U-Pb geochronology of 1.1 Ga diabase in the southwestern United States: Testing models for the origin of a post-Grenville large igneous province

Voluminous mafic magmatism can be a manifestation of continental breakup through rifting, or it can be an expression of the interaction of mantle plumes with continents, whether or not it leads to the formation of a new ocean basin (Halls and Fahrig, 1987; Ernst et al., 2001; Ernst and Buchan, 2003). Regional mafic dike and sill swarms are significant because they are associated with large igneous provinces that rapidly add significant volumes of mafic-ultramafic rock to the crust, such as in the Deccan, Siberian, and Karoo magmatic provinces (Cox, 1988; Zolotukhin and Al’Mukhamedov, 1988; Walker, 1993; Ernst et al., 2008), and therefore strongly influence crustal composition and architecture. Large igneous provinces are also important because they represent nearly instantaneous magmatic events over extremely large areas and therefore can be used as geologic markers to constrain continental reconstructions (Ernst et al., 2001; Harlan et al., 2003; Bleeker and Ernst, 2006). Large igneous provinces occur at an average rate of about one every 20–30 m.y., since the Early Proterozoic, and the North American record is particularly well understood (e.g., Ernst and Bleeker, 2010).

In North America, voluminous late Mesoproterozoic large igneous province magmatism is represented by the 1.27 Ga Mackenzie dike swarm, associated sills, remnant volcanic rocks, and layered intrusions of northern Canada (e.g., Fahrig, 1987; Baragar et al., 1996), the 1.115–1.085 Ga Keweenawan large igneous province of the Midcontinent rift (e.g., Davis and Paces, 1990; Hollings et al., 2010), and a broad region of 1.1 Ga mafic magmatism in the southwestern United States; the latter is the focus of this paper.

The “southwestern U.S. diabase province” (Fig. 1) was originally defined based on exposures of ca. 1.1 Ga basalt intrusions and flows in Arizona and southeastern California (Hammond, 1990; Howard, 1991). Dikes, sills, and sheets, or horizontal intrusions cutting existing fabrics, comprise a large igneous province across the region that was inferred to be broadly coeval with Mesoproterozoic sedimentary rocks in the Grand Canyon (e.g., Wrucke, 1989) and silicic intrusions such as the Pikes Peak granite in Colorado (Scharer and Allegre, 1982; Smith et al., 1999) and Aibo granite in northern Sonora, Mexico (Anderson and Silver, 2005; Farmer et al., 2005). The published ages of the mafic rocks range from 1040 to 1140 Ma (Table 1; also see summary in Wrucke, 1989), but high-precision U-Pb ages are rare (Heaman and Grotzinger, 1992) or only published in abstract form (e.g., Shastri et al., 1991).

Geochronology of mafic rocks using baddeleyite (ZrO₂) is becoming more common (e.g., Söderlund et al., 2013), and there is a systematic effort being applied to diabase dike and sill swarms around the world funded by an industry consortium (www.supercontinent.org), wherein more than 150 U-Pb baddeleyite ages have been generated since 2010 (Ernst et al., 2013a). Previous U-Pb baddeleyite dating has been applied to ca. 1.1 Ga intraplate mafic magmatism in Australia (Warakurna large igneous province; Wingate et al., 2004) and in southern Africa (Unkondo large igneous province; Hanson et al., 2004); both events are broadly similar in age to the southwestern U.S. diabase province. These studies have found a generally limited range of ages, and both large igneous provinces have been interpreted as being derived from a mantle plume source, as was the Keweenawan large igneous province of the Midcontinent rift (Hutchinson et al., 1990; Nicholson and Shirey, 1990).

For this study, we obtained high-precision U-Pb ages on baddeleyite from diabase in the southwestern U.S. diabase province to refine the timing of this event. We also obtained low-precision ages on single zircons that allow us to extend the limits of this province eastward into New Mexico. We address possible correlations with mafic magmatism in west Texas and the relationship of this event to the Grenville orogeny and to coeval silicic magmatism from northern Mexico to Colorado.

The ca. 1.1 Ga magmatism in the southwest United States occurred after a voluminous granite-rhyolite event in Laurentia that ranges in age from 1480 to 1380 Ma in which plutons of ca. 1.4 Ga age intruded ca. 1.8 Ga host rocks of the Mojave Province and 1.7–1.6 Ga rocks of the Yavapai and Mazatzal Provinces (Anderson, 1983; Karlstrom et al., 1987, 1990, 2001; Anderson and Morrison, 2005; Amato et al., 2011). The Grenville orogeny occurred from 1.3 to 1.1 Ga (Tollo et al., 2004) and involved collisions related to the assembly of the supercontinent Rodinia (Li et al., 2008). In the southwest United States, magmatism synchronous with Grenville tectonism (Fig. 1A) occurred until at least 1120 Ma (Mosher et al., 2004), though ⁴⁰Ar/³⁹Ar thermochronology from metamorphic rocks in west Texas has been interpreted to indicate Grenville deformation lasted until 1000–980 Ma (Grimes and Copeland, 2004). Post-tectonic magmatism in the Grenville province locally occurred in the Llano Uplift from 1092 to 1070 Ma (Mosher, 1998, and references therein).

As originally recognized and named by Hammond (1990), the southwestern U.S. diabase province (Fig. 1B) extends from southeastern California to southwestern Arizona (Howard, 1991). Barnes et al. (1999) grouped the Pikes Peak batholith into the province, along with mafic igneous rocks in west Texas. Our data confirm that diabase intrusions from western New Mexico also belong to this igneous province, which we now refer to as the Southwestern Laurentia large igneous province (SWLLIP). Throughout this region, mafic (diabase) dikes, sills, and horizontal sheet intrusions cutting existing fabrics have all been linked to this event, as have feeder dikes to sills (Wrucke, 1989; Hammond, 1990; Howard, 1991). The Cardenas basalt flow in the Grand Canyon provides evidence for eruptions around the same time period (Larson et al., 1994; Weil et al., 2003; Timmons et al., 2005). The diabase has been interpreted as intruding at crustal depths of up to 15 km (Howard, 1991), and the thickness of the intrusions ranges from a few meters to 400 m (Shride, 1967). Phanerozoic tectonic events, particularly Basin and Range extension, have locally changed the original orientation of some intrusions (Howard, 1991). However, several of the unaffected dikes in Arizona consistently strike northwest (Conway et al., 1993), potentially representing a major swarm associated with this magmatic province. The Bagdad Reflector Sequence in Arizona is inferred to represent mafic sills (Clayton, 1989; Litak and Hauser, 1992), and upper-crustal sills are present in the Texas Panhandle (Li et al., 2007).

The diabase intrusions are medium to coarse grained with subophitic textures, and fine-grained to glassy chilled margins are common (Wrucke, 1989). Mineralogy includes plagioclase + augite ± olivine ± orthopyroxene ± titanite + Fe-Ti oxides + apatite. The rocks are low-K basalts with SiO₂ generally ranging from 45 to 49 wt% and Mg# (molar MgO/[MgO + FeO]) ranging from 70 to 80 (Hammond, 1990). Trace-element signatures are thought to represent closed-system processing with minimal host-rock interactions (Nehru and Prinz, 1970; Hammond, 1986, 1990). Some intrusions, however, have high Rb and Zr contents, possibly reflecting crustal assimilation or low degrees of partial melting (Hammond, 1990). Alkalic diabase has incompatible element concentrations similar to alkalic ocean-island basalts, whereas the tholeiites are more similar to island-arc basalts (Hammond, 1990). However, many other large igneous provinces also have arc signatures, notably, negative Nb and Ta anomalies (Puffer, 2001), which may represent interaction of large igneous province magmatism with lithosphere that had been affected by a prior subduction event (e.g., Neumann et al., 2011).

The few previously published ages for diabase from this province range from 1140 Ma to 1040 Ma (Table 1). They were obtained through a variety of whole-rock and multicrystal analytical techniques, including thermal ionization mass spectrometry (TIMS) dating of zircon, baddeleyite, and titanite, ⁴⁰Ar/³⁹Ar dating of whole-rock basalt, K/Ar dating of biotite, and Rb/Sr isochron dating of basalt. Many of these ages are imprecise, and others were published only in abstract form. The most reliable ages include upper-intercept U-Pb TIMS ages on baddeleyite from two sills in Death Valley, California, at 1087 ± 3 Ma and 1069 ± 3 Ma (Heaman and Grotzinger, 1992). Note that the sample yielding the younger age has zircon rims growing on the baddeleyite crystals, with a lower-intercept age of 65 Ma, interpreted as dating the zircon growth, and the upper-intercept age representing an estimate for the emplacement age. Zircon rims on baddeleyite in West African dikes have been shown to yield an age that is too young by ∼20 m.y. (Söderlund et al., 2013); it should be considered whether the 1069 Ma age for the Death Valley sill may be similarly too young. Shastri et al. (1991) reported (in an abstract) TIMS ages of 1100 ± 2 Ma of baddeleyite and zircon from Sierra Ancha in central Arizona and 1080 ± 20 Ma of titanite from the Peacock Mountains in northern Arizona. The ⁴⁰Ar/³⁹Ar date of contact metamorphic biotite adjacent to the Cardenas basalt in the Unkar Group of the Grand Canyon yielded an age of 1103 ± 10 Ma (recalculated by Heizler [2013, personal commun.] fromWeil et al. [2003] and Timmons et al. [2005]). A mafic sill was encountered in drill core in the northern Texas Panhandle region, and this intrusion was named the Panhandle igneous complex (Barnes et al., 2002). Compositions range from olivine gabbro to a quartz diorite, which has a U-Pb age of 1080 ± 17 Ma (Li et al., 2007). It is more ambiguous as to whether the mafic and ultramafic intrusions that make up the Pecos mafic intrusive complex should be included as part of the Southwestern Laurentia large igneous province. We agree with Barnes et al. (1999) that one of the ages from the Pecos mafic intrusive complex, 1163 ± 4 Ma from a single zircon (Keller et al., 1989), places the intrusion well outside of the range of ages of the other diabase samples. However, Keller et al. (1989) also obtained an age of 1095 ± 14 Ma from titanite in another sample. Future studies may resolve the ambiguity resulting from only two dates from rocks that were derived from small-volume drill-core samples without geologic context.

Coeval silicic magmatism (Table 1) is known from a few localities in northern Sonora, Mexico, where granites have ages ranging from 1104 to 1075 Ma, and southwestern New Mexico, which has a granite dated at 1220 Ma (Rämö et al., 2003) and a rapakivi granite at 1077 ± 4 Ma (Amato and Mack, 2012). In west Texas, granites (the Red Bluff granitic suite) with associated mafic rocks have U-Pb zircon ages of 1125–1100 Ma (Shannon et al., 1997; Li et al., 2007; Howard, 2013), and in the syntectonic Llano Uplift, granites have been dated between 1119 and 1116 Ma, and undeformed, post-tectonic granites have been dated from 1092 to 1070 Ma (Walker, 1992; Mosher, 1998, and references therein; Mosher et al., 2008; Barker and Reed, 2010). The Pikes Peak batholith of central Colorado has granites and syenites with ages ranging from 1098 to 1085 Ma (Schärer and Allegre, 1982; Smith et al., 1999; Howard, 2013), as well as undated gabbro and diabase dikes.

Magmatism in this area has been previously interpreted as a large igneous province on the basis of spatial extent, mafic composition, and intraplate setting (e.g., Ernst and Buchan, 2001; Ernst et al., 2008). In addition, the presence of associated silicic magmatism is a common feature in large igneous provinces, wherein the silicic magmas are inferred to be derived from partial melting of fusible lower crust by mafic underplating (Bryan and Ernst, 2008; Bryan and Ferrari, 2013). The tectonic setting of the diabase province has been proposed to be dominantly extensional (Dickinson, 1989; Hammond, 1990; Conway et al., 1993; Barnes et al., 1999). Howard (1991) noted that the occurrence of horizontal sheets and sills is more consistent with a compressive or isotropic stress state, and that magmatism may be related to a plume. Conway et al. (1993) and Timmons et al. (2005) suggested that 1.1 Ga graben formation, sedimentation, and magmatism were an extensional response to Grenville shortening, with other workers suggesting they occurred in response to collision with a continental indentor (Adams and Keller, 1995; Mosher, 1998).

Unlike large igneous provinces such as the 200 Ma Central Atlantic magmatic province associated with the opening of the Atlantic Ocean, the Southwestern Laurentia large igneous province was not a precursor to continental breakup, as the supercontinent Rodinia did not disassemble until the middle Neoproterozoic, over 300 m.y. later. Other major large igneous provinces, such as the Siberian Traps, did not lead to breakup, nor did the coeval Keweenawan large igneous province—the former probably because of compressive stresses related to Pangea assembly (Ernst, 2014) and the latter because of stresses related to the Grenville orogeny (Tollo et al., 2004).

The ca. 1.1 Ga dikes and sills in Arizona have been the subject of several paleomagnetic studies evaluating the polarity of magnetization, the presence of reversals, and asymmetry of reversals, and these data are compared to results from the Keweenawan region (Heaman et al., 2007). Harlan (1993) noted that samples from Arizona yielded both normal and reverse polarities and that the similarity of the pole location to that from the Keweenawan samples indicated that the two areas experienced “essentially synchronous” magmatism, a conclusion also reached by Weil et al. (2003) based on samples from the Grand Canyon. Donadini et al. (2011) noted that the two polarities in Arizona were asymmetric, similar to that exhibited by Keweenawan rocks (Pesonen and Nevanlinna, 1981).

Some previously reported ages may be accurate but imprecise, limiting their usefulness in assessing the relationship between Southwestern Laurentia large igneous province magmatism and the Grenville orogeny. For our interpretations, we do not use ages procured through any method with 2σ uncertainties of 20 m.y. or greater (Table 1). Other ages were obtained by techniques such as K/Ar or Rb/Sr dating that are not generally considered reliable. In addition, other ages have only been published in abstract form with no analytical data. These are briefly reviewed here and result in a narrower range for Southwestern Laurentia large igneous province magmatism than the 1140–1040 Ma range that is possible using all of the published ages regardless of their quality.

These filters eliminate Rb/Sr whole-rock dates from the Cardenas basalt flow (Elston and McKee, 1982; Larson et al., 1994), a low-precision ⁴⁰Ar/³⁹Ar date on diabase from New Mexico (Strickland et al., 2003), and two older K/Ar biotite dates because it is not possible to address Ar loss or other complexities associated with this technique (Damon et al., 1962; Banks et al., 1972). A K/Ar biotite date (Damon et al., 1962) of 1140 Ma at Sierra Ancha is much older than the Silver (1978) TIMS zircon date of 1120 ± 10 Ma, though the analytical data were not included for this date or the 1100 ± 15 Ma date from the Little Dragoon Mountains in the same publication. The 1100 ± 2 Ma date (inShastri et al., 1991, abstract, no data) was described as having an inherited component.

The two TIMS dates from the Pecos ultramafic complex are precise but differ by nearly 70 m.y., thus creating ambiguity about the timing of magmatism in this intrusion only sampled via drill core, and the 1163 Ma date is from only a single zircon (Keller et al., 1989). A laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) U/Pb zircon date of 1066 ± 8 Ma from a biotite granite from the Little Hatchet Mountains in New Mexico (McLemore et al., 2012) was only published in abstract form and did not date the diabase itself, which was interpreted as mingling with the host granite. Finally, a high-precision U/Pb date of 1069 ± 3 Ma on baddeleyite from a sill in southeastern California (Heaman and Grotzinger, 1992) is likely too young, based on 11%–19% discordance, interpreted by the authors as resulting from Pb loss from alteration of the baddeleyite crystals or younger zircon coatings on the baddeleyite.

Geochemical analyses were performed using a Rigaku X-ray fluorescence (XRF) mass spectrometer (major and trace elements) at New Mexico State University. Samples were crushed with a tungsten carbide shatterbox. Reference materials (BHVO-1 and BHVO-1P) were measured before and after all unknown analyses. Accuracy on major-element concentrations is generally <1.1%. Sr and Nd isotopic compositions of selected samples were obtained by isotope dilution–thermal ionization mass spectrometry (ID-TIMS) at the University of Colorado at Boulder. Rock powder was dissolved in HF and perchloric acid in Teflon beakers. Rb, Sr, Sm, and Nd were separated, and their isotopic compositions were determined using the techniques described in Farmer et al. (1991).

U-Pb geochronology was carried out on zircon and baddeleyite. Zircons were separated from ∼5–10 kg of sample using a jaw crusher and disc grinder and sieved to 40 mesh (∼400 µm), followed by separation using a Gemeni table and a magnetic separator, taking the nonmagnetic fraction at 1.8 A with a 20° slope. This split was put into sodium polytungstate (2.85 g/cm³) and methylene iodide (3.3 g/cm³). The high-density fraction was put into nitric acid if sulfides were present; otherwise, zircons were handpicked and mounted for analysis.

Sensitive high-resolution ion microprobe–reverse geometry (SHRIMP-RG) dating of zircon from igneous rocks was conducted at the Stanford–U.S. Geological Survey Ion Probe Facility. The primary beam excavated an area of ∼25–30 µm across to a depth of ∼1 µm. The analytical routine followed Williams (1998) and Strickland et al. (2011). The SQUID program was used for data reduction (Ludwig, 2005). Isotopic compositions were calibrated by replicate analyses of zircon standard R33, which has an age of 419 Ma (Black et al., 2004). The ²⁰⁶Pb/²³⁸U ages were corrected for common Pb using the ²⁰⁷Pb method. Common Pb compositions were estimated from Stacey and Kramers (1975).

Laser ablation–multicollector (LA-MC) ICP-MS dating was conducted on zircons from igneous rocks at the Arizona Laserchron Laboratory at the University of Arizona. Beam diameter was generally 30–40 µm. Errors on spot ages of individual zircon grains are reported in the text and tables at 1σ, and we report weighted mean ages in the text and figures at the 2σ level. Data are presented on concordia diagrams using Isoplot (Ludwig, 2003).

ID-TIMS was used to date baddeleyite. Baddeleyite (ZrO₂) was extracted from the samples by Ulf Söderlund at Lund University using the method described in Söderlund and Johansson (2002), in which a heavy-mineral concentrate of the powdered rock is obtained using a Wilfley table, and then baddeleyite grains are picked in ethanol under a binocular microscope. Using this technique, several thin, dark-brown baddeleyite crystals and fragments were found for each of the samples, most being less than 100 µm in largest dimension, except for those from sample 09SR-3, which yielded significantly larger crystals. These concentrates were then processed and analyzed at the Berkeley Geochronology Center. Approximate (±50%) weights of individual grains, typically less than 0.5 µg, were estimated using digital photomicrographs. Determining the weight by this method with greater precision was not possible given the thinness of the grains; their smallest dimension was difficult to determine. In any case, this volume estimate was only used to determine absolute amounts of U and Pb, and it does not enter into the age calculations, which were determined by isotopic ratios. The larger crystals from sample 09SR-3 had a frosted appearance, indicating the presence of a coating of microcrystalline zircon. This is generally undesirable because it indicates a high likelihood of Pb loss from the fine zircon coatings, but every crystal was coated, and therefore this could not be the basis for excluding any from analysis. The presence of any zircon coating was difficult to assess for the smaller baddeleyite crystals. The crystals were divided into individual fractions of one or more grains, then washed in HCl and HNO₃, and transferred individually into miniature capsules with a ²³³U-²³⁵U-²⁰⁵Pb spike for HF vapor transfer dissolution over 6 d at 220 °C. No chemical separation procedures were required. The residue was then taken up in HCl and H₃PO₄, and the samples were loaded on Re filaments with silica gel. Isotope ratios were measured sequentially in peak jumping mode using a VG Sector 54 TIMS equipped with a WARP filter. The ion signal was amplified using a Daly detector and a photomultiplier, operated in pulse counting mode. Mass fractionation for Pb was corrected by 0.05%/AMU ± 0.09 based on repeated analyses of NBS 981. U decay constants are from Jaffey et al. (1971). Data were calculated and plotted using Isoplot (Ludwig, 2003).

We also consider whether the ²⁰⁷Pb/²⁰⁶Pb ages, rather than the ²⁰⁶Pb/²³⁸U ages, might better approximate the intrusive age of the samples. Large uncertainties (particularly in the ²⁰⁷Pb/²³⁵U ratio) for individual analyses result in relatively imprecise ²⁰⁷Pb/²⁰⁶Pb ages, as well as high probabilities of fit and low values of mean square weighted deviation (MSWD). The ²⁰⁷Pb/²⁰⁶Pb age incorporates all of the data, but it will always be older than the oldest ²⁰⁶Pb/²³⁸U age for these rocks. Although it is apparent that Pb loss has occurred in many of the baddeleyite crystals analyzed, the discordia line through the origin that defines the ²⁰⁷Pb/²⁰⁶Pb age is nearly parallel to Concordia, and it is therefore impossible to accurately assess the degree of Pb loss. However, for each sample, multiple fractions have the same ²⁰⁶Pb/²³⁸U age, and the likelihood that multiple fractions have lost the same amount of Pb is low. Therefore, we prefer to quote the weighted mean of the oldest coherent grouping of ²⁰⁶Pb/²³⁸U ages rather than the older ²⁰⁷Pb/²⁰⁶Pb ages. The data from each sample include a coherent cluster (n ≥ 3) of oldest concordant analyses; we report the ²⁰⁶Pb/²³⁸U ages for these clusters as well as the ²⁰⁷Pb/²⁰⁶Pb ages incorporating all data for each sample. Multiple baddeleyite crystals (up to four) were included in fractions from all samples except for sample 09SR-3. This was required due to the small size of the crystals and their relatively low U concentrations (often <100 ppm U). All uncertainties are reported at the 2σ level.

Diabase dikes and sills were sampled from localities in Arizona and southern New Mexico (Table 2). Most exposures were sampled from the center and at or near the margin of each intrusion. Characteristic features of the dikes are shown in Figure 2.

The Hualapai Mountains of northwestern Arizona are located in the eastern Mojave Province and expose ca. 1.8 Ga metamorphic rock and younger plutons (Stensrud and More, 1980; Siwiec, 2004). Both dikes and sills of diabase are present. The sample (09HU-14) is a fresh, holocrystalline equigranular diabase from a 6-m-thick sill. Mineralogy includes plagioclase + augite + olivine + minor biotite. Plagioclase laths are up to 3 mm long. The sample has 50.1% SiO₂, making it the most silicic sample we analyzed, and an Mg# of 79 (Table 3). It has an initial ⁸⁷Sr/⁸⁶Sr ratio (at 1085 Ma) of 0.7034 and ε_Nd of −1.4, which is the most evolved isotopic composition of all of the samples (Table 4).

The diabase sill at Sierra Ancha in central Arizona has been extensively studied and previously dated using a variety of techniques (Nehru and Prinz, 1970; Smith and Silver, 1975; Silver, 1978; Shastri et al., 1991). The region is part of the southern Yavapai Province. Sills up to 400 m thick intrude both below and within the Pioneer Formation, a middle Proterozoic clastic unit (Skotnicki, 2002). The sample we studied (09SANC-8) is from a sill (thickness not recorded), and petrography reveals plagioclase laths up to 3 mm in length and significantly more olivine than other localities, with 2-mm-wide crystals comprising up to 40% of the rock. Clinopyroxene subophitically surrounds plagioclase and olivine. Some seritization of plagioclase has occurred. The sample has 47.8% SiO₂ and an Mg# of 69 (Table 3). Its ⁸⁷Sr/⁸⁶Sr composition is unreliable because of alteration, but its ε_Nd is 3.2 (Table 4).

Salt River Canyon is relatively close to Sierra Ancha and is part of the Yavapai Province. The diabase has a similar occurrence, forming thick sills intruding Mesoproterozoic Apache Group sedimentary units (Wilson et al., 1959; Davis et al., 1981). The first sample (09SR-3) was collected from a 5-m-thick sill. Two additional samples were collected from a different sill that was 5 m thick, with sample 09SR-6m from the middle of the sill and 09SR-6t collected near the upper margin. All samples consist of plagioclase up to 3 mm in length, augite, and olivine, which comprises ∼20% of the rock. Sample 09SR-3 has 48.6% SiO₂ and an Mg# of 79 (Table 3). Samples 09SR-6m and 09SR-6t have 47.7% and 46.6% SiO₂, respectively, and Mg# values are both ∼71. Sample 09SR-6t was analyzed for isotopic composition. It has an initial ⁸⁷Sr/⁸⁶Sr of 0.7043, and its ε_Nd of 4.7 makes it the most isotopically primitive of all of the samples in this study (Table 4).

The sample collected in a metamorphic core complex in the Santa Catalina Mountains near Oracle, Arizona (3-4-07-1), is from a 15-m-thick dike that cuts the ca. 1.4 Ga Oracle granite in the Mazatzal Province (Force, 1997). This sample was not dated but was analyzed for geochemistry. The rock is medium grained with 45% plagioclase, 45% clinopyroxene, and 10% olivine. It has 48.7% SiO₂, Mg# of 77, initial ⁸⁷Sr/⁸⁶Sr of 0.7041, and ε_Nd of 0.9 (Tables 3–4).

The Pinaleño Mountains of southeastern Arizona expose a metamorphic core complex in the Mazatzal Province. Dikes and larger intrusions of diabase cut Proterozoic granite. Sample 09PL-5 was collected from the center of a large (∼1 km by 2 km) intrusion, is medium to coarse grained, has plagioclase up to 4 mm long, and contains 30% olivine, and clinopyroxene subophitically encloses both plagioclase and olivine. It contains 46.9% SiO₂ and has Mg# of 83, the highest in this study (Table 3). It has an atypically high Rb concentration of 179 ppm, likely the result of alteration, and thus its initial ⁸⁷Sr/⁸⁶Sr is unreliable. It has ε_Nd of 1.4 (Table 4).

Two localities in southern New Mexico were sampled for this study. In the Burro Mountains of the southern Mazatzal Province, diabase is found as northwest-striking dikes cutting Paleoproterozoic metamorphic rocks (e.g., Hedlund, 1980; Amato et al., 2008) and 1460 Ma granite (Amato et al., 2011). Most of the dikes dip <45°NE, which, after restoration of post-Cretaceous tilting, indicate an originally subhorizontal orientation (Dimitroff et al., 2013). Samples BDB-1 and BDB-7 are fine to medium grained with 47% SiO₂ and Mg# = 77 (Table 3). These two samples have initial ⁸⁷Sr/⁸⁶Sr values of 0.7032 and 0.7031 and ε_Nd of 0.8 and 2.0, respectively (Table 4). Sample 03BM-88, also from the Burro Mountains, is fine grained with SiO₂ of 48.1% and Mg# = 76. Its initial ⁸⁷Sr/⁸⁶Sr value is lower than 0.702, suggesting alteration. Although all three samples have Sr concentrations of 244–284 ppm, the Rb concentration of 26 ppm in sample 03BM-88 is more than twice the concentrations of 8–9 found in the other two samples. Thus, addition of Rb via postemplacement processes has likely affected the ⁸⁷Sr/⁸⁶Sr ratio of this sample. The ε_Nd of sample 03BM-88 is −0.2.

The easternmost sample in New Mexico is from the Mazatzal Province at Antelope Hill, located 5 km east of the Organ Mountains. This sample (05SA-15a) is from an ∼20-m-wide dike cutting an early Mesoproterozoic granite (Seager, 1981). Most of the dikes in the region strike northeast (Seager, 1981), and this dike strikes N60E and dips 65°NW. It consists of plagioclase and clinopyroxene with minor biotite and no olivine. It has SiO₂ of 47.1% and Mg# of 67, the lowest of all of the samples in this study (Table 3).

Eight diabase samples were dated by U-Pb for this study. Four yielded zircons, and these were dated with low-precision single-crystal methods (SHRIMP and LA-MC-ICP-MS). Four others yielded baddeleyite, and these were dated by the ID-TIMS method.

A sill at Salt River Canyon, Arizona (sample 09SR-6t), contains zircon fragments 100 µm wide, or as smaller crystals ∼50 µm wide and 100 µm long. Most zircons have an irregular shape with embayed margins. Cathodoluminescence images show a weak zonation in most crystals and an oscillatory zonation in the larger crystal fragments. No xenocrystic cores were observed. Of the 32 analyses (Table 5), one was likely xenocrystic, as it had a U/Th of >80 and yielded a ²⁰⁷Pb/²⁰⁶Pb age of 1461 Ma (Fig. 3). The other analyses had U/Th of 0.5–1.5 and are all concordant around 1.1 Ga. A concordia age of the oldest 26 analyses is 1111 ± 3 Ma, the upper concordia intercept is 1114 ± 7 (with a lower-intercept age within error of 0 Ma), and the weighted mean of the oldest ²³⁸U/²⁰⁶Pb ages (n = 21 out of 31 total analyses) is 1114 ± 13 Ma (MSWD = 0.4). Zircons from the middle of the same sill (sample 09SR-6m) were analyzed by SHRIMP. The intact grains were euhedral and typically 50 µm wide and 100 µm long, had U/Th of 0.5–1.1, and had minor zonation. Three of the analyses were discordant and/or younger than the main population (n = 4 out of 7), which had a weighted mean ²³⁸U/²⁰⁶Pb age of 1090 ± 5 Ma (MSWD = 0.1).

The sample from Antelope Hill in southern New Mexico (09SA-15a) yielded only five zircons, four of which were apparently xenocrystic, as they yielded three ²⁰⁷Pb/²⁰⁶Pb ages ranging from 1411 to 1440 Ma and one at 1760 Ma. The remaining analysis had a ²⁰⁷Pb/²⁰⁶Pb age of 1053 ± 52 Ma (Table 6).

A sample from the Burro Mountains in southwestern New Mexico (03BM-88) yielded 11 zircons. Backscattered electron images of Burro Mountains sample 03BM88 indicates that both zircon and baddeleyite are present as separate phases. Zircons were small and subhedral. Ten of the analyses were in the 1.6–1.8 Ga age range (Table 6). One was only 83% concordant with a ²³⁸U/²⁰⁶Pb age of 816 ± 24 Ma, and it had a ²⁰⁷Pb/²⁰⁶Pb age of 973 ± 41 Ma. Another was concordant with a ²⁰⁷Pb/²⁰⁶Pb age of 1113 ± 48 Ma. The zircon yielding a concordant 1.1 Ga age is 100 µm by 30 µm, has concentric oscillatory zonation, and has rounded, irregular edges suggesting resorption.

For the Pinaleño Mountains sample (09PL-5), two to four baddeleyite crystals were included in each fraction, with a total of six fractions analyzed (Fig. 4; Table 7). The ²⁰⁷Pb/²⁰⁶Pb age for all data is 1087 ± 5 Ma (n = 6, MSWD = 0.09, probability of fit = 99%). There is a coherent cluster of three oldest concordant data with a weighted mean ²⁰⁶Pb/²³⁸U age of 1080 ± 3 Ma (n = 3, MSWD = 1.4).

For sample 09HU-14 from the Hualapai Mountains, nine fractions of three baddeleyite crystals each were analyzed (Fig. 4; Table 7). These crystals had low U concentrations (typically <100 ppm), and this is reflected in relatively large uncertainties of Pb/U ratio determinations, particularly the ²⁰⁷Pb/²³⁵U ratio (between 2% and 18%, 2σ). This in turn produces large uncertainties in the ²⁰⁷Pb/²⁰⁶Pb age of 1093 ± 17 Ma (n = 9, MSWD = 0.45, probability of fit = 88%). Again, all of the data overlap concordia, and the oldest coherent cluster yields a weighted mean ²⁰⁶Pb/²³⁸U age of 1088 ± 3 Ma (n = 3, MSWD = 0.53).

The baddeleyite crystals from Salt River Canyon sample 09SR-3 were significantly larger than those from other samples and had higher U concentrations (between ∼100 and 250 ppm U), and so were measured as single grains in five fractions (Fig. 4; Table 7). The ²⁰⁷Pb/²⁰⁶Pb age of all data combined is 1089 ± 6 Ma (n = 5, MSWD = 0.12, probability of fit = 97%). The three oldest data have a weighted mean ²⁰⁶Pb/²³⁸U age of 1079 ± 4 Ma (n = 3, MSWD = 1.2); the continuous trend toward younger ages as a result of Pb loss makes it difficult to assess whether or not it is certain they are of a single population. The two oldest data for 09SR-7 overlap completely within their individual uncertainties and yield a weighted mean ²⁰⁶Pb/²³⁸U age of 1080 ± 2 Ma (n = 2, MSWD = 0.04).

From Sierra Ancha, sample 09SANC-8 was analyzed using three baddeleyite crystals per fraction (Fig. 4; Table 7). U concentrations were generally low (∼100 ppm U), resulting in larger uncertainties in the ²⁰⁷Pb/²³⁵U ratio (between 2% and 5%, 2σ). Including all data gives a ²⁰⁷Pb/²⁰⁶Pb age of 1111 ± 22 Ma (n = 6, MSWD = 0.38, probability of fit = 86%). The ²⁰⁶Pb/²³⁸U measurements were much more precise, and a cluster of the five oldest data gives a weighted mean ²⁰⁶Pb/²³⁸U age of 1094 ± 2 Ma (n = 5; MSWD = 0.86).

It is likely that Pb loss has occurred in baddeleyite crystals in every sample, with the Pb likely diffusing from coatings of microcrystalline zircon. This resulted in younger ²⁰⁶Pb/²³⁸U ages, regardless of whether a fraction consists of multiple or single grains. We consider the ²⁰⁶Pb/²³⁸U weighted mean ages for the oldest analyses from each sample to be a robust age of crystallization. This interpretation is supported by the collinearity of all the data in each sample and the general agreement of the ²⁰⁷Pb/²⁰⁶Pb ages with the older ²⁰⁶Pb/²³⁸U ages. Therefore, for the purposes of our discussion, the ages of the samples are, in order of increasing age: 09SR-3 = 1080 ± 2 Ma; 09PL-5 = 1080 ± 3 Ma; 09HU-14 = 1088 ± 3 Ma; and 09SANC-8 = 1094 ± 2 Ma (Table 8).

Our four new precise TIMS baddeleyite U/Pb ages demonstrate a range of ages from 1094 ± 2 Ma to 1080 ± 2 Ma for the Southwestern Laurentia large igneous province. Further dating may help refine or expand this age range.

This range encompasses the previously published TIMS baddeleyite date of 1087 ± 3 Ma, based on the upper-intercept age of three fractions that range from 4% to 9% discordant, from a sill in southeastern California (Heaman and Grotzinger, 1992). A SHRIMP upper concordia intercept age of 1080 ± 17 Ma (²⁰⁷Pb/²⁰⁶Pb weighted mean age is 1082 ± 10 Ma) was obtained on a quartz diorite from a mafic sill in the subsurface of the panhandle region of Texas (Li et al., 2007) and demonstrates that mafic magmatism of this age is present in Texas. The Timmons et al. (2005) ⁴⁰Ar/³⁹Ar date of 1103 ± 10 Ma from the Grand Canyon is from contact metamorphic biotite growth in sedimentary rocks adjacent to a mafic sill. The uncertainty quoted here is from Heizler (2013, personal commun.) and is larger than that quoted in the initial publication and now includes an estimate of the systematic errors. It is possible that this magmatism could extend the range of ages in the Southwestern Laurentia large igneous province, but there are some uncertainties comparing low-temperature argon ages with U-Pb ages, and thus we are hesitant to use this as a firm constraint on the duration of the event while recognizing it is likely part of the same province as the diabase intrusions.

Hammond (1990) divided the diabase province into the Death Valley area, the Colorado River trough area, and central Arizona, but although the whole-rock and isotopic characteristics of the samples did not vary significantly based on location, she did note that the samples fell into two main groups based on their ε_Nd values. The more primitive group had values around +3 to +4, and the other had values <+2. In our study (Table 4), the majority of samples fell into the less-evolved group, with the exception of the sample from Salt River Canyon, which has an ε_Nd of +4.7. In general, the samples with the highest ε_Nd also had the lowest initial ⁸⁷Sr/⁸⁶Sr, with the most primitive values around 0.7030–0.7036. It is likely that the variation in ε_Nd reflects interaction with Paleoproterozoic crust. The majority of this crust, particularly in central Arizona, consists of juvenile arc rocks, and the remainder is ca. 1.4 Ga granites. The juvenile arc crust may have Nd concentrations and Sm/Nd similar to the younger rocks, so the main effect on the samples would be to keep the whole-rock compositions similar but lower the ε_Nd. The sample at Salt River Canyon thus is likely more representative of the initial magma composition and is similar to basalts derived from enriched mantle such as that which produces ocean-island basalts (OIB) from a plume source. The values determined from samples in this study are similar to values of 3.3 and 4.6 determined on diabase associated with the Red Bluff granites of west Texas (Barnes et al., 1999) but distinct from values reported from the Keweenawan basalts.

The original correlation between the diabase dikes of Arizona and California was made by Wrucke and Shride (1972), but the two comprehensive studies linking these dikes and sills as part of one regional event were made by Hammond (1990) based on geochemistry and Howard (1991) who used field relationships. These studies included localities as far northeast as the Panamint Range and Death Valley in California, north to the Grand Canyon in northern Arizona, and southwest in the Pinaleño Mountains and Little Dragoon Mountains in southwestern Arizona. Hedlund (1978, 1980) noted extensive northwest-striking diabase cutting Mesoproterozoic rocks but not cutting minor Cambrian–Ordovician Bliss sandstone, thus restricting the age of the dikes to between 1.4 and 0.5 Ga. Similarly, Seager (1981) mapped abundant mafic dikes cutting Proterozoic granites in south-central New Mexico that were absent in the overlying Paleozoic section. Argon dating of diabase in northwestern New Mexico demonstrated the presence of 1.1 Ga dikes there (Strickland et al., 2003). Conway et al. (1993) suggested that the province extended into west Texas. Timmons et al. (2005) showed normal faults of Mesoproterozoic age in New Mexico and inferred that undated mafic rocks in south-central New Mexico were ca. 1.1 Ga in age.

Our zircon dating confirms that the mafic intrusions in the Burro Mountains and at Antelope Hill have a similar age as the samples nearby in eastern Arizona, and thus are likely part of the Southwestern Laurentia large igneous province, but the issue of inheritance must be addressed. Whereas TIMS baddeleyite dating did not show evidence for inheritance, the single-crystal U-Pb dating of zircon indicates that diabase contains zircons inherited from country rock.

The sample from Antelope Hill in New Mexico (09SA-15a) had one zircon at 1.76 Ga, which is older than the Mazatzal Province crust it intrudes, but these ages are known from detrital zircon studies of Proterozoic metasedimentary rocks in southern New Mexico (Amato et al., 2008). Three zircons recovered from this dike have ages between 1440 and 1411 Ma, and these are consistent with derivation from the host granite. Only one of the zircons yielded an age similar to the other dikes in this study (1053 ± 52 Ma, 91% concordant), but we provisionally suggest that this set of dikes in the San Andres range is part of the Southwestern Laurentia large igneous province.

The sample from Salt River Canyon (09SR-6t) yielded a much higher proportion of zircons that were ca. 1.1 Ga. Only one of the 32 analyses resulted in an older age of 1.46 Ga. This is a common age of granites in central Arizona (e.g., Anderson, 1989; Skotnicki, 2002), and thus it is also interpreted as a xenocrystic zircon. The other sample from this sill (09SR-6m) had no xenocrystic zircons among the seven that were analyzed.

In the Burro Mountains, the diabase sample (03BM-88) had mostly xenocrystic zircons, with 10 out of 12 analyses yielding ages from 2.0 to 1.4 Ga. The two oldest grains at 2.0 Ga and 1.8 Ga are older than the host Mazatzal Province crust but could have been derived from 1.65 Ga metasedimentary rocks that have detrital zircons of this age (Amato et al., 2008). Five grains have ages around 1.6 Ga, and these ages are common in the Burro Mountains basement rocks (Amato et al., 2008). Three zircons are around 1.4 Ga, and plutons of this age are the most abundant Proterozoic rock in the area (Amato et al., 2011). Despite the fact that only two zircons are ca. 1.1 Ga, these are the youngest ages in the sample, and thus we are confident that these dikes are part of the Southwestern Laurentia large igneous province.

In summary, the Southwestern Laurentia large igneous province should be extended eastward from its original limit in eastern Arizona at least into western New Mexico (Zuni Mountains and Antelope Hill), where diabase is exposed on the surface, and probably into the Texas Panhandle region, where coeval gabbro and quartz diabase are encountered in drill-core samples. The total known extent of the province is thus now up to 1500 km from east to west and ∼500 km from northern Mexico to northern New Mexico and 1000 km from north to south if Pikes Peak granitic magmatism in Colorado is included. The expanded scale of this intraplate event, either 750,000 km², or 1,500,000 km² if Pikes Peak is included, provides additional support for earlier interpretations of this widespread event as forming a large igneous province (Ernst and Buchan, 2003; Ernst et al., 2008). The inclusion of significant silicic magmatism in this event, and the extension of this magmatism into Mexico suggest that the original name of southwestern U.S. diabase province (Hammond, 1990) is no longer appropriate, and that is why we are now using the term “Southwestern Laurentia large igneous province.”

Silicic magmatism in the range 1.2–1.1 Ga, coeval with the Southwestern Laurentia large igneous province diabase dikes and sills, is widespread in the southwestern United States and includes A-type granite, alkali-feldspar granite, syenite, and anorthosite (e.g., Smith et al., 1999; Li et al., 2007). Of these intrusions, those in the San Gabriel Mountains (Barth et al., 2001) are >100 m.y. older than the diabase and thus are clearly not related. The granitoids at Red Bluff in west Texas (Li et al., 2007; Howard, 2013) are ∼30–40 m.y. older and also apparently not related, though Barnes et al. (1999) inferred that some undated basalts may be related to the granitoids. The peak magmatism occurred from 1098 to 1085 Ma (Scharer and Allegre, 1982; Smith et al., 1999; Howard, 2013), which is coeval with the Southwestern Laurentia large igneous province. Numerous other granitic intrusions in this region are within 10 m.y. of (and thus essentially coeval with, within 1% uncertainty) the ca. 1085 Ma diabase magmatism (Table 1; Fig. 5). These include, from south to north: (1) plutons in northern Mexico, such as the Aibo granite and others (Espinoza et al., 2003; Iriondo et al., 2003; Farmer et al., 2005; Amato et al., 2009); (2) a rapakivi granite in southern New Mexico, dated at 1077 ± 4 Ma (Amato and Mack, 2012); and (3) plutons in central Texas (e.g., Walker, 1992; Mosher, 1998, and references therein; Li et al., 2007).

A link between the mafic and silicic magmatism has been proposed (e.g., Barnes et al., 1999; Whitmeyer and Karlstrom, 2007). Models suggest that postcollisional, mantle-derived mafic magmatism can provide the heat for crustal anatexis, resulting in alkaline silicic magmatism (e.g., Jahn et al., 1999; Bonin, 2004). Another model applied to Pikes Peak granitoids was based on fractional crystallization of mafic rocks (Smith et al., 1999). Smith et al. (1999) surmised that monzogranites in the Pikes Peak batholith were derived from mixing with mafic or intermediate magmas, and that sodic syenites and granites were derived via fractionation of mantle-derived mafic magmas. For evidence, they cited the high ε_Nd values of +2.2 to −0.7 in the silicic rocks and the presence of relatively small-volume gabbro and diabase dikes. They suggested that significant volumes of mafic rock exist below the current exposure level (Smith et al., 1999).

What is interesting is that with the exception of the intrusion in the Little Hatchet Mountains of southern New Mexico, where mafic magmas are inferred to have mingled with the 1.08 Ga anorthosite (McLemore et al., 2012), Pikes Peak (Smith et al., 1999), and possibly the 1.12 Ga Red Bluff granitic suite (Shannon et al., 1997; Barnes et al., 1999), most areas of 1.1 Ga magmatism have either mafic rocks or silicic rocks, but not both. Virtually all of the reported diabase localities lack silicic rocks, with the exception of small-volume granophyre in the Sierra Ancha region (Smith and Silver, 1975). It is possible that undated mafic rocks in some of the areas with silicic magmatism are also 1.1 Ga, but in general it appears that although broadly the magmatism at 1.1 Ga was “bimodal,” there are few places where bimodal magmatism has been reported from the same locality. This may be related to the level of exposure: Perhaps in most areas of Arizona and southeastern California, higher-level silicic 1.1 Ga plutons or volcanic rocks were exhumed and eroded. Evidence countering this hypothesis is the apparently shallow level of intrusion of some of the diabase sills in Arizona (Wrucke, 1989). Another explanation for the relatively few areas with bimodal magmatism is that melting caused by a mantle plume (see next section) would cause widespread underplating, resulting in partial melting of fusible lower crust. In areas where the lower crust was less fusible, the mafic magma would rise to shallower levels and be emplaced as dikes and sills. These areas would be dominantly mafic with only a minor silicic component. In contrast, in areas where the lower crust was fusible, significant silicic magmatism would be generated, but the mafic magma would freeze out, with the result that silicic magmatism would dominate at shallower levels, and the mafic magmas would only dominate in the lower crust.

We suggest that the similarity in ages between the diabase dikes and silicic intrusions over a broad area implies that the entire magmatic province is at least 1500 km × 1000 km (1.5 million km²) and that the silicic magmatism was generated by both partial melting of crustal rocks induced by mafic magmas and through fractionation of mafic rocks. Such a process is inferred for many large igneous provinces (e.g., Bryan and Ernst, 2008; Bryan and Ferrari, 2013; Ernst, 2014). Mafic underplating associated with the large igneous province can cause partial melting of the lower crust, producing silicic melts of A, I, or S type, depending on the protolith composition.

Ambiguity regarding the tectonic setting and correlations of the Southwestern Laurentia large igneous province magmatism mainly stems from the large uncertainties and resulting wide range of ages for this event. Our data, together with the 1087 Ma baddeleyite age of Heaman and Grotzinger (1992), suggest that rather than a duration stretching up to 100 m.y., from 1140 to 1040 Ma (Table 1), it is more likely that mafic magmatism occurred mainly from 1094 to 1080 Ma. Additional ages may expand or contract this range. This limited range is coeval with abundant silicic plutonism, and together these rocks form a large igneous province that is up to 1500 km × 1000 km in extent. The province could be larger, but the western and southwestern boundaries are defined by a rift margin that formed during the Neoproterozoic breakup of Rodinia (e.g., Bond and Kominz, 1984; Cawood et al., 2007; Li et al., 2008), and the eastern and northern boundaries are unknown because of younger cover rocks, including the Colorado Plateau.

We address three possible models for the origin of the Southwestern Laurentia large igneous province: extension orthogonal to Grenville shortening (Conway et al., 1993; Timmons et al., 2005), post–Grenville-collision delamination (Mosher et al., 2008), and a plume source (Howard, 1991), with a variation that links the Southwestern Laurentia large igneous province plume to the one that created the Keweenawan large igneous province.

Wrucke (1989) suggested that the diabase was related to either passive emplacement or occurred during continental rifting. Hammond (1990) proposed that limited continental rifting caused the diabase magmatic event. Conway et al. (1993) used the northwest-striking dikes in the Hualapai and Pinaleño Mountains to suggest that the intrusions were an extensional response to shortening from the Grenville orogeny. Adams and Miller (1995) suggested that the Pecos mafic intrusive complex, which has the dimensions and apparent orientation to be a major rift structure, resulted from rifting during the Grenville collision, but ambiguities in its age preclude confirmation of this hypothesis (Keller et al., 1989).

Mosher (1998) inferred that silicic magmatism in Texas was related to a collisional indentor (with limits at the west Texas border with New Mexico), with orthogonal rifting inboard of the collision. Timmons et al. (2005) interpreted 1.2–1.1 Ga faulting, sedimentation, and magmatism in the Grand Canyon as being related to protracted Grenville orogenic events. Whitmeyer and Karlstrom (2007) noted that extension in the province probably occurred as the result of Grenville convergence.

A connection between Grenville shortening and diabase magmatism must therefore rely on precise dating of both events. Our new dates have begun to refine the timing of mafic magmatism in the province, but when did the Grenville orogeny end in the southwest United States? The timing of the Grenville orogeny in the southwest United States is mainly known from data in the Llano Uplift in central Texas and in the Van Horn and Franklin Mountains in west Texas. These localities are 300–1000 km away from the main study area in central Arizona. Sedimentation in the Grenville foreland region of west Texas occurred until at least 1250 Ma and possibly as late as 1120 Ma (Bickford et al., 2000). The youngest syntectonic magmatism in the Llano Uplift occurred in the interval between 1120 and 1116 Ma (Walker, 1992; Mosher, 1998; Mosher et al., 2004; Barker and Reed, 2010), whereas magmatism inferred to be post-tectonic relative to the Grenville collision in central Texas occurred from 1098 to 1070 Ma (Walker, 1992; Mosher, 1998). Mosher (1998) and Mosher et al. (2008) inferred that the culminating Grenville collision occurred by 1120 Ma. Shannon et al. (1997) used the geochemistry of the ca. 1120 Ma Red Bluff granitic suite in west Texas to suggest that it formed during extension, as it had no subduction signature. Similarly, poorly dated volcanism in west Texas at 1111 ± 43 Ma (Roths, 1993; mis-cited by Bickford et al., 2000, their Fig. 6, as 1111 ± 20 Ma).

In an alternative model, Grimes and Copeland (2004) suggested that shortening occurred until ca. 1050–980 Ma (∼115 m.y. later than estimates from other geologic data), but this requires that their complex ⁴⁰Ar/³⁹Ar spectra can be interpreted as reliable cooling ages that necessarily resulted from exhumation related to Grenville shortening. We also suggest there is significant uncertainty about the 1035 Ma ⁴⁰Ar/³⁹Ar muscovite and hornblende ages (weighted mean of laser fusion ages that range from <1000 Ma to >1100 Ma) reported by Bickford et al. (2000) in support of movement on a thrust fault indicating a resurgence of shortening proposed by Soegaard and Callahan (1994).

We agree with most previous workers that shortening had largely ceased sometime in the interval of 1120–1116 Ma, possibly as late as 1100 Ma, or just before the oldest post-tectonic dikes and plutons. We also postulate that the majority of mafic magmatism in the Southwestern Laurentia large igneous province therefore occurred after the Grenville orogeny had ceased in the southwest United States, and that silicic magmatism older than ca. 1120 Ma was related to Grenville subduction/collision and that younger silicic magmatism was related to the influx of mantle-derived magmatism. Regardless, we suggest that the paucity of mafic magmatism during the bulk of the protracted Grenville event is not consistent with a Rhine graben model of extension orthogonal to a collisional front (Adams and Miller, 1995; Mosher, 1998). In addition, mafic magmatism (Fig. 1) occurred over 1000 km from the western edge of the continental indentor proposed by Mosher (1998). It is worth noting that the Keweenawan event in the Midcontinent rift was previously suggested to have been caused by an “impactogen” (Anderson and Burke, 1983), but subsequent work has shown that the magmatism likely formed during a lull in Grenville shortening, with renewed shortening and inversion of the rift postdating the main phase of magmatism (Cannon, 1994; Hauser, 1996).

Further tests of this model would require further dating of diabase across the province and improved constraints on the end of the Grenville orogeny in southwestern Laurentia.

Mosher et al. (2008) proposed that following the Grenville collision at 1120 Ma, lithospheric delamination occurred, resulting in uplift and upwelling of mafic magma that led to crustal melting. Lithospheric delamination following collision can cause the asthenosphere to rise and generate bimodal mafic and alkaline felsic magmatism (Bonin, 2004). This process was inferred to have occurred in Algeria, where gabbro and anorthosite formed after the Pan-African orogeny when lithospheric delamination was inferred to have occurred and was linked to postcollisional transtension (Liegeois and Black, 1987; Ait-Djafer et al., 2003). Slab roll-back and breakoff were considered to explain magmatism in the Eastern Carpathians following a Miocene collision (Girbacea and Frisch, 1998).

With application to the case of the Southwestern Laurentia large igneous province, the inferred area of delamination should underlie the entire area of mafic and silicic magmatism. A key question is whether the seismic data for the lithospheric structure of this southwest part of Laurentia are consistent with a widespread thinning of the lithosphere in this region at this time. However, the lithospheric structure at 1.1 Ga may not be easily determined given the subsequent history of Jurassic and Tertiary extension in the region along with Laramide-age shortening. The subsurface maps of Artemieva (2009) do not show any obvious thinning of the modern lithosphere in southwest Laurentia, but that region has experienced multiple episodes of Phanerozoic shortening and extension, and thus the modern lithosphere probably does not reflect the thickness of the lithosphere in the Mesoproterozoic.

We consider a model that is related to the previous model, but in our model the delamination is caused by margin-parallel shearing resulting from oblique convergence. The NW-SE orientation of the normal faults, dike intrusions, and basins shown by Timmons et al. (2005) represents northeast-southwest extension that could be caused by orthogonal extension caused by northwest shortening, in a pure shear model. However, an alternative explanation is that this extension was caused by dextral shearing along the margin, in a simple-shear model (Fig. 6). This deformation would generate northwest-trending structures such as dikes and normal faults, based on the orientation of the least-compressive stress. In this model, the dikes and faults are not perpendicular to the Grenville front, but instead they are oblique to the margin. Apache Group sediment was inferred to have been deposited in extensional basins (see review in Wrucke, 1989), and we suggest these may have been transtensional basins associated with a major shear zone. Soegaard and Callahan (1994) described sedimentological and structural evidence in support of ca. 1100 Ma late Grenville basin formation within a transpressive tectonic setting. Bickford et al. (2000) noted a convergence direction oblique to the margin in the southwest United States that would result in right-lateral shearing. Grimes and Mosher (2003) noted a transition from orthogonal to dextral oblique shortening in west Texas Grenvillian structures, and this observation is consistent with our model.

The apparent contradiction of horizontal sill and sheet intrusions together with vertical dike formation (Howard, 1991) could be resolved in this model. The dikes intrude as the result of horizontal extension associated with the shear zone. The sills and sheets can form during the uplift as the asthenosphere is upwelling and creating lithospheric buoyancy throughout the province. However, regional dike swarms are generally considered to be emplaced in a direction consistent with the maximum compressive stress (e.g., Fahrig, 1987; Halls and Fahrig, 1987; Ernst, 2014). There have been numerous proposals for emplacement of regional-scale dike swarms in transtensional environments, but in each case, further work, including precise dating to sort out the membership of dikes into different swarms, consistently affirms that the trend of the dike swarms is consistent with emplacement direction controlled by the maximum compressive stress direction, with the direction of σ3 being normal to the trend of the dike (Ernst, 2014).

Further work to test this model would be to assess whether dike orientations and normal faults have an en echelon pattern, which is more likely in a simple shear model.

The arrival of a mantle plume, and subsequent partial melting, is a well-established method for generating a large igneous province (Coffin and Eldholm, 1994; Campbell, 2001; Courtillot and Renne, 2003; Ernst and Buchan, 2003). Howard (1991) alluded to a plume origin for the Southwestern Laurentia large igneous province, based on the abundance of horizontal sheets, and Barnes et al. (1999) noted that the Pecos mafic intrusive complex in Texas (Fig. 1) could have both formed as the result of a mantle plume. As noted above, the ages of the mafic/ultramafic rocks in the basement of Texas are uncertain.

The Midcontinent rift system, also referred to as the Keweenawan Rift, of central North America is a Proterozoic (ca. 1.1 Ga) flood basalt large igneous province that formed along a 2000-km-long continental rift (Davis and Paces, 1990; Hutchinson et al., 1990; Nicholson and Shirey, 1990; Klewin and Shirey, 1992; Seifert and Olmsted, 2004; Heaman et al., 2007; Hollings et al., 2010). Some nonplume models have been proposed for the Keweenawan large igneous province, such as the “impactogen” model of Anderson and Burke (1983; but see also Cannon [1994] for a rebuttal) or the oblique collision model of Donaldson and Irving (1972), or the pull-apart basin model of Gordon and Hempton (1986). However, most models for the Keweenawan large igneous province infer its formation above a mantle plume (Hutchinson et al., 1990; Nicholson and Shirey, 1990), and volume trends of magmatism are consistent with a hotspot source that may have created rifts at the boundary of a microplate (Merino et al., 2013).

Tholeiitic basalts are dominant in the Keweenawan large igneous province, although mafic sills and silicic volcanism are also significant components (Davis and Green, 1997; Seifert and Olmsted, 2004). Main-stage rifting occurred from 1108 to 1086 Ma (Davis and Green, 1997), although some magmatism associated with the rift may have begun as early as 1150 Ma (Heaman et al., 2007) and is thought to be related to the 1140 Ma Abitibi dike swarm, which extends northeast to east-northeast from the rift for a distance of up to 700 km (Ernst and Buchan, 1993). The overlap of igneous ages between the Keweenawan large igneous province and our new ages for the Southwestern Laurentia large igneous province is consistent with the paleomagnetic data of Harlan (1993). However, ∼50%–80% of the igneous activity in the Midcontinent rift occurred at 1108 ± 1 Ma (Davis and Green, 1997), and this is within 15 m.y. of the earliest magmatism in the Southwestern Laurentia large igneous province.

Both the Keweenawan large igneous province and the Southwestern Laurentia large igneous province could have formed as the result of mantle plumes. It is possible that a second independent plume in the southwest Unite States occurred at the exact time as (or slightly later than) the Keweenawan plume. However, the near synchronicity in ages suggests to us that a simpler model is one in which a single plume was responsible for magmatism in both areas (Fig. 7).

We propose a single underlying plume ascending beneath the south-central part of the Laurentian craton and spreading along the base of the lithosphere out to a typical diameter of ∼2000 km (e.g., Campbell, 2007). A plume head that has arrived beneath the thick lithospheric root of south-central Laurentian craton spreads in all directions beneath the lithosphere, and the pattern of spreading is influenced by the topography of the lithospheric-asthenospheric boundary. Plume material can slide upward along the sublithospheric boundary toward “thin spots” (Thompson and Gibson, 1991). As illustrated in Artemieva (2009, her Fig. 5A), the base of the lithosphere in the region between the Keweenawan and Southwestern Laurentia large igneous province regions is between 150 and 100 km. There is an overall gradient (shallowing) to the south, and so a plume head would dominantly spread southward up to shallower depths as it spread in that direction. Not resolvable on this diagram are the locations of regions of ∼60–70 km, which represent depths of significant generation of flood basalt magmatism (e.g., White and McKenzie, 1989). However, such regions must exist for voluminous melt generation by any model. What is encouraging for a plume origin for the Southwestern Laurentia large igneous province is that the topography of the base of the lithosphere underlying the central United States is overall sloping upward to the south.

With such a model, the other scenario to consider is the portion of the plume head spreading northward and its potential role in generating the Keweenawan large igneous province of the Midcontinent rift region. It is notable that depth of the base of the lithosphere significantly increases on the north side of the Midcontinent rift system. This change is due to location of the southern boundary of the Superior craton, which, like all Archean cratons, has a thicker lithospheric root than does Proterozoic or Phanerozoic crust. The increase in thickness is ∼50 km (Artemieva, 2009). In our model, the portion of the spreading plume head that travels northward will be mostly stopped at the Great Lake regions, because of the thicker lithosphere to the north.

Extensive magmatism in the Midcontinent rift requires a mechanism for lithospheric thinning to allow plume head material to reach levels at ∼60 km depth to allow voluminous melting. Merino et al. (2013) suggested that Keweenawan magmatism was concentrated along a rift system opened in a microplate type model with the Wisconsin block rotating away from the Superior Province. So we are suggesting that such extension in the Merino et al. (2013) model will result in sufficient lithospheric thinning to allow the underlying plume material to ascend to a shallow enough depth for extensive melting to produce the Keweenawan large igneous province.

The plume model does not preclude other models such as extension or delamination. For example, we postulate that the two areas with extensive ca. 1.1 Ga magmatism (Keweenawan large igneous province and Southwestern Laurentia large igneous province) were previously thinned by extension. Magmatism from the plume source would be localized in the crustal regions that had experienced extension above the plume. The delamination and subsequent exhumation proposed by Mosher (1998) for the time period 1120–1070 Ma overlap completely with the Southwestern Laurentia large igneous province magmatic episode. Similarly, the hybrid model of Şengör (2001) suggests that a combination of delamination and a plume source can produce large igneous province–scale magmatism.

There are other large igneous province events in which coeval mafic magmatism is found on more than one side of a crustal block, supporting the idea of a mantle plume rising underneath and spreading out in multiple directions upward along the lithospheric-asthenospheric boundary (the thin spots of Thompson and Gibson, 1991), perhaps channelized along lithospheric valleys (e.g., Ebinger and Sleep, 1998). One example is the 1880 Ma circum-Superior large igneous province of the Superior craton (Baragar et al., 1996; Heaman et al., 2009; Ernst and Bell, 2010; Minifie et al., 2013). A Cenozoic example is the Yellowstone plume, which has been postulated to have been bifurcated by the descending slab (Kincaid et al., 2013).

We suggest that the existing timing constraints are consistent with the Southwestern Laurentia large igneous province magmatism occurring after the last main phase of Grenville shortening in the southwest United States. Thus, we suggest that postcollisional models for this magmatism are more applicable than those that infer a connection between Grenville shortening and extension inboard of an indentor (e.g., Mosher, 1998). We prefer a model in which magmatism resulted from lithospheric delamination, possibly caused by right-lateral shearing during oblique convergence at the end of the Grenville orogeny. This may have then allowed a single plume that impinged upon thick lithosphere in the midcontinent to spread out along the base of the lithosphere, with a portion moving southwest to the thinned southwest edge of Laurentia, and partially melting to form the Southwestern Laurentia large igneous province, and another portion of the flattened plume head spreading northward to an area of rifting and lithospheric thinning (e.g., Merino et al., 2013), melting to form the Keweenawan large igneous province in the Midcontinent rift, which has similar but slightly older ages. Mafic magmas induced crustal melting resulting in the post-Grenville silicic magmatism in the Southwestern Laurentia large igneous province.

If this large igneous province age can be found on other crustal blocks, then the barcode matching approach (Bleeker and Ernst, 2006) could help to identify nearest neighbors to southwestern Laurentia at this time. Globally, there are several large igneous provinces of approximately the same age (Fig. 8). In western Australia, the Warakurna large igneous province contains mafic sills, dikes, and mafic-ultramafic intrusions that were emplaced at 1078–1070 Ma over a large area, suggesting a mantle plume source (Wingate et al., 2004; Ernst et al., 2008). These ages are close to, but younger than, Southwestern Laurentia large igneous province magmatism. The Rodinia reconstruction of Li et al. (2008) used geologic and paleomagnetic data to suggest that Australia had not accreted to Laurentia by 1100 Ma and that there are geologic problems with the Australia–Southwestern U.S. (AUSWUS) model (Brookfield, 1993; Karlstrom et al., 1999; Burrett and Berry, 2000).

The Umkondo large igneous province in southern Africa formed at 1112–1102 Ma with a mean at 1108 Ma (Hanson et al., 1998; Hanson et al., 2004). This age range puts the magmatic event in the Kalahari craton ∼20–25 m.y. older than magmatism in the Southwestern Laurentia large igneous province. Hanson et al. (2004) suggested that coeval magmatism between the Umkondo large igneous province in the Kalahari craton and magmatism in the Keweenawan (Midcontinent rift) of Laurentia likely represents a single large igneous province that formed during assembly of Rodinia and that magmatism in the Southwestern Laurentia large igneous province was coeval with both of those regions and possibly also related to the same large igneous province. Paleomagnetic data show that Kalahari and Laurentia were separated by a gap of 1700–3000 km at 1100 Ma (Li et al., 2008), and thus the plume would have affected a broad region prior to Rodinia assembly, and it would have been independent of the lithospheric plate movements (Hanson et al., 2004). The gap between the two cratons could have been oceanic lithosphere, or, alternatively, another craton, such as that hosting the Warakurna large igneous province, could have been present between them (Hanson et al., 2004). We suggest that the ages are sufficiently different (Kalahari: 1078–1070 Ma; Southwestern Laurentia large igneous province: 1094–1080 Ma) and the plate configuration issues are sufficiently insurmountable that these swarms should not be considered piercing points for Rodinia reconstructions.

Magmatism occurred at ca. 1110 Ma in other areas, globally. Ernst et al. (2013b) obtained a new age for a dike swarm in the Angola part of the Congo craton and suggested a reconstruction with other areas with 1110 Ma magmatism: the Mahoba suite of ENE-WSW–trending dikes in the Bundelkhand craton of northern India, which is also ca. 1110 Ma (Pradhan et al., 2012), and mafic intrusions in the Bolivian part of the Amazonian craton (Hamilton et al., 2012). However, in all of these areas (Congo, Indian, Amazonian, and Kalahari cratons), the 1110 Ma magmatism is too old to correlate with the Southwestern Laurentia large igneous province, and the identity of the crustal block formerly adjacent to the southwest part of Laurentia remains undetermined.

We used high-precision TIMS to date baddeleyite in diabase from four localities in Arizona to obtain ²⁰⁶Pb/²³⁸U dates of 1080 ± 2 Ma, 1080 ± 3 Ma, 1088 ± 3 Ma, and 1094 ± 2 Ma yielding a range of 1094–1080 Ma. These are the first high-precision ages on rocks from this magmatic province to be published since two baddeleyite ages of 1087 Ma and 1069 Ma were reported (Heaman and Grotzinger, 1992), and we suggest that the 1069 Ma date represents Pb loss. Previously published U-Pb ages are consistent with a model in which coeval silicic rocks were derived as crustal melts by heating of lower crust from mantle-derived mafic magmas. The former “southwestern U.S. diabase province” of Hammond (1990) is herein renamed the Southwestern Laurentia large igneous province because the province includes rocks in Sonora, Mexico, and includes rocks of silicic composition in addition to the diabase.

We prefer alternate models to the one in which the large igneous province was caused by extension inboard and orthogonal to Grenville shortening (e.g., Timmons et al., 2005) because we suggest that the bulk of Southwestern Laurentia large igneous province magmatism postdated the end of the Grenville orogeny (Mosher, 1998) in southwestern Laurentia. We suggest that magmatism was induced by lithospheric delamination following the Grenville collision, possibly related to dextral shearing resulting from oblique convergence. A mantle plume may have been present, and given the partial overlap in age between the Southwestern Laurentia large igneous province and the plume-related Keweenawan large igneous province of the midcontinent region, we propose a model in which a single mantle plume impinged beneath the lithosphere of south-central Laurentia. Given appropriate topography on the lower surface of the lithosphere, the plume material may have slid upward along the sublithospheric boundary in more than one direction toward regions of thinner lithosphere that would permit large igneous province–scale partial melting. We suggest the part of the plume sliding upward toward the north could have partially melted and produced the Keweenawan large igneous province in association with plate-boundary extensive forces that caused rifting and decompression melting of the underlying plume head. Another part of this plume could have migrated southward to regions of thinner lithosphere along the southern margin of Laurentia, producing the Southwestern Laurentia large igneous province. Although the Southwestern Laurentia large igneous province is similar in age to ca. 1110 Ma large igneous provinces of the Kalahari, Congo, or Amazonian cratons, and also similar in age to the 1075 Ma Warakurna large igneous province, the age match is not sufficiently robust to argue for them being linked in paleocontinental reconstructions.

Future tests of these models will come from geochemistry, to constrain the magmatic sources and try to identify a plume signature, structural geology, to evaluate pure shear versus simple shear models for extension, and geochronology, to address any age trends and the overall duration of the event.

This is publication 34 of the Large Igneous Provinces–Paleocontinental Reconstruction–Resource Exploration Project (www.supercontinent.org), which provided funding for geochronology. Wouter Bleeker, a coleader of this project, participated in helpful discussions on the geochronology of this paper. Bright was supported by funding from the New Mexico Geological Society and the Department of Geological Sciences at New Mexico State University. Amato acknowledges sabbatical support from the Cooperative Institute for Research in Environmental Sciences at the University of Colorado at Boulder. Assistance to George Gehrels was provided by the staff at the Laserchron Laboratory at University of Arizona and at the USGS/Stanford sensitive high-resolution ion microprobe facility. Ulf Söderlund at Lund University extracted baddeleyite from the samples. Baddeleyite isotope dilution–thermal ionization mass spectrometry analyses were conducted at the laboratory of Roland Mundil at the Berkeley Geochronology Center. We thank Lang Farmer and Emily Verplanck at the University of Colorado at Boulder for their assistance in collecting the Sr and Nd isotopic data. Jon Spencer collected the diabase sample at Oracle. Yvette Lopez and Nancy McMillan collected and provided X-ray fluorescence (XRF) data for two samples from the Burro Mountains. Sam Bothern helped obtain some of the XRF data at New Mexico State University.