1 Department of Geology, University of Cincinnati, Cincinnati, OH 45221, USA. E–mail: warren.huff@uc.edu

2 Department of Geological Sciences, The Ohio State University, 155 S. Oval Mall, Columbus, OH 43210, USA.

3 Illinois State Geological Survey, 615 E. Peabody Dr., Champaign, IL 61820, USA.

4 Centro de Investigaciones Geológicas, Universidad Nacional de La Plata, Calle 1, Nº 644, 1900 La Plata, Argentina.

5 Department of Earth and Environmental Sciences, University of Illinois at Chicago, 845 W. Taylor St., Chicago, IL 60607. USA

6 Department of Earth Sciences, University of Southern California, 3651 Trousdale Pkwy, Los Angeles, CA 90089. USA

Key words: K-bentonites. Volcanism. Paleogeography. Ordovician. Precordillera.

The discovery of an extensive succession of Ordovician K–bentonites in the Argentine Precordillera appeared, at first glance, to fit nicely with an emerging model of the Precordillera as a former fragment of Laurentia. The tectonic and paleogeographic implications of that putative connection have been thoroughly explored by numerous authors (Astini and Thomas, 1999), and a Laurentian origin has been provisionally accepted by many Ordovician paleogeographers. Most of the K–bentonites occur in the eastern thrust belts, where they are quite common in the San Juan Limestone and the overlying Gualcamayo Formation, but a few ash beds are known also from the central thrust belts. The oldest known K–bentonites occur in the middle Arenig I. victoriae lunatus graptolite (Oe. evae conodont) Zone in the topmost part of the San Juan Limestone and the youngest of about middle Llanvirn age in the Pt. elegans graptolite Zone in the Gualcamayo Formation. Upon closer inspection, tentative correlations with supposed coeval relic ash beds in the Solitario region of west Texas (Dickerson and Keller, 1998) have proved misleading and, thus, there are no known equivalent K–bentonite sequences in North America. The Argentine sequence is unique in its abundance of K–bentonite beds, yet when compared with K–bentonite sequences of similar age elsewhere it provides no supporting evidence of a close association between the Precordillera and other Ordovician sedimentary basins at that time. The ash distribution pattern is more consistent with paleogeographic reconstructions which envision early Ordovician drifting of the Precordillera in fairly close proximity to one or more additional volcanic arcs and with eventual collision along the Andean margin of Gondwana during the mid–Ordovician Ocloyic event of the Famatinian orogeny. While coeval ash beds do not appear to be present in North America, there is nevertheless a widespread suite of Middle and Late Ordovician K–bentonite beds that has been extensively documented (Kolata et al., 1996). The absence of corresponding ash bed successions does not, of course, render the entire Laurentian model implausible, but it does require that a corresponding tectonomagmatic environment be established to serve as a source region for these explosively erupted ashes. Further, it complicates the presumed timing of rifting and drifting of the Precordillera by placing it in the proximity of active collision margin volcanism in early Middle Ordovician time. We present here a brief comparative analysis of geochemical and mineralogical features of both Precordilleran and Laurentian Ordovician K–bentonites in order to show there are substantial differences between them that imply quite different tectonomagmatic histories.

Many Ordovician K–bentonites typically include a matrix of illite/smectite mixed–layer clay plus a phenocryst assemblage common to many felsic volcanic rocks including biotite, beta–form quartz, alkali and plagioclase feldspar, apatite, and zircon, with lesser amounts of hornblende, clinopyroxene, sphene, and Fe–Ti oxides. The proportions of the mineral phases and variations in their crystal chemistry are commonly unique to small groups of K–bentonite beds, and frequently to individual beds themselves. Precordillera K–bentonites are somewhat unique in containing abundant clinopyroxenes, along with other phenocrysts common in other regions (Cingolani et al., 1997; Huff et al., 1995; Krekeler et al., 1995; Prokopenko et al., 1997). Further, glass melt inclusions preserved in quartz crystals provide helpful information about the composition of the parental magma at the time of quartz crystallization. These two aspects of K–bentonite composition together provide insight into some of the principal differences between the Argentine and North American sequences.

Figure 2. Composition of quartz–hosted melt inclusions from Cerro Viejo, Argentina, and Shakertown, Kentucky.

Pyroxene was identified in at least five K–bentonite beds at Talacasto, San Juan Province, Argentina, where they constitute as much as 80 percent of the non–opaque heavy mineral fraction (Prokopenko et al., 1997). The crystals appear as angular, highly fragmented grains ranging from pale green to greenish–brown in color under the petrographic microscope. Microprobe analyses (Figure 1) show that they are calcium–rich clinopyroxenes with a diopside composition. Their occurrence is somewhat unusual when compared to the North American and Baltoscandic K–bentonites which do not have them, but it is consistent with the general mineralogy of volcanic rocks generated in highly silicic magma chambers (Hildreth, 1979).

The directions of compositional trends of clinopyroxenes from the Talacasto section present some systematic changes that appear to vary with stratigraphic position. For example, augites with the least Ca–enrichment from the two lowest beds in the section (Figure 1) display a tholeitic differentiation trend (Sack and Ghiorso, 1994) whereas crystals from the uppermost parts of the section are much more clustered and show little compositional variation or show trends along the Ca–Fe line of the clinopyroxene quadrilateral. Clinopyroxenes crystallizing out of felsic magmas would be expected to show an enrichment in Fe and Ca if mixing with more mafic magmas was involved. Thus, pyroxene compositions in the Talacasto section suggest a somewhat more complex history of magmatic development than previously shown by other Ordovician K–bentonites along the Iapetus margin.

Quartz is very common in both North and South American Ordovician K–bentonites, and frequently occurs in beta–form morphology, characteristic of volcanogenic quartz (Bohor and Triplehorn, 1993). Some quartz crystals contain pristine glass melt inclusions, which are clear, rounded or ovoid in shape, and range between 20 and 75 µm in diameter. We report here the results of microprobe analyses of some melt inclusions in both Argentine and North American K–bentonites which serve to constrain the composition of the pre–eruption parental magma. Microprobe analysis of 14 glass melt inclusions in quartz grains from Cerro Viejo and 25 melt inclusions for the Caradocian Millbrig K–bentonite at Shakertown, Kentucky, yielded oxide totals near 96.6%, and the data were plotted on a total alkalies versus silica (TAS) diagram (Figure 2). Both ashes were clearly derived from silicic magmas, but are distinguishable from one another on the basis of both silica and total alkali content. Felsic magmas frequently contain dissolved water in the range of 3–5 wt percent, and are capable of producing large–scale explosive eruptions, particularly when associated with caldera–forming events (Rose and Chesner, 1987). Silicic magmas may also contain some amount of exsolved volatile constituents which would have been released at the time of eruption and thus not be preserved in the trapped glass, so while these analyses may not reflect complete parental magma compositions they are sufficiently distinct and informative to permit assignment to a tectonomagmatic origin.

An unresolved question concerning the Precordilleran K–bentonites is the location of the volcanoes that produced these ashes. In the case of Quaternary and Holocene ashes, regional mapping combined with geochemical and mineralogical fingerprinting is frequently capable of pinpointing the location of the source vents. However, with ancient volcanics those source vents have generally been destroyed long since and one is confronted with problems similar to most palaeogeographic reconstructions. Assessment of variation in the number of beds, their stratigraphic distribution, and their relative thickness within a region may provide some clues to the direction of ash transport (Huff et al., 1992; Huff et al., 1996; Kolata et al., 1996). Such assessment in the Precordillera, however, is complicated by the fact that the K–bentonites occur in thrust belts that were subjected to eastward transport of large but uncertain distances during Andean tectonism.

More than 60 K–bentonite beds are currently known from the Ordovician of North America (Kolata et al., 1996) and some of these beds, such as the Deicke and the Millbrig K–bentonites, are widely distributed. However, a comparison of the stratigraphic distributions of ash beds in the Precordillera and North America show considerable differences (Bergström et al., 1996). In the latter region, very few ash beds are known from Lower Ordovician and the lower Middle Ordovician. These include single beds in Mississippi (Kolata et al., 1996; Thomas, 1988) in the Marathon area, western Texas (King, 1937), an undetermined number of beds of early Middle Ordovician age in a giant sinkhole near Douglas Lake, eastern Tennessee (Laurence, 1944), and a few ash beds on western Newfoundland (Kolata et al., 1996). Keller and Dickerson (Keller and Dickerson, 1996) recently misidentified some calcareous mudstones of early Middle Ordovician age from the Solitario region, western Texas, as K–bentonites. Hence, we currently know of no obvious counterpart in North America to the extensive K–bentonite bed complex of Early and early Middle Ordovician age in the Precordillera. However, it may be significant to note that much of the lower Middle Ordovician, which contains abundant K–bentonites in the Precordillera, is not represented in the successions in the Southern and Central Appalachians, where this interval is cut out by the prominent Sauk–Tippecanoe unconformity (Faill, 1997; Ross et al., 1982).

The principal K–bentonite complex in eastern North America is in the upper Middle Ordovician (Mohawkian) where there are more than 40 ash beds (Kolata et al., 1996) two of which may reach a thickness of 0.1–1 m. No equivalents to this K–bentonite complex, as well as to the few North American Upper Ordovician ash beds, are known in the Precordillera. This can be taken as an indication that the close Middle Ordovician juxtaposition of the Precordillera and easternmost North America shown in several recent paleogeographic reconstructions is unlikely to be correct.

Numerous K–bentonites occur in the San Juan and Gualcamayo Formations in the eastern thrust belts of the Precordillera, and a few ash beds are also known from the central thrust belts. However, none are known from the western belt despite extensive field work in that area. As is typical of many Ordovician K–bentonites found elsewhere, they consist of a matrix of illite/smectite mixed–layer clay plus a phenocryst assemblage including biotite, beta–form quartz, alkali and plagioclase feldspars, apatite, and zircon, hornblende, clinopyroxene, sphene, and Fe–Ti oxides. Pristine melt inclusions in quartz crystals are rhyolitic in composition.

The Argentine sequence is unique in its abundance of K–bentonite beds, yet when compared with K–bentonite sequences of similar age elsewhere it provides no supporting evidence of a close association between the Precordillera and other Ordovician sedimentary basins along the Iapetus margin at that time. Palaeogeographic reconstruction for the early Palaeozoic is, at best, a "weight of evidence" process, in which the more pieces of information there are available, the more solid the arguments for a particular scenario. K–bentonite stratigraphic and geographic distribution patterns can be helpful in this process by providing direct evidence of subduction–related explosive volcanism and the accompanying palaeowind directions, sedimentary conditions and magmatic processes which existed at the time. Based on that record, it is clear that subduction–related volcanism in the Precordillera commenced in the mid–Arenig and continued to the late–Llanvirn, coeval with volcanism in the Famatina range. We do not find supporting evidence that the Precordilleran terrane was attached to Laurentia prior to mid–Ordovician time (Astini et al., 1995; Thomas and Astini, 1996). Rather, we find the ash pattern is more consistent with the palaeogeographic reconstructions that envision drifting of the Precordillera in fairly close proximity to one or more additional volcanic arcs with eventual collision along the Andean margin of Gondwana during the Ocloyic orogeny in Middle Ordovician time. The extensive K–bentonite record in the Guandacol region suggests that the source volcanoes may have been located north or northeast of the Precordillera. Conceivably, the Puna–Famatina terrane could have been one of these volcanic arcs and might have served as one source of the K–bentonite ashes, possibly in concert with active arc magmatism on the Gondwana plate itself.

This research has been suported by NSF grants INT–9513128 and EAR–9204893 to WDH, SMB, and DRK. We also acknowledge support from CONICET to CC as part of a joint NSF–CONICET agreement.

Accepted: June 15, 2003