Quaternary International 178 (2008) 276–287 285 Acknowledgments I am indebted to T. Gerke for assistance in generating a large portion of the microprobe data, and to J. Haynes for sharing additional microprobe data he collected during his post-doctoral tenure at the Smithsonian Institution. This research was funded in part by National Science Foundation Grant INT-9513128. References Fig. 10. Mg number versus Al2O3 in biotite. The symbols are the same as in Fig. 8. portions accumulating earlier than others, and possibly even from different source volcanoes. The Millbrig is similarly seen to have some components that represent earlier evolutionary stages than others and, at least geochemically, overlap with the Deicke field. This finding is consistent with the field observations of the Millbrig which, in the southern Appalachians where it has its maximum thickness, generally is seen to consist of several fining upward subunits with coarse biotite and feldspar at the base of each layer (Haynes, 1994). 6. Conclusions The Deicke K-bentonite appears to be a single event deposit whereas the Millbrig and Kinnekulle beds represent multiple event deposits. The Millbrig in eastern North America and the Kinnekulle in northern Europe both display macroscopic and microscopic evidence of multiple event histories, a characteristic that is only explainable by invoking a history of episodic ash accumulation in areas with essentially no background sedimentation (Kolata et al., 1998; Ver Straeten, 2004). Portions of the Millbrig and Kinnekulle beds have biotites that are compositionally indistinguishable from one another, although the majority of samples analyzed show a clear distinction between the two beds. Tectonomagmatic discrimination diagrams combined with Mg number data indicate that the Deicke– Millbrig–Kinnekulle sequence represents the transformation from calc-alkaline to peraluminous magmatic sources, consistent with a model of progressively evolving magmatism during the closure of the Iapetus Ocean. Published isotopic age dates are inconclusive as to the precise ages of each bed. Thus, it can be concluded that the Millbrig and Kinnekulle beds are coeval and represent separate but simultaneous episodes of explosive volcanism, although it cannot be excluded that parts of these beds were derived from the same eruption(s). Abdel-Rahman, A.-F.M., 1994. Nature of biotites from alkaline, calcalkaline, and peraluminous magmas. Journal of Petrology 35, 525–541. Adams, J.A.S., Osmond, J.K., Edwards, G., Henle, W., 1960. Absolute dating of the Middle Ordovician. Nature 188, 636–638. Baadsgaard, H., Lerbekmo, J.F., 1983. Rb–Sr and U–Pb dating of bentonites. Canadian Journal of Earth Sciences 20, 1281–1290. Benson, D.J., 1986. Stratigraphic setting of the Middle Ordovician of the Alabama Appalachians. In: Benson, D.J., Stock, C.W. (Eds.), Proceedings of the Alabama Geological Society Guidebook, Birmingham, pp. 3–14. Bergström, S.M., Nilsson, R., 1974. Age and correlation of the Middle Ordovician bentonites on Bornholm. Bulletin of the Geological Society of Denmark 23, 27–48. Bergström, S.M., Huff, W.D., Kolata, D.R., Bauert, H., 1995. Nomenclature, stratigraphy, chemical fingerprinting, and areal distribution of some Middle Ordovician K-bentonites in Baltoscandia. Geologiska Föreningens i Stockholm Förhandlingar 117, 1–13. Bergström, S.M., Huff, W.D., Kolata, D.R., Yost, D.A., Hart, C., 1997. A unique Middle Ordovician K-bentonite bed succession at Röstånga, S. Sweden. Geologiska Föreningens i Stockholm Förhandlingar 119, 231–244. Brun, J., Chagnon, A., 1979. Rock stratigraphy and clay mineralogy of volcanic ash beds from the Black River and Trenton Groups (Middle Ordovician) of southern Quebec. Canadian Journal of Earth Sciences 16, 1499–1507. Byström-Asklund, A.M., Baadsgaard, H., Folinsbee, R.E., 1961. K/Ar age of biotite, sanidine, and illite from Middle Ordovician bentonites at Kinnekulle, Sweden. Geologiska Föreningens i Stockholm Förhandlingar 83, 92–96. Chowns, T.M., Carter, B.D., 1983. Middle Ordovician section along Mount Alto Road at southwest end of Horseleg Mountain. In: Chowns, T.M. (Ed.), Proceedings of the Geology of Paleozoic Rocks in the Vicinity of Rome, Georgia. Guidebook for the 18th Annual Field Trip of the Georgia Geological Society, pp. 70–73. Chowns, T.M., McKinney, F.K., 1980. Depositional facies in MiddleUpper Ordovician and Silurian rocks of Alabama and Georgia. In: Frey, R.W. (Ed.), Excursions in Southeastern Geology. Geological Society of American Annual Meeting. American Geological Institute, Falls Church, VA, pp. 323–348. Compston, W., Williams, I.S., 1992. Ion probe ages for the British Ordovician and Silurian stratotypes. In: Webby, B.D., Laurie, J.R. (Eds.), Global Perspectives on Ordovician Geology, Proceedings of the 6th International Symposium on the Ordovician System, Sydney. Balkema, Rotterdam, pp. 59–67. Cressman, E.R., 1973. Lithostratigraphy and depositional environments of the Lexington Limestone (Ordovician) of central Kentucky. United States Geological Survey Professional Paper 768, p. 61. Drahovzal, J.A., Neathery, T.L., 1971. The Middle and Upper Ordovician of the Alabama Appalachians. In Guidebook, 9th Annual Field Trip. Alabama Geological Society, p. 104. Ghosh, P.K., 1972. Use of bentonites and glauconites in potassium 40/ argon 40 dating in Gulf Coast stratigraphy. Unpublished Ph.D. Thesis, Rice University, Houston, TX. ARTICLE IN PRESS 286 W.D. Huff / Quaternary International 178 (2008) 276–287 Harland, W.B., Smith, A.G., Wilcock, B., 1964. The Phanerozoic TimeScale; A Symposium. Geological Society of London Special Publication 1, p. 458. Haynes, J.T., 1994. The Ordovician Deicke and Millbrig K-Bentonite Beds of the Cincinnati Arch and the Southern Valley and Ridge Province. Geological Society of America Special Paper 290, pp. 1–80. Haynes, J.T., Melson, W.G., Kunk, M.J., 1995. Composition of biotite phenocrysts in Ordovician tephras casts doubt on the proposed transAtlantic correlation of the Millbrig K-bentonite (United States) and the Kinnekulle K-bentonite (Sweden). Geology 23, 847–850. Huff, W.D., Kolata, D.R., 1990. Correlation of the Ordovician Deicke and Millbrig K-bentonites between the Mississippi Valley and the southern Appalachians. American Association of Petroleum Geologists Bulletin 74, 1736–1747. Huff, W.D., Bergström, S.M., Kolata, D.R., 1992. Gigantic Ordovician volcanic ash fall in North America and Europe: biological, tectonomagmatic, and event-stratigraphic significance. Geology 20, 875–878. Huff, W.D., Müftüoglu, E., Kolata, D.R., Bergström, S.M., 1999. K-bentonite bed preservation and its event stratigraphic significance. Acta Universitatis Carolinae–Geologica 43, 491–493. Jaanusson, V., 1964. The Viruan (Middle Ordovician) of Kinekulle and northern Billingen, vol. 43. Bulletin of the Geological Institute, University of Uppsala, pp. 1–73. Kay, G.M., 1935. Distribution of Ordovician altered volcanic materials and related clays. Geological Society of America Bulletin 46, 225–244. Kolata, D.R., Frost, J.K., Huff, W.D., 1986. K-bentonites of the Ordovician Decorah subgroup, upper Mississippi valley: correlation by chemical fingerprinting. Illinois State Geological Survey, Champaign-Urbana, IL. Kolata, D.R., Huff, W.D., Bergström, S.M., 1996. Ordovician Kbentonites of eastern North America. Geological Society of America Special Paper 313, pp. 1–84. Kolata, D.R., Huff, W.D., Bergström, S.M., 1998. Nature and regional significance of unconformities associated with the Middle Ordovician Hagan K–bentonite complex in the North American midcontinent. Geological Society of America Bulletin 110, 723–739. Kunk, M.J., Sutter, J.F., 1984. 40Ar/39Ar age spectrum dating of biotite from Middle Ordovician bentonites, eastern North America. In: Bruton, D.L. (Ed.), Aspects of the Ordovician System. Paleontological Contributions, vol. 295. University of Oslo, pp. 11–22. Kunk, M.J., Sutter, J., Obradovich, J.D., Lanphere, M.A., 1985. Age of biostratigraphic horizons within the Ordovician and Silurian systems. In: Snelling, N.J. (Ed.), The Chronology of the Geological Record, vol. 10. British Geological Survey Memoir, pp. 89–92. Kunk, M.J., Bergström, S.M., Sutter, J.F., 1986. A comparison of Middle Ordovician bentonites in Sweden and North America using 40Ar–39Ar age spectrum dating techniques. Terra Cognita 6, 141. Kunk, M.J., Samson, S.D., Roddick, J.C., 1988. 40Ar/39Ar age spectrum dating of biotite and sanidine and U–Pb dating of zircon of midOrdovician bentonites from Sweden: a comparison with results from eastern North America. Abstracts and Program, fifth International Symposium on the Ordovician System, St. John’s, Newfoundland, p. 49. McFarlan, A.C., 1943. Geology of Kentucky. University of Kentucky, Lexington. Milici, R.C., 1969. Middle Ordovician Stratigraphy in central Sequatchie Valley, Tennessee. Southeastern Geology 11, 111–127. Milici, R.C., Smith, J.W., 1969. Stratigraphy of the Chickamauga Supergroup in its type area. Georgia Geological Survey Bulletin 80, 1–35. Min, K., Renne, P.R., Huff, W.D., 2001. 40Ar/39Ar dating of Ordovician K-bentonites in Laurentia and Baltoscandia. Earth and Planetary Science Letters 185 (1–2), 121–134. Nelson, W.A., 1921. Notes on a volcanic ash bed in the Ordovician of Middle Tennessee. Tennessee Division of Geology Bulletin 25, 46–48. Nelson, W.A., 1922. Volcanic ash beds in the Ordovician of Tennessee, Kentucky and Alabama. Geological Society of America Bulletin 33, 605–615. Rader, E.K., 1982. Valley and Ridge Stratigraphic Correlations, Virginia, vol. 37. Virginia Division of Mineral Resources Publication. Ragland, P.C., 1989. Basic Analytical Petrology. Oxford University Press, New York. Rampino, M.R., Self, S., Stothers, R.B., 1988. Volcanic winters. Annual Reviews of Earth and Planetary Science 16, 73–99. Renne, P.R., Ludwig, K.R., Karner, D.B., 2000. Progress and challenges in geochronology. Science Progress 83 (1), 107–121. Rodgers, J., 1953. Geologic map of east Tennessee with explanatory text. Tennessee Department of Conservation Division of Geology Bulletin 58, 168. Ross Jr., R.J., Naeser, C.W., Bergström, S.M., Cressman, E.R., 1981. Bentonites in the Tyrone Limestone dated by fission-tracks. Geological Society of America, Abstracts with Programs 13, 541. Ross Jr., R.J., Naeser, C.W., Izett, G.A., Obradovich, J.D., Bassett, M.G., Hughs, C.P., Cocks, L.R.M., Dean, W.T., Ingham, J.K., Jenkins, C.J., Rickards, R.B., Sheldon, P.R., Toghill, P., Whittington, H.B., Zalasiewicz, J., 1982. Fission-track dating of British Ordovician and Silurian stratotypes. Geological Magazine 119, 135–153. Ryder, R.T., Repetski, J.E., Harris, A.G., 1997. Stratigraphic framework of Cambrian and Ordovician rocks in the Central Appalachian Basin from Campbell County, Kentucky, to Tazewell County, Virginia. In: Proceedings of the US Geological Survey Miscellaneous Investigations Series Map I-2530, revised and digitized by Erika E. Lentz, 2004. /http://pubs.usgs.gov/imap/i-2530/S. Saltzman, M.R., Bergström, S.M., Huff, W.D., Kolata, D.R., 2003. Conodont and graptolite biostratigraphy and the Ordovician (Early Chatfieldian, Middle Caradocian) d13C excursion in North America and Baltoscandia: implications for the interpretation of the relations between the Millbrig and Kinnekulle K-bentonites. In: Albanesi, G.L., Beresi, M.S., Peralta, S.H. (Eds.), Ordovician from the Andes: Proceedings of the Ninth International Symposium on the Ordovician System, vol. 17. INSUGEO: Serie Correlacion Geologica, pp. 137–142. Samson, S.D., Patchett, P.J., Roddick, J.C., Parrish, R.R., 1989. Origin and tectonic setting of Ordovician bentonites in North America: isotopic and age constraints. Geological Society of America Bulletin 101, 1175–1181. Schmitz, M.D., Bowring, S.A., Ludwig, K.R., Renne, P.R., 2003. Comment on ‘‘Precise K–Ar, 40Ar–39Ar, Rb–Sr and U–Pb mineral ages from the 27.5 Ma Fish Canyon Tuff reference standard’’ by M.A. Lanphere and H. Baadsgaard. Chemical Geology 199 (3–4), 277–280. Thorslund, P., 1948. The Chasmops Series of the Kullatorp Core. In: Waern, B., Thorslund, P., Hennigsmoen, G. (Eds.), Deep boring through Ordovician and Silurian strata at Kinnekulle, Vestergötland, vol. 32. Bulletin of the Geological Institute, University of Uppsala, pp. 343–373. Toulkeridis, T., Clauer, N., Chaudhari, S.K., Goldstein, S.L., 1998. Multimethod (K–Ar, Rb–Sr, Sm–Nd) dating of bentonite minerals from the eastern United States. Basin Research 10, 261–270. Tucker, R.D., McKerrow, W.S., 1995. Early Paleozoic chronology: a review in light of new U–Pb zircon ages from Newfoundland and Britain. Canadian Journal of Earth Sciences 32, 368–379. Tucker, R.D., 1992. U–Pb dating of plinian-eruption ashfalls by the isotope dilution method; a reliable and precise tool for time-scale calibration and biostratigraphic correlation. Geological Society of America, Abstracts with Programs 24 (7), 198. Tucker, R.D., Krogh, T.E., Ross Jr., R.J., Williams, S.H., 1990. Time-scale calibration by high-precision U–Pb zircon dating of interstratified volcanic ashes in the Ordovician and Lower Silurian stratotypes of Britain. Earth and Planetary Science Letters 100, 51–58. Ulrich, E.O., 1888. Correlation of the Lower Silurian sections of Tennessee and of the Ohio and Mississippi valleys with those of New York and Canada. American Geologist 1, 100–110. Ver Straeten, C.A., 2004. K-bentonites, volcanic ash preservation, and implications for Early to Middle Devonian volcanism in the Acadian ARTICLE IN PRESS W.D. Huff / Quaternary International 178 (2008) 276–287 orogen, eastern North America. Geological Society of America Bulletin 116, 474–489. Weaver, C.E., 1953. Mineralogy and petrology of some Ordovician Kbentonites and related limestones. Geological Society of America Bulletin 64, 921–943. Williams, I.S., Tetley, N.W., Compston, W., McDougall, I., 1982. A comparison of K–Ar and Rb–Sr ages of rapidly cooled igneous rocks: two points in the Palaeozoic time scale re-evaluated. Journal of the Geological Society, London 139, 557–568. 287 Willman, H.B., Kolata, D.R., 1978. The Platteville and Galena Groups in northern Illinois. Illinois State Geological Survey, Champaign-Urbana, IL. Wilson Jr., C.W., 1949. Pre-Chattanooga stratigraphy in central Tennessee. Tennessee Division of Geology Bulletin 56, 1–407. Wolcott, D.E., Cressman, E.R., Connor, J.J., 1972. Trend-surface analysis of the thickness of the High Bridge Group (Middle Ordovician) of central Kentucky and its bearing on the nature of the post-Knox unconformity. United States Geological Survey Professional Paper 800-B, pp. 25–33.