Journal of Petrology Advance Access originally published online on December 10, 2004
Journal of Petrology 2005 46(2):407-439; doi:10.1093/petrology/egh082
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology vol. 46 issue 2 © Oxford University Press 2004; all rights reserved
Petrogenesis of Pre-caldera Mafic Lavas, Jemez Mountains Volcanic Field (New Mexico, USA)
1 DEPARTMENT OF GEOLOGY, WASHINGTON STATE UNIVERSITY, PULLMAN, WA 99164, USA
2 DEPARTMENT OF GEOSCIENCES, OREGON STATE UNIVERSITY, 104 WILKINSON HALL, CORVALLIS, OR 97331, USA
3 DEPARTMENT OF GEOLOGICAL AND ENVIRONMENTAL SCIENCES, CALIFORNIA STATE UNIVERSITY, CHICO, CA 95929, USA
4 EARTH AND ENVIRONMENTAL SCIENCES, LOS ALAMOS NATIONAL LABORATORY, LOS ALAMOS, NM 87545, USA
5 DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF CALIFORNIA, SANTA CRUZ, CA, 95064, USA
6 DEPARTMENT OF GEOLOGY, McMASTER UNIVERSITY, 1280 MAIN STREET WEST, HAMILTON, ONTARIO L8S 4M1, CANADA
RECEIVED MARCH 30, 2004; ACCEPTED SEPTEMBER 20, 2004
 |
ABSTRACT |
|---|
The Miocene–Quaternary Jemez Mountains volcanic field (JMVF), the site of the Valles caldera, lies at the intersection of the Jemez lineament, a Proterozoic suture, and the Cenozoic Rio Grande rift. Parental magmas are of two types: K-depleted silica-undersaturated, derived from the partial melting of lithospheric mantle with residual amphibole, and tholeiitic, derived from either asthenospheric or lithospheric mantle. Variability in silica-undersaturated basalts reflects contributions of melts derived from lherzolitic and pyroxenitic mantle, representing heterogeneous lithosphere associated with the suture. The K depletion is inherited by fractionated, crustally contaminated derivatives (hawaiites and mugearites), leading to distinctive incompatible trace element signatures, with Th/(Nb,Ta) and La/(Nb,Ta) greater than, but K/(Nb,Ta) similar to, Bulk Silicate Earth. These compositions dominate the mafic and intermediate lavas, and the JMVF is therefore derived largely, and perhaps entirely, from melting of fertile continental Jemez lineament lithosphere during rift-related extension. Significant variations in Pb and Nd isotope ratios (206Pb/204Pb = 17·20–18·93; 143Nd/144Nd = 0·51244–0·51272) result from crustal contamination, whereas 87Sr/86Sr is low and relatively uniform (0·7040–0·7048). We compare the effects of contamination by low-87Sr/86Sr crust with assimilation of high-87Sr/86Sr granitoid by partial melting, with Sr retained in a feldspathic residue. Both models satisfactorily reproduce the isotopic features of the rocks, but the lack of a measurable Eu anomaly in most JMVF mafic lavas is difficult to reconcile with a major role for residual plagioclase during petrogenesis.
KEY WORDS: Jemez Mountains volcanic field; Rio Grande rift; lithospheric mantle; crustal contamination; trace elements; radiogenic isotopes
|
INTRODUCTION |
|---|
The Jemez Mountains volcanic field (JMVF) in New Mexico is well known as the site of the Valles caldera, one of three large Quaternary continental rhyolitic caldera systems in the USA (the other two are Long Valley and Yellowstone). The JMVF has a history of volcanism reaching back over 20 Myr prior to the caldera-forming eruptions of the Bandelier Tuff, involving magma compositions ranging from nephelinite to high-silica rhyolite. This longevity of active volcanism, despite significant westward movement of the North American continent during this time, indicates that the controls on the location of JMVF volcanism reside within the lithosphere. Our purpose is to examine the bulk-rock geochemistry of pre-Valles JMVF mafic volcanic rocks with the intention of identifying mantle components that have played important roles in magmatism, the extent to which the mantle-derived magmas have been compositionally modified by interaction with regional crust, and to describe their contribution to the construction of a moderate-sized (
2000 km3), long-lived volcanic field that ultimately hosted caldera-forming eruptions of catastrophic magnitude. We focus on lavas with <57% SiO2; silicic rocks will be discussed in a future paper. Although lavas in this compositional range make up less than 10% of the exposed JMVF pile, they span most of the lifetime of the field, and a mafic component is often present in the much more voluminous mixed andesites and dacites; there can be little doubt that the JMVF is ‘fundamentally basaltic’ (Hildreth, 1981
). This paper is intended to provide a basis for future, more detailed studies of the temporal development of the JMVF, and to provide a foundation for understanding the petrogenesis of the climactic rhyolitic magmas.
The earliest lavas in the area, dated at 25–16·5 Ma, are interbedded with Santa Fe Group basin-fill sediments of the adjacent Rio Grande rift (Bailey et al., 1969; Smith et al., 1970; Gardner et al., 1986; Woldegabriel et al., 2003). Subsequent activity between 13 and 6 Ma built a large (1000 km3) andesite-dominated volcanic ridge on the basin margin (Fig. 1), now represented by lavas of the Paliza Canyon Formation and other units of the Keres Group (Bailey et al., 1969; Smith et al., 1970), mostly exposed in the southern Jemez Mountains. The northern part of the Jemez Mountains consists mainly of lavas making up the Polvadera Group of Bailey et al. (1969). The earliest, basalt-dominated, Polvadera activity occurred between 14 and 7 Ma (Goff et al., 1989; Aldrich & Dethier, 1990), to produce the Lobato Basalt. Coincident with a lull in crustal extension (Gardner et al., 1986), the dacite-dominated Tschicoma Formation was erupted between 7 and 2 Ma (Goff et al., 1989) in the northern and northeastern Jemez. The latter stages of Polvadera Group activity shifted towards rhyolitic compositions between 4 and 2 Ma (Turbeville et al., 1989). During the same period, mafic volcanism around the eastern and northern margins of the JMVF (Smith et al., 1970; Baldridge, 1979) produced the Cerros del Rio and El Alto volcanic fields, which consist of lavas and scoria with 45–64% SiO2. It should be emphasized that JMVF formations exhibit considerable stratigraphic, geochronological and geographical overlap. Prior work has indicated that the pre-caldera volcanic rocks record varying styles and degrees of interaction between mantle-derived magmas and continental crust (Gardner, 1985; Singer & Kudo, 1986; Duncker et al., 1991; Ellisor et al., 1996; Wolff et al., 2000).
|
The Tewa Group of Bailey et al. (1969)
consists of the rhyolitic products of the Valles caldera, which are dominated by the Bandelier Tuff (Fig. 1a). The major caldera-forming units of the Bandelier Tuff, the Otowi and Tshirege Members, erupted at 1·61 and 1·23 Ma, respectively (Izett & Obradovich, 1994; Spell et al., 1996). Smaller volume rhyolitic lavas and tuffs were emplaced before, between, and after the Bandelier eruptions. Aspects of the petrology and geochemistry of the caldera-related rhyolites have been considered by Smith (1979), Stix et al. (1988), Spell & Kyle (1989), Spell et al. (1990, 1993, 1996), Dunbar & Hervig (1992), Stix & Gorton (1993), Wolff & Gardner (1995), Stimac (1996), Wolff et al. (1999, 2002) and Wolff & Ramos (2003). |
REGIONAL TECTONIC SETTING, GEOLOGY, AND MAGMATISM |
|---|
The Rio Grande rift is the only major manifestation of Basin-and-Range extension that lies east of the Colorado Plateau. It consists of a north–south-trending series of Cenozoic en echelon sedimentary basins that bisects New Mexico and southern Colorado. The rift cuts across regional trends in mid-Proterozoic metamorphic and granitic rocks and overlying Upper Paleozoic to Cenozoic sedimentary rocks. The JMVF is built on the western shoulder of the Española basin, at the intersection of the rift and the Jemez lineament (Fig. 1a). The latter is an alignment of late Cenozoic volcanic fields extending from SE Arizona to NE New Mexico, which coincides with the surface boundary of a suture zone complex between the Proterozoic Southern Yavapai (1·8–1·7 Ga) and Mazatzal (1·65 Ga) lithospheric provinces (Shaw & Karlstrom, 1999
; Karlstrom et al., 2002). Compared with the abundant volcanism along the lineament, the Rio Grande rift is relatively starved of magma; indeed, the JMVF is by far the largest volcanic field associated with the rift.
Recent geophysical investigations, summarized by Duecker et al. (2001) and Karlstrom et al. (2002), have shed light on lithospheric structure beneath the Jemez lineament just east of the rift (CD-ROM project seismic line, Fig. 1a). Seismic reflection data show south-dipping mid-crustal reflections that project toward a south-dipping mantle boundary broadly coincident with the Jemez lineament. The boundary separates high-velocity mantle to the south from low-velocity mantle to the north, extends to >200 km depth, and is interpreted as the Southern Yavapai–Mazatzal suture (Karlstrom et al., 2002). The low-velocity mantle to the north of the lineament is one of a number of NE-trending similarly aligned features beneath the lineament, which Duecker et al. (2001) argued are more significant regional mantle structures than those resulting from rifting. Significantly for magma genesis, the low-velocity feature is internally layered, suggesting that it resides in the lithosphere. It extends from the Moho at 40–50 km to a depth of 120 km, the inferred base of the lithosphere.
Rifting in northern New Mexico began around 30 Ma with crustal extension, locally exceeding 100% (Morgan et al., 1986), over a width of 170 km. The original Española basin, coincident with but broader than the present basin, formed with a western boundary fault in the Nacimiento Mountains (Aldrich, 1986; Morgan et al., 1986). Following a period of quiescence from at least 18 to 13 Ma, extension was reinitiated within a narrower zone (50 km) along the pre-existing axis, with the western rift boundary trending north–south through the center of the JMVF. A further lull in extensional activity between 7 and 4 Ma (Gardner et al., 1986) ended with the development of the Pajarito fault zone (Fig. 1), marking another eastward shift of the Española basin's western boundary. Temperatures may exceed 900°C in the lower crust beneath the rift (Baldridge et al., 1984; Clarkson & Reiter, 1984; Morgan & Golombek, 1984). This is within the range of solidus temperatures of crustal lithologies, hence thermal perturbations caused by intrusions of mantle-derived primitive basaltic magma into the sub-rift crust are very likely to induce partial melting of the Proterozoic crustal rocks.
The JMVF is built on a substrate of Upper Paleozoic sedimentary strata that rest on the Proterozoic basement, which consists of granitoid and metamorphic rocks locally dated at 1·62–1·44 Ga (Brookins & Laughlin, 1983). The Paleozoic succession is associated with basin development during the ancestral Rocky Mountains orogeny (Pazzaglia et al., 1999) and is overlain by a veneer of Cenozoic sedimentary units, the uppermost of which is the eastward-thickening rift-filling Santa Fe Group (Bailey et al., 1969; Smith et al., 1970). Geothermal wells (Eichelberger & Koch, 1979; Nielson & Hulen, 1984) and Continental Scientific Drilling Project corehole VC-2B (Hulen & Gardner, 1989) penetrated the Proterozoic basement beneath the present-day Valles caldera. The depth to the Proterozoic basement beneath the caldera varies from 1·5 km beneath the west side of the caldera to 5 km in the east (Goff et al., 1989).
Proterozoic rocks exposed or intersected by drillholes within the Valles caldera are exclusively granitoid, although metavolcanic rocks were penetrated by the Fenton Hill well, just outside the western caldera margin (Laughlin et al., 1983), and amphibolites occur along with granitoids as lithic fragments in the Bandelier Tuff (Eichelberger & Koch, 1979). Metavolcanic and metasedimentary rocks are also exposed in uplifts around the borders of the Española basin. Some Proterozoic lithics from the Bandelier Tuff are partly melted, and the Otowi Member magma chamber was actively stoping country rock at the time of eruption (Eichelberger & Koch, 1979). Hence, these lithics are derived from near the zone of magma storage beneath the volcanic field, and represent crustal lithologies actually involved in magmatism. This point is considered later in the paper.
In summary, JMVF volcanism probably results from extension acting on the presumably weak lithosphere of the Jemez lineament, which itself is a localizing force on regional Cenozoic magmatism. Several components may contribute to magmatism: convecting asthenospheric mantle, Proterozoic oceanic lithosphere, and a wide variety of crustal lithologies. The mantle source regions of basalts in the Rio Grande rift and adjacent regions have been variously considered to lie in the lithosphere and/or convecting asthenosphere (Perry et al., 1987; Leat et al., 1988, 1989; Duncker et al., 1991; Johnson & Beard, 1993; McMillan, 1998; Wolff et al., 2000; Baldridge, 2004). Largely on the basis of trace element and isotopic signatures in the lavas, three or four chemical types of mantle source have been invoked: the convecting asthenosphere, similar to the source for modern mid-ocean ridge basalts (MORB) (Leat et al., 1988, 1989; McMillan, 1998; McMillan et al., 2000); Proterozoic lithosphere enriched by basaltic melt (McMillan, 1998; Baldridge, 2004); and lithosphere enriched by fluids from subducting slab(s) during the Proterozoic (Leat et al., 1988, 1989; McMillan, 1998) and/or the Cenozoic, perhaps further modified by loss of large ion lithophile elements (LILE) in pre-magmatic fluids (Duncker et al., 1991). Importantly, Johnson & Beard (1993) used Hf isotopes to show that the Proterozoic lithospheric mantle component in Rio Grande rift lavas originally melted in the spinel stability field and is therefore of probable oceanic origin. Although all investigators recognize a role for crustal melting in regional magmatism, there is no consensus on whether or not trace element and isotopic features of the most primitive basalts bear a significant crustal imprint. For example, the arc-similar signatures of northern Rio Grande rift tholeiitic lavas with LILE/HFSE (high field strength element) ratios greater than those of typical MORB and ocean island basalts (OIB) have been variously ascribed to crustal contamination of asthenosphere-derived magma (Duncker et al., 1991) or a source in subduction-modified mantle (McMillan, 1998).
|
ANALYTICAL METHODS |
|---|
Major elements and selected trace elements were analyzed by X-ray fluorescence (XRF) at New Mexico Institute of Mining and Mineral Technology [NMIMT; procedures of Hallett & Kyle (1993)
] and in the Geoanalytical Laboratory at Washington State University [WSU; procedures of Johnson et al. (1999)]. Results from the two laboratories are comparable (Johnson et al., 1999). All Keres, Polvadera Group and basement samples were additionally analyzed for trace elements by inductively coupled plasma mass spectrometry (ICP-MS) at WSU [see Knaack et al. (1994) for procedures]. Additional data (including Sr and Nd isotopes) on Santa Fe Group and some Cerros del Rio lavas and basement samples are taken from Gibson et al. (1993) and Duncker et al. (1991), although most of Duncker et al.'s Cerros del Rio samples have been reanalyzed at WSU. In most diagrams presented here, all available data are plotted. In cases where high precision is an issue, for example U/Nb ratios, only WSU data are plotted.
Sr, Nd and Pb isotopes on Keres Group samples, and Sr and Nd isotopes on Taos Range basement rocks, were analyzed at University of California Los Angeles (UCLA). Following dissolution, samples for Pb analysis were brought up in a HBr–HNO3 solution. Teflon columns (500 µl) with 100–200 mesh AGIX-8 anion exchange resin were cleaned with H2O, dilute HCl and HNO3, then pre-treated with the HBr–HNO3 solution, and the sample solution was loaded onto the columns and washed with further HBr–HNO3. Pb was collected from the resin with H2O. Otherwise, the procedures of Davidson et al. (1993) were followed.
Sr and Nd isotopes on Polvadera Group samples, and Pb isotopes on Taos range basement rocks and some Polvadera Group samples, were analyzed under the supervision of Dr Robert Creaser at the University of Alberta using standard dissolution and ion exchange procedures; isotope ratios were determined on a Micromass S54 thermal ionization mass spectrometry (TIMS) instrument. Sr and Nd ratios were corrected for mass-dependent fractionation by normalizing to 86Sr/88Sr = 0·1194 and 146Nd/144Nd = 0·7219, respectively. During the period of analysis, SRM 987 averaged 87Sr/86Sr = 0·71027 and the Shin Etsu Nd standard averaged 143Nd/144Nd = 0·512046. Measured Pb ratios were normalized to values for NBS981 of 206Pb/204Pb = 16·9356, 207Pb/204Pb = 15·4891 and 208Pb/204Pb = 36·7006.
Pb isotope ratio analysis of Cerros del Rio and Santa Fe Group lavas were performed at McMaster University. Samples with loss on ignition (LOI) values >2%, together with samples that were suspected of having undergone post-eruptive alteration (e.g. carbonation) and those that required multiple analysis because of high Fe contents, were leached in warm 6N HCl overnight and rinsed in quadruple-distilled water prior to digestion. Standard digestion techniques using HF, HNO3, and HCl were employed. Pb was separated using standard anion exchange techniques using 0·7N HBr. Purified samples were then loaded on single Re filaments using silica gel, and analyzed on a VG 354 mass spectrometer using a single collector. During the period of analysis, NBS981 yielded mean values of 206Pb/204Pb = 16·901, 207Pb/204Pb = 15·447, and 208Pb/204Pb = 36·558 (n = 111). Data have been corrected for fractionation of 0·1% per a.m.u. In-run precisions of measured isotope ratios averaged <0·06% (2
). Total process blanks were less than 1 ng and are considered negligible. Pb isotopes on JMVF and Santa Fe range basement rocks were analyzed on a VG 54E mass spectrometer at Oxford University, using procedures similar to those at McMaster University.
Pb isotopes on Polvadera Group samples, and Sr, Nd and Pb isotopes on Bandelier Tuff basement lithics were separated using standard ion exchange procedures and analyzed using a ThermoFinnigan Neptune multicollector (MC)-ICP-MS at WSU. Pb procedures have been described by Wolff & Ramos (2003). Replicate isotopic analyses on selected samples indicate that the results from McMaster University, UCLA, University of Alberta, and WSU agree within error of the TIMS analyses; Pb isotope ratios analyzed by MC-ICP-MS are considerably more precise (Wolff & Ramos, 2003).
|
GENERAL CHARACTERISTICS OF JMVF MAFIC LAVAS |
|---|
In this paper, we consider all lavas with <57% SiO2 (nominally basalts and basaltic andesites). This cut-off is purely one of convenience; a complete continuum exists from the most mafic to the most silicic lavas in the JMVF. Many mafic lavas have alkaline affinities and are classified as hawaiites, mugearites and benmoreites (Fig. 2). Because geochemical groupings among JMVF mafic lavas (Fig. 3) were first recognized in the youngest flows (Duncker et al., 1991
; Wolff et al., 2000), we describe the formations in order from youngest to oldest. Data are given in Table 1.
|
|
|
Cerros del Rio and El Alto fields
The Cerros del Rio mafic volcanic field occupies the south–central part of the Española basin. Our samples are mostly those of Duncker et al. (1991)
, for which we present new trace element and Pb isotope data. All come from the northern part of the field, within a 15 km radius of White Rock (Fig. 1b). Lavas with <57% SiO2 fall into two compositional groups: tholeiites, similar to the Miocene tholeiites, and nepheline- to weakly hypersthene-normative hawaiites–mugearites. The tholeiites were erupted between 2·48 and 2·33 Ma (Woldegabriel et al., 1996), late in the history of the field, from vents located near its buried northwestern extremity east of the Rio Grande (Dethier, 1997).
Hawaiites and mugearites form a compositional continuum (Fig. 2). Olivine is the only phenocryst in the most mafic hawaiites, but is joined by augite and hypersthene as silica increases. Quartz xenocrysts of crustal origin are common. These lavas dominate the White Rock Canyon sections sampled by Duncker et al. (1991) and most were erupted over a short time period (2·57–2·46 Ma, Woldegabriel et al., 1996). They have distinctive incompatible trace element abundances, with approximately Bulk Silicate Earth (BSE) K/Nb and sub-BSE Rb/Nb, but elevated La/Nb and Th/Nb (Fig. 3b), and thus do not correspond to any common globally recognized mafic magma type (i.e. arc basalt, intraplate basalt, or MORB). Wolff et al. (2000) proposed that these compositions originated through mixing of nephelinite–basanite, represented by the Santa Fe Group lavas (see below), with crustal melts.
Lavas and scoria of the much smaller El Alto field, in the northern JMVF, are petrographically and chemically similar to Cerros del Rio tholeiites, hawaiites and mugearites (Wolff et al., 2000). This field may be slightly older than most of the Cerros del Rio lavas; one flow has been dated at 3·2 Ma (Baldridge et al., 1980).
Polvadera Group
The Polvadera Group consists of the Lobato Basalt, Tschicoma, Puye, and El Rechuelos Rhyolite Formations (Bailey et al., 1969). Of these, only the Lobato Basalt contains mafic lavas, although basaltic andesite and mugearite enclaves occur in some Tschicoma Formation dacites. The bulk of the Lobato Basalt was erupted between 10·8 and 7·8 Ma (Manley, 1982; Goff et al., 1989), but flows interbedded with Santa Fe Group sediments in the northern JMVF, mapped as Lobato by Smith et al. (1970) and Aldrich & Dethier (1990), have been dated as old as 14·1 ± 0·3 Ma (Dethier et al., 1986; Aldrich & Dethier, 1990). Most of our Lobato samples come from the Clara Peak shield section of Goff et al. (1989), and are tholeiites with a few weakly ne-normative lavas; this predominance of tholeiites is probably representative of the Lobato basalt as a whole (Goff et al., 1989). Lobato tholeiites are olivine-phyric with sparse xenocrysts of quartz. MgO contents for the samples we have analyzed fall in the range 3·9–7·9%, and the lavas are generally compositionally similar to the later Cerros del Rio and El Alto tholeiites (Figs 2 and 4a).
|
Mafic enclaves with
53% SiO2 in Tschicoma Fm. dacite lavas (7–2 Ma; Singer & Kudo, 1986; Goff et al., 1989) are rounded to angular and typically 2–10 cm (maximum 50 cm) in size. Crenulate and sometimes diffuse margins indicate that they were at least partially liquid at the same time as the host dacite. Their alkali contents overlap with the Paliza Canyon mugearites and they are slightly less alkaline than the otherwise similar Cerros del Rio mugearites (Fig. 2), but have the same distinctive trace element patterns as both these groups of lavas (Fig. 3b).
Paliza Canyon Formation
The Keres Group, which displays a spectrum of magma types from basalt to rhyolite, formed between 13 and 6 Ma, coincident with an episode of crustal extension in the adjacent Española basin (Gardner et al., 1986). Following Goff et al. (1990), all Keres Group mafic lavas are assigned to the Paliza Canyon Formation.
The formation is dominated by andesite, with subordinate volumes of basalt and dacite flows. Mafic lavas include olivine tholeiites, olivine basalts, hawaiites, mugearites and benmoreites (Fig. 2) with MgO contents in the range 2·4–7·0% and fall into the same petrographic and trace-element groupings as the Cerros del Rio lavas, although SiO2 contents are slightly higher and all Paliza Canyon samples we have analyzed are at least weakly hypersthene-normative. Olivine tholeiites have subdued enrichments in incompatible trace elements and generally resemble tholeiites of other formations (Fig. 3a); one sample contains a quartzite fragment. Other basalts and hawaiites resemble the equivalent Cerros del Rio lavas (Fig. 3b), although orthopyroxene is common in the mugearites and benmoreites; these lavas have the distinctive high La/Nb and Th/K ratios. Plagioclase phenocrysts (An30–60) typically show complex zoning and are resorbed with reaction rims suggestive of varying degrees of magma mixing and hybridization.
Santa Fe Group lavas
The oldest lavas that crop out within the confines of the JMVF are thin mafic flows interbedded with Santa Fe Group sediments, found just below the base of the Keres Group in the area of St. Peter's Dome (Fig. 1; see also Goff et al., 1990). At 25–16·5 Ma (Gardner et al., 1986; Woldegabriel et al., 2003), they represent infrequent activity heralding the main onset of JMVF magmatism by several million years. Among them is a larnite-normative nephelinite flow (sample JM93141 in Table 1) which, at 16% MgO, is the most primitive lava so far described from the Jemez Mountains. It has 15% olivine phenocrysts and microphenocrysts up to 1 mm in diameter, and trace clinopyroxene. Most of the olivine grains show euhedral outlines with tuning fork and re-entrant rich morphologies indicative of rapid crystallization. Core compositions of olivines are Fo85–90, with most in a restricted range Fo87–89 (average Fo87·5). Many grains have Cr-spinel and Cr-rich magnetite inclusions. The average olivine composition is close to equilibrium with the calculated groundmass [assuming magmatic Fe2+/(Fe2+ + Fe3+) = 0·85 in the liquid], but is forsterite poor compared with the calculated composition (Fo91) in equilibrium with the whole rock. We conclude that some accumulation of olivine has occurred; the range of olivine compositions is satisfied by 11–13% MgO in the pristine magma, which would correspond to removal of 10–15% olivine from the analyzed bulk composition.
The early St. Peter's Dome lavas, among which are intermediate compositions petrographically and chemically similar to Keres Group basaltic andesites and mugearites, form part of a more widespread, small-volume, early to middle Miocene volcanic episode in the Española basin (Gibson et al., 1993) that includes tholeiites and quartz-normative basaltic andesites as well as basanites and nephelinites. The tholeiites and basaltic andesites have 2·9–7·4% MgO, trace element patterns resembling enriched MORB (E-MORB; Fig. 3a), and appear to have been only slightly modified by interaction with continental crust (Gibson et al., 1993, and below); most resemble the younger tholeiites.
|
IDENTIFICATION AND ORIGIN OF JMVF PRIMARY MAGMAS |
|---|
Nephelinite–basanite
The Española basin silica-undersaturated lavas were described by Gibson et al. (1993)
. Hereafter, we refer to these, and the St. Peter's Dome flows, collectively as ‘Santa Fe Group lavas’, consistent with the usage of Gardner et al. (1986) and Goff et al. (1990). They are of two types (Fig. 4): a more primitive group of nephelinites and basanites (MgO 13–16%; Ni 290–420 ppm; Cr 530–550 ppm) among which we place the St. Peter's Dome nephelinite, and a lower-MgO group of basanites and alkali basalts (MgO 9–12%; Ni 160–270 ppm; Cr 150–350 ppm). The latter have, on average, lower CaO, higher SiO2, REE, Nb/Zr, 206Pb/204Pb, 208Pb/204Pb and lower 143Nd/144Nd (Fig. 4), and appear to be derived from a more enriched source than the higher-MgO, more silica-undersaturated group. The generally more enriched character of this group could be due to crustal contamination; however, relationships between Pb isotope ratios and compatible element abundances are inconsistent with contamination, which in the JMVF acts to decrease 206Pb/204Pb and 208Pb/204Pb (see below). In particular, compositions that represent the contaminated equivalents of the nephelinites (Wolff et al., 2000) have lower Pb isotope ratios at lower MgO contents (Fig. 5). Hence, it is more likely that the low-MgO primitive group is derived from a more enriched mantle source with higher U/Pb, Th/Pb and lower Sm/Nd than the high-MgO group.
|
Recently, mixed lherzolite–pyroxenite mantle has been invoked as a significant source for OIB (Hauri, 1996
; Hirschmann & Stolper, 1996; Sigmarsson et al., 1998; Lassiter et al., 2000). Pyroxenite veins may make up a few percent of the upper mantle (Hirschmann & Stolper, 1996), although the lower solidus of pyroxenite compared with peridotite may cause it to yield a disproportionately high fraction of the total melt from a volume of veined mantle. Potential sources of pyroxenite include recycled lower-crustal cumulates, frozen melts, and subducted oceanic crust; the last may be especially significant in an ancient suture zone such as the Jemez lineament. Pyroxenite melts have higher SiO2 and Al2O3, and lower CaO and MgO than small-degree melts of lherzolite. Some of the most compelling evidence for a pyroxenite contribution to mafic magma comes from primitive silica-undersaturated suites such as Hawaiian post-erosional lavas (Lassiter et al., 2000), and the alkalic series of Grand Comore (Class et al., 1998b, 1999; C. Class, unpublished data), in which major elements are correlated with Os isotope ratios. Both of these examples are located on young oceanic lithosphere, where the very high Re/Os of pyroxenite compared with peridotite results in measurable differences in 187Os/188Os after a short ingrowth period (10 Myr), in contrast to Sr, Nd and Pb isotope ratios. The major element variations of the Santa Fe Group undersaturated lavas closely mimic both the Hawaiian and Grand Comore suites (Fig. 6). Although we have no Os isotope data, the much greater age of the Proterozoic sub-JMVF lithospheric mantle (a lithospheric mantle source for these lavas is justified below) than the oceanic lithosphere beneath Hawaii and the Comores allows measurably greater ingrowth of 143Nd, 206Pb and 208Pb in the pyroxenite as compared with the lherzolite source (Figs 4 and 5). One contrast between the Santa Fe Group lavas and the oceanic suites is that the pyroxenite-derived melts in the latter have lower, not higher, abundances of incompatible trace elements, as the consequence of a higher degree of partial melting of the pyroxenite (Lassiter et al., 2000). We do not regard this as a serious obstacle to the applicability of the ‘mixed-source’ model in this case, for two reasons: (1) the isotopic data demonstrate that pyroxenite is substantially enriched in incompatible elements compared with the lherzolite source (Fig. 4); (2) the implicit assumption of pyroxenite–lherzolite melting under the same thermal conditions breaks down if the spatial scale of heterogeneity is at the kilometer scale or more, which is likely to be the case beneath the JMVF if the main pyroxenite reservoir is old Proterozoic oceanic crust (see below).
|
Normalized trace-element plots of Santa Fe Group nephelinites and basanites are shown in Fig. 7. To minimize the effects of minor fractional crystallization on incompatible trace element abundances, compositions have been corrected for minor olivine fractionation or accumulation. The more primitive group have been corrected to 12% MgO based on olivine–liquid relations in JM93141, whereas the lower-MgO, CaO lavas were corrected (somewhat arbitrarily) to an average value of 10% MgO. In these plots, the order of elements from left to right reflects increasing compatibility during partial melting of a four-phase lherzolite (ol + opx + cpx + sp or gt; Sun, 1980
; Thompson et al., 1984). Overall, the patterns fan slightly from right to left (Fig. 7), as expected for different degrees of partial melting. The relative depletion in K for the primitive lavas of both groups is the most striking feature of the normalized patterns.
|
Depletions in K relative to comparably incompatible elements are common among intraplate alkaline lavas, both in the ocean basins and on the continents. Explanations for the K depletion fall into two groups: (1) prior selective K depletion of the mantle source; (2) K is retained during partial melting by a residual potassic mineral such as amphibole or phlogopite. In the first case, K loss occurs prior to magmatism and there is no implied constraint on the physical nature of the mantle source. Incorporation of ancient subducted oceanic crust, stripped of K by dehydration during subduction, into the intraplate magma mantle source is a possible mechanism (Weaver, 1991
). A major objection to this model in the present case is the lack of similarity in the behaviors of K and Ba (Fig. 7), which should be depleted to similar extents during slab dehydration and hence show coupled variation in lavas derived from sources with an ancient subducted component.
In the second case, the nature of the mantle source is constrained by the stability fields of amphibole and phlogopite in a lherzolitic assemblage. Primitive K-depleted basanites and nephelinites, very similar to the Santa Fe Group lavas, have been described from Grand Comore by Späth et al. (1996), Class & Goldstein (1997) and Class et al. (1998a), and from Kenya by Späth et al. (2001). In each area, those workers conclusively showed by careful modeling of cogenetic suites that melting of an anhydrous four-phase lherzolite source cannot account for the behavior of K, and that a hydrous K-bearing phase, more likely to be amphibole than phlogopite, must be present in the residue. The Santa Fe Group lavas were erupted at different times from widely separated locations and cannot be regarded as a cogenetic suite amenable to modeling as the products of a single partial melting event. None the less, we can show that the behavior of Ba, Rb and K are consistent with residual amphibole and allow a minor role, at most, for residual phlogopite. High-quality experimentally derived trace element partitioning data highlight the contrasting effects of phlogopite and amphibole on the compositions of liquids with which they equilibrate. Dalpé & Baker (2000) found that partition coefficients between amphibole and alkali basalt melt at mantle pressures (1·5–2·5 GPa) are several times higher for K than for Rb and Ba. In contrast, phlogopite–melt partition coefficients are similar for all three elements (LaTourrette et al., 1995). Rb and Ba are both less depleted and more variable than K in the undersaturated Santa Fe Group lavas (Fig. 7), consistent with amphibole being the dominant residual LILE-bearing phase rather than phlogopite. Illustrative results from partial melting calculations using assemblages with amphibole and phlogopite are shown in Fig. 7. We conclude that the most likely origin of these magmas is low-degree partial melting of amphibole-bearing mantle.
Amphibole is stable under the P–T conditions of the lithospheric mantle, but not in the convecting upper mantle or upwelling plumes (Class & Goldstein, 1997, and references cited therein). Therefore, the presence of amphibole in the mantle source region for the early Española basin nephelinites and basanites is evidence for melting of the subcontinental mantle lithosphere. Essentially the same magma type, modified by crustal contamination, appears to have been available for much of the lifetime of the JMVF (Wolff et al., 2000, and below). Hence, the mantle lithosphere appears to have been a major source of mafic magma during construction of the volcanic field. It should be noted that the K depletion is present in both groups of primitive undersaturated lavas. Therefore, we conclude that the nephelinite–basanite–alkali basalt magmas, which form an important parental magma type throughout the history of the JMVF, are derived from mixed lherzolite–pyroxenite amphibole-bearing lithospheric mantle. Given the tectonic setting of the volcanic field, Proterozoic oceanic lithosphere associated with the Jemez lineament suturezone is the likely source for these magmas. This conclusion is consistent with that of McMillan (1998), who invoked basalt-veined lithospheric mantle as the source for Española Basin lavas, and the Hf isotope evidence presented by Johnson & Beard (1993) for a component of ancient suboceanic mantle in Cenozoic lavas of the Rio Grande rift and surrounding region.
Olivine tholeiite
Unlike the silica-undersaturated lavas, there are no primitive compositions among the hy-normative basalts; none have MgO >8%, and all exhibit features consistent with at least minor crustal contamination. The latter is probably a consequence of lower incompatible trace element contents and hence sensitivity of the tholeiitic liquids to contamination, compared with the silica-undersaturated magmas. The Santa Fe Group basaltic andesites have higher 143Nd/144Nd (0·51280–0·51272) than Lobato and Cerros del Rio tholeiites (0·51273–0·51262) despite higher SiO2 (Duncker et al., 1991; Gibson et al., 1993), and are less contaminated with crust. Abundances of compatible and incompatible trace elements are similar in tholeiites regardless of age. The normalized trace element patterns show mild depletions of Nb and Ta relative to LILE of similar compatibility (Figs 4a and 8), consistent with minor contamination, but lack the strong K depletion that characterizes the silica-undersaturated primitive magmas. We concur with Duncker et al. (1991) that the most likely parental basalt composition resembles E-MORB or tholeiitic OIB, similar to the MORB-like tholeiitic parent invoked by Dungan et al. (1986) as the mantle-derived component in the crustally contaminated Pliocene Servilleta lavas of the Taos Plateau (Fig. 1). This is consistent with the Sr and Nd isotopic composition of the least enriched of the early tholeiites (87Sr/86Sr = 0·7038, 143Nd/144Nd = 0·51280, Gibson et al., 1993), which lies within the E-MORB range. The convecting upper mantle is therefore a plausible source region for the parents of these magmas. None the less, the general Sr and Nd isotopic similarity to the silica-undersaturated magmas allows the parental magmas of the tholeiites to be partial melts of the same mantle, without residual amphibole (Fig. 8). Given the lower enrichments of incompatible elements in the tholeiites, this is most easily explained by a higher degree of partial melting with complete consumption of amphibole.
|
The relationship between the two primary magma types and their derivatives can be illustrated by consideration of critical trace element ratios. Using high-quality trace element data, Hofmann et al. (1986)
showed that U/Nb and Pb/Ce ratios in ocean basin basalts, both MORB and OIB, exhibit little variation (U/Nb = 0·022 ± 0·005, Pb/Ce = 0·041 ± 0·009). Both ratios are much higher in average continental crust (U/Nb 0·1, Pb/Ce 0·25). The MORB and OIB values are considered representative of the convecting upper mantle. In Fig. 9, we have plotted U/Nb vs Pb/Ce (to avoid difficulties arising from inter-laboratory discrepancies, especially for U and Pb, only samples analyzed by ICP-MS at WSU are plotted). The data fall into two positively correlated arrays: a lower-U/Nb group defined by the Lobato Basalt, and a higher-U/Nb group dominated by the Paliza Canyon and Cerros del Rio–El Alto hawaiites and mugearites. Most olivine tholeiites from Cerros del Rio and El Alto plot in the former group. Santa Fe Group nephelinites plot at the low-Pb/Ce, low-U/Nb end of the higher-U/Nb group, whereas a Santa Fe Group tholeiite bears the same relation to the low-U/Nb group. The increase in U/Nb and Pb/Ce among both groups is attributed to crustal contamination of mantle-derived magmas (this is justified in the next section), and examples of contamination trajectories are shown in Fig. 9. Also plotted is the limited MORB and OIB range (Hofmann et al., 1986). Keeping in mind that crustal contamination has affected even the most primitive compositions among the Lobato lavas and Cerros del Rio–El Alto tholeiites, MORB–OIB forms a plausible end-member for the lower-U/Nb array. A MORB-OIB end-member cannot satisfy the K–Th–Nb features of the high-U/Nb group, nor can the nephelinite–basanite magmas for the low-U/Nb group (Wolff et al., 2000, and below). The high U/Nb of the Santa Fe nephelinites is consistent with a role as parental magmas for the the Cerros del Rio hawaiites and Paliza Canyon lavas. The convergence of Cerros del Rio hawaiites and mugearites with Lobato basalts at low U/Nb suggests some mixing of components derived from both sources.
|
We have shown that the trace element abundances of the nephelinite–basanite magmas are satisfied by low-degree partial melting of a MORB–OIB-like mantle source composition with residual amphibole (Fig. 7). The scatter among the high-U/Nb group in Fig. 9 may require a contribution from primitive magmas with still higher U/Nb than analyzed nephelinites, or may be the result of mixing with partial rather than bulk melts of crust, exemplified by the energy-constrained AFC curve shown in Fig. 9 (see below). The elevated U/Nb of the most primitive lava could be due to enrichment of the nephelinite–basanite source with LILE prior to magmatism, although if this were due to metasomatism by a fluid, then elevated Pb/Ce might also be expected, yet Pb/Ce in the nephelinite–basanite lavas is in the MORB–OIB range. Modeled partial melts for these magmas (Fig. 7) show an increase in U/Nb over primitive mantle, a consequence of significantly higher amphibole/melt partition coefficients for Nb than for U; Th/Nb is also elevated among this group. Therefore, the U/Nb–Pb/Ce relations are fully consistent with derivation from a MORB–OIB-like mantle source in the presence of residual amphibole. Hence, the contrasts between the two primary magmas (one silica undersaturated, one tholeiitic) can be attributed solely to the degree of melting and the consequent presence or absence of residual amphibole, which requires a lithospheric source for the nephelinite–basanite magmas. Although we cannot rule out an origin for the tholeiites in the convecting upper mantle, the simplest origin for the range of compositions among JMVF mafic magmas is therefore variable partial melting of amphibole-bearing lithospheric mantle with the incompatible element characteristics of asthenosphere, as modified by contamination with continental crust. Again, this is consistent with a source lying in ancient subducted oceanic lithosphere trapped in the regional suture zone beneath the JMVF.
These observations do not preclude an origin for the tholeiites in subduction-modified mantle (e.g. McMillan, 1998) with elevated levels of Pb and other strongly fluid-mobile elements. However, the extent of crustal contamination required to produce the entire range of Lobato Basalt and similar lavas (lower-U/Nb group, Fig. 9) is 25%, similar to the average amount of crust needed to produce the hawaiites from nephelinite–basanite magmas, and there is no reason to suppose a different crustal transport–residence history for the two groups of magmas.
To summarize, the two distinct groupings of JMVF mafic magmas, hawaiite–mugearite and tholeiite, are most simply explained as the respective products of strongly silica undersaturated and tholeiitic parents produced by different degrees of melting of lithospheric mantle and variably contaminated by continental crust. Crustal involvement increased dramatically with the tempo of magma production during the transition from scattered rift-related volcanism to the onset of construction of the JMVF, as a result of accumulation of heat in the crust with time. In the rest of the paper, we consider crustal lithologies and their interaction with the mantle-derived magmas.
|
EVALUATION OF CRUSTAL COMPONENTS IN JMVF MAGMAS |
|---|
 TOPABSTRACTINTRODUCTIONREGIONAL TECTONIC SETTING,...ANALYTICAL METHODSGENERAL CHARACTERISTICS OF JMVF...IDENTIFICATION AND ORIGIN OF... EVALUATION OF CRUSTAL COMPONENTS... SUMMARY OF MAIN GEOCHEMICAL...REFERENCES |
|---|
Although recognizing that not all crustal lithologies that interacted with mantle-derived JMVF magmas may be available at the surface, we have attempted to constrain possible contaminants by analyzing Proterozoic basement rocks from three areas (Fig. 1): the Taos range,
100 km NE of the JMVF (eight samples of diverse lithologies); around the Española basin (two amphibolites from the Santa Fe Range and two granites from west of the basin); and the Bandelier Tuff (four lithic fragments recovered from the Otowi Member). Our rationale for studying rocks from the Taos range is its position just north of the Jemez lineament combined with the southerly dip of the terrane boundary; hence formations exposed at the surface might have equivalents at depth beneath the JMVF.
Radiogenic isotope and trace element characteristics of regional crustal rocks
The basement lithologies fall into three broad categories: amphibolites (mafic–intermediate igneous protoliths); granitoid (silicic igneous protoliths), and metasediments. Sr, Nd and Pb isotope data for the basement rocks are compared with JMVF volcanics in Figs 10 and 11. Trace element and isotope characteristics are diverse, but we note the following features (Figs 10 and 11; Table 2).
|
|
|
(1) Almost all have LILE/Nb and Pb/Ce ratios well in excess of BSE values.
(2) Siliciclastic metasediment bears a highly distinctive imprint (elevated Zr and Hf) of detrital zircon.
(3) Lithologies represented by mafic amphibolites (Taos and Santa Fe ranges) that we have analyzed are eliminated as components in JMVF magmas by their Pb isotope characteristics.
(4) Lead isotope ratios in three basement fragments recovered from the Bandelier Tuff (two granitoids and an intermediate amphibolite), and a single granitoid from the Taos Range, are similar to the majority of JMVF silicic rocks and hence are candidates for crustal components in the magmas (Fig. 11). However, all four have highly radiogenic Sr, with 87Sr/86Sr >0·724.
How significant are the crustal components in JMVF magmas?
Previous geochemical studies of the JMVF (Singer & Kudo, 1986; Duncker et al., 1991; Wolff et al., 2000) have invoked a significant role for continental crust in JMVF magmas of nearly all compositions. In this section, as a preliminary to identifying the crustal components we briefly show that our much larger dataset supplies additional evidence for the presence of abundant crustally derived material.
Petrographic evidence
The Cerros del Rio hawaiites, mugearites and benmoreites contain unequivocal evidence for the presence of assimilated crust, in the form of rounded quartz xenocrysts, up to 2·5 mm in diameter, that have 
18O values up to 3
higher than the host rock (Duncker et al., 1991). Some of the quartz xenocrysts exhibit internal microcataclastic zones and oriented rutile needles, indicative of a metamorphic origin. The presence of large quartz grains requires assimilation of granitoid or metasedimentary rock. Quartz xenocrysts and quartzite fragments also occur in Paliza Canyon and Lobato basalts.
Geochemical evidence
All JMVF formations show positive correlations between LILE/Nb, Pb/Ce and indices of fractionation such as SiO2 (Fig. 12), despite the inherent inability of the observed phenocryst assemblages (ol ± cpx ± opx ± plag ± Fe–Ti oxides) to induce significant fractionation of K and U from Nb, and of Pb from Ce. The correlation between K/Nb and SiO2 extends smoothly from the Cerros del Rio and El Alto hawaiites/mugearites through Paliza Canyon into the silicic lavas of the Keres and Polvadera Groups (Fig. 12a). K/Nb ratios of basement rocks extend up to 10 000. Pb/Ce vs SiO2 shows closely similar behavior (Fig. 12b); Pb/Ce ratios of basement rocks reach 1·69. K/Nb is negatively correlated with 143Nd/144Nd, both overall and among individual formations (Fig. 13). We conclude that the LILE geochemistry of JMVF pre-caldera rocks, including andesites and more silicic compositions that are not treated in detail here, is dominated by crustal additions to mantle-derived magmas.