The Gulf of California is an archetype of continental rupture through transtensional rifting, and exploitation of a thermally weakened arc to produce a rift. Volcanic rocks of central Baja California record the transition from calcalkaline arc magmatism, due to subduction of the Farallon plate (ca. 24–12 Ma), to rift magmatism, related to the opening of the Gulf of California (<12 Ma). In addition, a suite of postsubduction rocks (<12 Ma), referred to as “bajaites,” are enriched in light rare-earth and other incompatible elements (e.g., Ba and Sr). These are further subdivided into high-magnesian andesite (with 50%–58% SiO2 and MgO >4%) and adakite (>56% SiO2 and MgO <3%). The bajaites correlate spatially with a fossil slab imaged under central Baja and are inferred to record postsubduction melting of the slab and subduction-modified mantle by asthenospheric upwelling associated with rifting or slab breakoff. We report on volcanic rocks of all three suites, which surround and underlie the Santa Rosalía sedimentary rift basin. This area represents the western margin of the Guaymas basin, the most magmatically robust segment of the Gulf of California rift, where seafloor spreading occurred in isolation for 3–4 m.y. (starting at 6 Ma) before transtensional pull-apart basins to the north and south ruptured the continental crust. Outcrops of the Santa Rosalía area thus offer the opportunity to understand the magmatic evolution of the Guaymas rift, which has been the focus of numerous oceanographic expeditions.

We describe 21 distinct volcanic and hypabyssal map units in the Santa Rosalía area, using field characteristics, petrographic data, and major- and trace-element geochemical data, as well as zircon isotopic data and ten new 40Ar-39Ar ages. Lithofacies include lavas and lava domes, block-and-ash-flow tuffs, ignimbrites, and hypabyssal intrusions (plugs, dikes, and peperites). Calcalkaline volcanic rocks (13.81–10.11 Ma) pass conformably upsection, with no time gap, into volcanic rocks with rift transitional chemistry (9.69–8.84 Ma). The onset of rifting was marked by explosive eruption of silicic ignimbrite (tuff of El Morro), possibly from a caldera, similar to the onset of rifting or accelerated rifting in other parts of the Gulf of California. Epsilon Hf zircon data are consistent with a rift transitional setting for the tuff of El Morro. Arc and rift volcanic rocks were then juxtaposed by normal faults and tilted eastward toward a north-south fault that lay offshore, likely related to the north-south normal faults documented for the early history of the Guaymas basin, prior to the onset of northwest-southeast transtenional faulting. Magmatism in the Santa Rosalía area resumed with emplacement of high-magnesian andesite lavas and intrusions, at 6.06 Ma ± 0.27 Ma, coeval with the onset of seafloor spreading in the Guaymas basin at ca. 6 Ma.

The 9.69–8.84 Ma rift transitional volcanic rocks underlying the Santa Rosalía sedimentary basin provide a maximum age on its basal fill. Evaporites in the Santa Rosalía sedimentary basin formed on the margin of the Guaymas basin, where thicker evaporites formed. Overlying coarse-grained clastic sedimentary fill of the Santa Rosalía basin and its stratiform Cu-Co-Zn-Mn sulfides may have accumulated rapidly, coeval with emplacement of 6.06 Ma high-magnesian andesite intrusions and the ca. 6 Ma onset of seafloor spreading in the Guaymas basin.

Previous work established that volcanic rocks of the Santa Rosalía area (Fig. 1) record an arc-rift transition, from subduction of the Farallon plate, to rift magmatism related to the opening of the Gulf of California (Conly et al., 2005). Subduction and rifting in the Gulf of California region are estimated to be ca. 24–12 Ma and ca. 12 Ma to present, respectively, using plate tectonic models (e.g., Atwater, 1970; Gastil et al., 1979; Atwater, 1989; Stock and Hodges, 1989, 1990; Lonsdale, 1991; Atwater and Stock, 1998; Henry and Aranda-Gomez, 2000; Michuad et al., 2006; Umhoefer, 2011). Furthermore, the geochemistry of volcanic rocks of the Gulf of California region is generally consistent with a change from subduction (referred to as Comondú Group, Fig. 1) to rifting at ca. 12 Ma (e.g., Hausback, 1984; Sawlan and Smith, 1984; Sawlan, 1991; Martín-Barajas et al., 1995, 2000; Paz Moreno and Demant, 1999; Ferrari et al., 2005), although residual arc signatures persisted for 4 m.y. in Sonora (Till et al., 2009) and recurred in the Quaternary at Tres Virgenes volcanic field (Fig. 2; Sawlan, 1991; Schmitt et al., 2006, 2010). The Gulf of California ruptured much more rapidly than is typical of continental rifts, at least partly because it exploited the thermally weakened Comondú arc axis (Umhoefer, 2011), but also because pull-apart basins are the most efficient tectonic setting for localizing rapid thinning of the crust (Christie-Blick and Biddle, 1985; Nilsen and McLaughlin, 1985; Pitman and Andrews, 1985; Nilsen and Sylvester, 1995; Xie and Heller, 2009; Umhoefer, 2011; Allen and Allen, 2013). In this paper, we explore in detail the geology of the arc-rift transition, preserved in the Santa Rosalía area (Figs. 24).

The Santa Rosalía area also offers insight into the processes that produced the Guaymas basin (Fig. 1). This basin has been the focus of two Deep Sea Drilling Project and Integrated Ocean Drilling Program (DSDP/IODP) expeditions, as well as wide-angle and multi-channel seismic surveys (Curray et al., 1982; Lizarralde et al., 2007; Miller and Lizarralde, 2013; Teske et al., 2018). Tectonic reconstructions show that the 2-km-thick evaporite body imaged by multi-channel seismic surveys in the Guaymas basin was fringed by thinner (up to ∼70-m-thick) gypsum deposits exposed onshore in the Santa Rosalía basin (Fig. 1; Miller and Lizarralde, 2013). However, the evaporite and underlying crustal rocks in the Guayamas basin have not been sampled or dated. The Santa Rosalía area offers the opportunity to do this.

Seafloor spreading in the Gulf of California initiated in the Guaymas basin at ca. 6 Ma, and occurred in isolation for 3–4 m.y. before transtensional pull-apart basins to the north and south ruptured the continental crust (Lizarralde et al., 2007; Sutherland et al., 2012; Martín-Barajas et al., 2013; Umhoefer et al., 2018). Guaymas basin is also a more magma-rich rift than the other segments (Lizarralde et al., 2007). Why did this segment succeed first, in the center of a seaway that extended ∼700 km to the north and ∼750 km to the south? Lizarralde et al. (2007) attempted to address this question but arrived at no clear resolution. We show here that high-magnesian andesite magmatism in the Santa Rosalía area was coeval with the onset of seafloor spreading in the Guaymas basin at 6 Ma. The distribution of high-magnesian andesites (shown on Fig. 1) correlates spatially with remnant slabs geophysically imaged under central Baja California. These andesites have generally been regarded as the product of postsubduction slab-window magmatism (Benoit et al., 2002; Calmus et al., 2003; Bellon et al., 2006; Pallares et al., 2007, 2008; Castillo, 2008; Negrete-Aranda and Cañón-Tapia, 2008; Negrete-Aranda et al., 2010, 2013; Calmus et al., 2011; Di Luccio et al., 2014). The high-magnesian andesites in the Santa Rosalía area lie far inboard from the paleotrench, relative to other high-magnesian andesites in central Baja California (Fig. 1). We discuss the possibility that this was controlled by a slab tear that crosses beneath the Baja California Peninsula between the Guadalupe and Magdalena plates. This tear was imaged by Di Luccio et al. (2014) as a low-velocity (hot) zone, and we discuss how this tear may have promoted early rupture and magma-rich seafloor spreading in the Guaymas basin.

Much attention has also been focused on the origin and age of Cu-Zn-Co-Mn stratiform deposits of the Boleo Formation sedimentary rift basin fill at Santa Rosalía (Wilson and Rocha, 1955; Schmidt, 1975; Holt et al., 2000; Ochoa-Landín et al., 2000; Conly et al., 2006, 2011). These deposits have been mined for more than a century and are currently the target of a multibillion-dollar mining operation, the Minera Boleo. Although these ore beds lie within clastic sedimentary rocks that rest upon and interfinger with the evaporates described above, their age is poorly constrained. We present arguments for the interpretation that the ore beds formed rapidly, during emplacement of slab window high-magnesian andesite intrusions and onset of Guaymas basin seafloor spreading at 6 Ma.

Regional-scale tectonic reconstructions have delineated four main phases in the evolution of the Gulf of California (Gastil et al., 1979; Angelier et al., 1981; Atwater, 1989; Stock and Hodges, 1989, 1990; Lonsdale, 1991; Axen, 1995; Axen and Fletcher, 1998; Henry and Aranda-Gomez, 2000; Martín-Barajas et al., 2000; Stock, 2000; Oskin et al., 2001; Oskin and Stock, 2003a, 2003b, 2003c; Michaud et al., 2006; Lizarralde et al., 2007; Umhoefer, 2011; Sutherland et al., 2012; Duque-Trujillo et al., 2015; Darin et al., 2016; Drake et al., 2017):

  • (1) Forearc volcaniclastic sedimentation in what is now the Gulf of California (ca. 28–19 Ma), between a subduction zone to the west and an extensional continental arc to the east that produced large-volume explosive caldera-forming silicic eruptions due to slab rollback (Sierra Madre Occidental). Distal and reworked equivalents of these ignimbrites were deposited in the forearc region, in what is now the Gulf of California region.

  • (2) Continued slab rollback caused westward sweep of the volcanic arc into what is now the Gulf of California region, referred to as the Comondú Group or Comondú arc (ca. 19–12 Ma), at least locally accompanied by E-W extension, both over a narrow (∼100 km) region. This sweep was generally accompanied by a shift to effusive volcanism with intermediate compositions. These rocks now form the main peninsular escarpment in Baja California (Sawlan and Smith, 1984).

  • (3) End of subduction and onset of continental rift and/or slab window volcanism in what is now the Gulf of California region. The relative importance of E-W extension and NW-SE transtension remains controversial for this time frame (ca. 12–6 Ma). Volcanic rocks of this age include basalts and the products of large-volume silicic explosive eruptions (ignimbrites). As subduction volcanism waned, the “bajaites” described above were erupted across central Baja California.

  • (4) Continental separation and onset of seafloor spreading, under a NW-SE transtensional regime with ongoing continental rift and slab window volcanism (ca. 8–6 Ma to present).

The Gulf of California is an outstanding place to study the geologic processes described above, because exposures are superb, and the young volcanic rocks that record these processes are relatively unaltered. For this reason, volcanic rocks have been studied along many parts of the Gulf of California (e.g., Hausback, 1984; Sawlan and Smith, 1984; McLean et al., 1987; Stock and Hodges, 1989; Sawlan, 1991; Martín-Barajas et al., 1995, 1997; Paz Moreno and Demant, 1999; Delgado Argote et al., 2000; Stock, 2000; Umhoefer et al., 2001; Escalona Alcazar et al., 2001; Ferrari et al., 2005; Schmitt et al., 2010; Bennett et al., 2013; Ferrari et al., 2013; Bennett et al., 2015; García-Sánchez et al., 2016; Osorio-Ocampo et al., 2016; Ferrari et al., 2018). However, only one paper has focused on magmatism in the Santa Rosalía area (Conly et al., 2005). This represents a gap in knowledge of the timing and nature of magmatism on the margin of the well-studied, magmatically robust Guaymas rift.

Analytical methods are detailed in Supplemental Item 11, hereafter referred to as Item S1. These include X-ray fluorescence (XRF) and inductively coupled plasma–mass spectrometry (ICP-MS) whole-rock geochemical analysis methods, 40Ar-39Ar geochronologic methods, and methods for obtaining U-Pb and Hf isotopic analyses on zircons. Analytical data are given in Item S3 (footnote 1) for 40Ar-39Ar, Item S4 for U-Pb zircons, and Item S5 for Hf isotopes in zircon.

For the first time in the Santa Rosalía area (Figs. 1 and 2), we present detailed volcanic lithofacies maps (Figs. 3 and 4), volcanic stratigraphy (Fig. 5 and Table 1), outcrop photos (Fig. 6 and Item S6), and photomicrographs (Fig. 7 and Item S7), as well as geochemical data on samples located by GPS (Item S2), with samples plotted on the maps (Figs. 3 and 4).

Our mapping has focused on two main areas (Fig. 2), Santa Rosalía north (Fig. 3 and Table 1A) and Santa Rosalía south (Fig. 4 and Table 1B), separated by sedimentary rocks of the overlying Santa Rosalía basin. The Santa Rosalía north geologic map (Fig. 2) was mapped on Google Earth base map with Digital Globe images from 2012. The Santa Rosalía south geologic map (Fig. 3) was mapped on Google Earth base map with Digital Globe imagery from 2016. Maps from Google Earth use the WGS 1984 coordinate system. Additionally, 30 m contour lines derived from ASTER GDEM V2.0 courtesy of Ministry of Economy, Trade, and Industry (METI) and National Aeronautics and Space Administration (NASA). Field mapping and data were entered at a variety of spatial scales using a tablet with GPS and the FieldMove program. Strikes and dips were taken with both Brunton compass and FieldMove Clino at all locations. Google Earth KMZ files are provided in Item S9 (footnote 1), with map layers and map units, sample locations, strike and dip measurements, and faults.

Volcanic lithofacies (e.g., lavas, block-and-ash-flow tuffs, ignimbrites, etc.) are described in the first section of this paper. Volcanic-volcaniclastic terminology follows that of Fisher and Schmincke (1984). Lithofacies repeat, but they are separated into distinct map units (Table 1) using contact relations and variations in chemistry, mineralogy, textures, and 40Ar-39Ar geochronologic ages. These map units are described in the second section and grouped by chemistry into three basic categories (calcalkaline, rift transitional, and high-magnesian andesites), but discussion of the geochemistry is deferred until after the map units are described.

We mapped the following volcanic lithofacies in the Santa Rosalía area: lava domes, block-and-ash-flow tuffs, lavas, peperites, hypabyssal intrusions, and ignimbrites. Lava domes are recognized as coherent, compositionally homogeneous bodies with high aspect ratio (i.e., thick and highly lenticular), mantled by coarse breccias of the same composition (Fig. 6A). These domes can commonly be mapped down-section into intrusions that crosscut stratigraphy. Block-and-ash-flow tuffs consist of unsorted, nonstratified deposits of block- to lapilli- to ash-sized clasts, all of the same composition (Fig. 6B). These deposits are widely regarded to result from non-explosive collapse of lava domes or lava flow fronts (e.g., Fisher et al., 1980; Sparks, 1997; Freundt et al., 2000). Lavas are distinguished from intrusions by the presence of flow-top breccias or vesiculated tops that form layers. Lavas may be further subdivided by textural characteristics; in this area, we subdivide them by marked differences in thickness. Thick lavas form stratified sections with individual flows ∼10 m thick, with coherent intervals separated by flow breccias with large blocks (Fig. 6C). Thin lavas form stratified sections with individual flows less than 2–3 m thick, with coherent intervals separated by scoria lapilli and small blocks (Fig. 6D). Peperite intrusions consist of glassy (quenched) blocks with jigsaw clast texture dispersed in a sedimentary host (Fig. 6E). Hypabyssal intrusions form coherent, homogeneous bodies that crosscut stratigraphy; the example shown in Figure 6F has miarolitic cavities. Ignimbrites consist of pumice (light-colored) or scoria (dark-colored) blocks in a matrix of bubble-wall shards of the same composition. They form thick massive, unsorted deposits and commonly contain volcanic rock fragments and broken crystals. Ignimbrites are commonly laterally extensive, so they are used as regional-scale stratigraphic markers (e.g., Busby et al., 2005; Murray et al., 2013; Busby et al., 2018a). For this reason, we describe this lithofacies in detail below, in stratigraphic and geochronological context.

The Santa Rosalía north area is subdivided into outcrop areas (e.g., Cerro Sombrero Montado, Cerro Juanita, El Morro, etc., Fig. 3, Table 1A) separated by sedimentary rocks of the Santa Rosalía basin (Fig. 2). The Santa Rosalía south area forms one contiguous outcrop area (Fig. 4). Within outcrop areas, map units are shown in stratigraphic order on the map keys (Figs. 3 and 4); however, none of the map units are continuous from one outcrop area to the next. Therefore, the composite stratigraphy shown on Figure 3 and Table 1A was constructed using the 40Ar-39Ar ages shown on the map key and Figure 5.

Table 1 provides an outline for the description of map units, grouping the map units by outcrop area and summarizing outcrop and petrographic characteristics of 47 samples. The units described in Table 1A (Santa Rosalía north, Fig. 3) are described first because that area ranges to older ages than rocks of Santa Rosalía south (Table 1B), which ranges to younger ages (Fig. 4). Compositional names (e.g., andesite and basaltic trachyandesite) were assigned using major- and trace-element data (Item S2 [footnote 1]), and we group these by geochemical characteristics into the following geochemical designations: (1) calcalkaline, (2) rift transitional rocks, further divided into ignimbrites and lavas, and (3) high-magnesian andesites. The distribution of rocks with these geochemical designations is shown in the insets of the geologic maps (Figs. 3 and 4).

40Ar-39 Ar ages are summarized in stratigraphic order, grouped and color-coded by geochemical designation, in Figure 5. Age spectra and 36Ar/40Ar versus 39Ar/40Ar correlation diagrams are shown in Figures 8A and 8B, respectively. Item S3 contains a summary table of the 40Ar-39Ar results, and for each sample, it presents a table with relevant 40Ar-39Ar, age spectra, 37ArCa/39ArK diagram, and 36Ar/40Ar versus 39Ar/40Ar correlation diagram.

Selected outcrop photos (Fig. 6) are presented herein. A more complete set is presented in Item S6, in order to help the reader visualize field characteristics. A complete set of photomicrographs is presented in Item S7 to provide data on volcanic textures, minerals, and alteration. Photomicrographs are provided for all dated samples (Item S7). Unless otherwise noted, all samples have plagioclase; the other mineral names are given in order of increasing abundance on Table 1 (least abundant mineral first, as is standard). In this section, the map unit names shown on Figures 3 and 4 and Table 1 are written in italics, for ease of reference.

Santa Rosalía North

Rocks of Uncertain Stratigraphic Position

Three geographic areas are arbitrarily placed at the base of Table 1A because their age is unknown.

The Mina Caopas gypsum lies in the north end of the Santa Rosalía sedimentary basin (Fig. 2) and is ∼100 m thick, making it one of the biggest gypsum producers in the world. At one locality in the bottom of an arroyo, the gypsum rests on a lobate body of basaltic andesite lava (Table 1). This body has abundant plagioclase, so we name it a hornblende clinopyroxene plagioclase porphyry to emphasize that fact (sample SR 10-1, Table 1). It has a coherent interior mantled by flow breccia and passes laterally into a block-and-ash-flow tuff with the same mineralogy and blocks up to 1.5 m in size (Items S6A-1 and S6A-2 [footnote 1]). This lava is calcalkaline (Table 1). Therefore, it is not likely that it would yield a useful maximum age on the base of the gypsum, since it is likely to be >12 Ma in age, although calcalkaline volcanism has recurred since subduction ended, as discussed above.

The northern core localities (cores LS-08 and LS-10, Fig. 2) were drilled beneath the Santa Rosalía basin sedimentary fill. They consist of coherent dacite (calcalkaline); LS-10 and LS-08, Table 1A), and may either represent lava with the flow-top breccia eroded away or hypabyssal intrusions. These samples are the only dacite lavas or intrusions we have found in the Santa Rosalía area, except for a small body of coherent calcalkaline clinopyroxene dacite at Arroyo Purgatorio (sample SR 17-1, Fig. 3; Table 1A).

The hematized basaltic andesite lavas map unit forms an extensive outcrop west of the Curuglu fault (Fig. 3). It is an ∼200-m-thick section of calcalkaline lavas (sample SR 09-2, Table 1A). They are bright red (Item S6B) and are undated because of the alteration. The section has the same strike and dip as the top of the Mina Santa Teresa section (described below), which may underlie it; but because it is so much more altered (and undated), we did not include it in the Mina Santa Teresa section.

Mina Santa Teresa

The Mina Santa Teresa section is mapped as one unit: trachyandesite lava dome and andesite block-and-ash-flow tuffs and lavas (calcalkaline in composition) with internal stratigraphic variation described here (Fig. 3 and Table 1A).

The base of the Mina Santa Teresa section is a hornblende clinopyroxene trachyandesite lava dome with a coherent interior (sample SR 08-6, Table 1A, Item 7A), mantled by breccias (Fig. 6A). The best age estimate of sample SR 08-6 is the plateau age of 13.08 ± 0.11 Ma calculated with the weighted mean of eight consecutive fractions of the step-heating experiment performed with a groundmass sample (Fig. 8A-3). The data used to calculate the plateau age cluster close to the x-intercept on the (36Ar/40Ar) versus (39Ar/40Ar) correlation diagram; to calculate the straight line fraction, “g” was ignored, yielding an isochron age of 13.03 ± 0.36 Ma (Fig. 8B-3). This less precise isochron age is in agreement within 1σ errors with the plateau age.

The lava dome passes gradationally upward (northward) into hornblende clinopyroxene andesite block-and-ash-flow tuff (sample SR 08-5, Table 1A). This is a nonstratified deposit of dense (nonvesicular) angular blocks up to 1 m across, supported in a matrix of lapilli tuff and tuff of the same composition (Fig. 6B). The block-and-ash-flow tuff is overlain by a hornblende clinopyroxene andesite lava (sample SR 09-1, Table 1A), which forms the top of the section. Plagioclase and clinopyroxene are fresh, and hornblende is altered (Item 7-B [footnote 1]). This hornblende clinopyroxene andesite lava is dated at 13.21 ± 0.21 Ma (sample SR 09-1, Fig. 5). Two 40Ar-39Ar experiments were performed on the same groundmass concentrate of this sample. The experiments yielded reproducible results with a well-defined plateau age of 13.21 ± 0.21 Ma (Fig. 8A-2) for 88% of the 39Ar released in eight fractions. The plateau age is statistically equal to the isochron age (Fig. 8B-2), calculated from data for the combined fractions of the two experiments performed.

The ages of sample SR 08-6, at the base of the Mina Santa Teresa section, and sample SR 09-1, at its top, overlap within two sigma errors (Fig. 5).

Mina Victoria

We mapped four units in the Mina Victoria area (Fig. 3), all calcalkaline (Table 1A). These are described in stratigraphic order.

Andesite peperite intrusion forms the base of the Mina Victoria (sample PEP, Table 1). It has black, glassy andesite blocks showing jigsaw clast texture partially dispersed in tan volcaniclastic sandstone, indicating interaction of the magma with wet sediment (Fig. 4E; Items S6D-1–S6D-3 and Item S7C). Sandstone injections into cracks of widely varying orientations in the blocks indicate that steam explosions were generated as the magma mixed with wet sediment, in places as sills. This intrusion appears to pass gradationally upward into the overlying trachyandesite lava dome map unit, which has the same mineralogy but slightly different chemistry (Table 1A; Item S2).

Trachyandesite lava dome is mapped as one unit (Fig. 3), but the coherent lava dome interior (sample SR 12-4) and the lava dome breccia (sample SR 09-5) are described separately in Table 1A. The coherent interior and the breccia have the same composition, mineralogy, and flow-banded texture (Items S6D-4 and S6D-5 [footnote 1]). The dome interior is devitrified to a gray-green color, with intense flow foliation at its base, becoming less jointed upward but retaining strong flow banding. The coherent dome interior rocks pass gradationally upward into the lava dome breccias, with breccias locally intruded by coherent material. The dome interior was sampled for geochronology; a photomicrograph of the sample is provided in Item 7D. Breccias of the lava dome are conformably overlain by the thick andesite lavas map unit.

The thick andesite lavas map unit (Fig. 3 and Table 1A) consists of ∼10-m-thick lavas, with coherent lava interiors alternating with coarse, nonvesicular flow breccias (Items SD6 and SD6-7 [footnote 1]). All of the lavas in this map unit are andesite but vary in their proportions of clinopyroxene, hornblende, and plagioclase (samples SR 09-7, SR12-2, and SR 12-5, Table 1A). Clinopyroxene is fresh, and hornblende is altered (Item S7E).

Thin scoriaceous basaltic trachyandesite lavas rest conformably on the thick andesite lavas. They contrast with the thick andesite lavas by being much thinner and more vesicular and are capped by scoria lapilli rather than nonvesicular large blocks as in the thick andesite lavas (Items S6D-8 and S6D-9). They are also more mafic and phenocryst poor (sample SR 12-6, Table 1A).

The coherent interior of the Mina Victoria trachyandesite lava dome yielded a 40Ar-39Ar age of 13.81 ± 0.10 Ma (sample SR 12-4, Fig. 5). The experiment conducted with a groundmass concentrate yielded a plateau age of 13.90 ± 0.08 Ma for seven consecutive fractions representing more than 85% of the 39Ar released (Fig. 8A-1). However, the isochron age of 13.81 ± 0.10 Ma (Fig. 8B-1) is taken as the best estimate because the data indicate that the (40Ar/36Ar)i is higher than the atmospheric value.

The Mina Victoria thick andesite lavas yielded an 40Ar-39Ar age of 10.1 ± 0.9 Ma (sample SR 12-2, Fig. 5). The sample came from the coherent interior of a lava in the middle of the section. 40Ar-39Ar experiments were performed with plagioclase and groundmass concentrates. The results of the groundmass and plagioclase are plotted together in the age spectrum and 36Ar/40Ar versus 39Ar/40Ar correlation diagram (Figs. 8A-4 and 8B-4). The results are characterized by low 40Ar radiogenic (40Ar*). The age spectra for the groundmass concentrate indicate severe argon loss that reaches a small plateau segment defined by three fractions with the highest 40Ar* (Fig. 8A-4). Their weighted mean yields a 12.12 ± 0.51 Ma age. The best age estimate of sample SR 12-2 is the plateau age of 10.1 ± 0.9 Ma obtained with all the fractions of the plagioclase step-heating experiment.

Sawlan and Smith (1984) reported K/Ar ages on a sample that plots within our thick andesite lavas (sample 82BSJ104B, Fig. 3B), when we apply the coordinates reported in their paper, and an air photo with a plotted sample location, as kindly provided to us by Michael Sawlan (2019, written commun.). The sample came from a dike, and XRF geochemistry indicates it is an andesite (Sawlan, 2019, written commun.; see geochemical data in Item S2 [footnote 1]). Because only six trace elements were analyzed for this sample, we were not able to confidently determine whether the dike is calcalkaline or rift transitional in chemistry. Sawlan and Smith (1984) reported duplicate whole-rock ages on that sample, of 12.3 ± 0.4 Ma and 12.5 ± 0.4 Ma. This K/Ar age appears to be too old since this dike is within the thick andesite lavas that we date at 10.1 ± 0.9 Ma.

We did not sample the thin scoriaceous basaltic trachyandesite lavas map unit for geochronology because it is very altered due to its scoriaceous nature.

East of Curuglu Fault

We mapped two units in the area along the east side of the Curuglu fault (Fig. 3); both units are calcalkaline (Table 1A). These are described in stratigraphic order.

Basaltic andesite to andesite tuff of Curuglu is a mafic ignimbrite that is >25 m thick, with its base truncated by the Curuglu fault. In outcrop, it is a pink scoria lapilli tuff (Item S6E-1 [footnote 1]), and in thin section, vitroclastic texture is obvious, including brown relict glass with microlites, vesicles, and bubble-wall shards (Item S7F). This mafic ignimbrite is divided into internally massive beds, or flow units, 8–10 m thick, with sharp contacts. The basal flow unit has abundant black to dark-gray aphyric angular blocks that are andesite (sample SR 08-1, Table 1A, Item S6E-1 [footnote 1]). The second flow unit lacks lithics; a whole-rock sample of this has abundant crystals of plagioclase and clinopyroxene and is basaltic andesite (sample SR 08-2, Table 1A, Items S2 and S7F). The second flow unit is density graded, with scoria block concentrations at its top (Item S6E-2). A single one of these scoria blocks was sampled, and it is an aphyric andesite (sample SR 08-3, Table 1A, Item S2). This suggests that the matrix underwent a high degree of crystal enrichment during eruption and transport. The second flow unit is capped by 35-cm-thick, well-sorted lapillistone (Item S6E-2), which records fluvial working before deposition of the third flow unit, which also lacks lithics. Locally, a 10-m-deep channel is incised into the top of the tuff of Curuglu, filled with crudely stratified volcanic lithic breccia-conglomerate and pebbly sandstone, interpreted as fluvial and debris flow deposits (Item S6E-3). The basaltic andesite to andesite tuff of Curuglu is now covered by the tailings pond for the mine.

Mafic ignimbrites (i.e., andesite or basaltic andesite scoria-rich pyroclastic flow deposits) are unusual, but they do occur in arc settings (e.g., Fisher and Schmincke, 1984; Freundt et al., 2000; McCurry and Schmidt, 2001; Busby and Bassett, 2007). They are very small in volume and therefore are not deposited far from the vent. This fact and the presence of block-sized andesite clasts in the basaltic andesite to andesite tuff of Curuglu indicate that this is a vent-proximal mafic ignimbrite. The tuff of Curuglu is conformably overlain by andesite lava.

The andesite lava map unit is a single, 22-m-thick hornblende clinopyroxene andesite lava (sample SR 08-4, Table 1A, Item S2). It is dominated by flow breccia (Item S6E-4), with <20% coherent flow interior. The groundmass and plagioclase are fresh, and the hornblende is altered (Item S7G [footnote 1]).

The geochronology sample from the andesite lava map unit (SR 08-4, Fig. 5) was taken as far away from the Curuglu fault as possible (∼100 m), to avoid alteration. Two experiments were performed on a groundmass concentrate and yielded reproducible results. The plateau age of 11.50 ± 0.18 Ma was calculated with the weighted mean of seven consecutive fractions that represent 84.36% of the 39Ar released (Fig. 8A-5). The combined data for the two experiments define a slightly younger isochron age because the (40Ar/36Ar)i is higher than the atmospheric value (Fig. 8B-5). Therefore, the isochron age of 10.83 ± 0.95 Ma is taken as the best age estimate for sample SR 08-4 (Fig. 5 and Table 1A).

El Morro

The El Morro area (Fig. 3) has outstanding sea cliff and arroyo exposures of ignimbrite referred to by Conly et al. (2005) as the El Morro Tuff, and informally referred to herein as the tuff of El Morro. It is a trachydacite to rhyolite nonwelded to welded ignimbrite and is the oldest map unit that we assign to the rift transitional geochemical group (Table 1A; Fig. 5; Item S2), discussed below. We summarize key features of this ignimbrite herein, and we provide more detailed outcrop and petrographic descriptions in the Supplemental Items (Items S6F-1–S6F-10 and S7H–S7K), in hopes that it will be recognized elsewhere and used as a stratigraphic marker. Only 50 m of the tuff of El Morro is exposed, with the base covered and the top eroded and locally overlain by Boleo Formation sedimentary rocks. We have divided out three map units within the tuff of El Morro (Fig. 3): pink nonwelded lithic rhyolite to trachydacite ignimbrite, purple welded rhyolite ignimbrite, and basaltic trachyandesite lavas interstratified with pink nonwelded lithic ignimbrite (Table 1A).

The pink nonwelded lithic rhyolite to trachydacite ignimbrite dominates the tuff of El Morro and is massive nonwelded ignimbrite (in exposures tens of meters thick) that is weakly welded in intervals where lithics are less abundant (Fig. 7A; Items S6F-1 and S7H–S7I). The purple welded rhyolite ignimbrite consists of lithic-free intervals that are densely welded (Fig. 7B; Items S6F-3, S6F-4, and S7K). The tuff of El Morro is nearly aphyric, with <1% plagioclase in the pumices and <5% plagioclase in the matrix. We analyzed a single pumice block and it is trachydacite in composition (sample SR 15- 6, Table 1A, Item S2 [footnote 1]). Whole-rock samples of pink nonwelded lithic rhyolite to trachydacite ignimbrite (which vary in lithic content) are rhyolite to trachydacite in composition (Table 1A). The lithic fragments are mainly angular clasts of the purple welded rhyolite ignimbrite (Item S6F-8); these clasts also occur sparsely in the purple welded rhyolite ignimbrite (Item S7K). One of these purple welded rhyolite ignimbrite clasts was sampled and is petrographically and chemically identical to the purple welded rhyolite ignimbrite where it forms an in situ sheet (Table 1A). Additionally, the pink nonwelded lithic rhyolite to trachydacite ignimbrite has lenses within it that consist of boulders of pink weakly welded ignimbrite, which are identical to the pink nonwelded lithic rhyolite to trachydacite ignimbrite, where it forms in situ sheets (Item S6F-7). These boulders, in turn, contain lithic fragments of the purple welded rhyolite ignimbrite (Item S6F-9). These relations indicate repeated cannibalization. This is typical of caldera fill, where earlier-erupted ignimbrite that welds on the walls of a caldera, or onto intracaldera blocks that collapse piecemeal, gets broken up and redeposited into the caldera during later stages of eruption and caldera collapse (Lambert, 1974; Lipman, 1976; Branney and Kokelaar, 1994; Schermer and Busby, 1994).

Along the sea cliffs on the east side of the Santa Agueda estuary (Fig. 3), the pink nonwelded lithic rhyolite to trachydacite ignimbrite is cut by large-scale blocky injection structures. These form clastic sills and dikes that are very irregular in shape and orientation (Items S6F-5 and S6F-6). The blocks are solely purple welded rhyolite ignimbrite, indicating that the injection structures formed before any younger units were deposited (for example, the basalt lava and the basaltic andesite lava at Cerro Juanita, Figs. 3 and 5). We interpret these injection structures to record landsliding soon after the purple and pink ignimbrites were deposited, either on an active fault block or in a subsiding caldera.

The pink nonwelded lithic rhyolite to trachydacite ignimbrite has two interstratified basaltic trachyandesite lavas, mapped separately (Fig. 3 and Table 1A). The ignimbrite that encloses them is massive, with no evidence of fluvial reworking; therefore, we infer that little time is represented for deposition of the tuff of El Morro, despite the presence of interstratified lavas. A white pumice block concentration is partially admixed with the uppermost part of the flow-top breccia on the upper lava (Items S6-F1 and S6-F2 [footnote 1]). This may be similar to lee-side coarse pumice lenses in the Taupo ignimbrite, which are found in the lee of topographic obstacles (Walker and Wilson, 1983), which the lava evidently formed.

We did not sample the tuff of El Morro ignimbrite for 40Ar-39Ar geochronology because it is too altered. Conly et al. (2005) referred to the tuff of El Morro as “conglomeratic tuff” (presumably lithic ignimbrite) and obtained a range of K-Ar ages on “clast separates” (indicating volcanic lithics?) of 11.71Ma ± 0.62–8.62 ± 0.21 Ma, and on matrix, overlapping ages of 8.94 ± 0.80–8.22 ± 0.39 Ma. The groundmass of the interstratified basaltic trachyandesite lavas is much fresher than the matrix in the enclosing tuff of El Morro ignimbrite (Item S7-J); so we dated the upper one of those two lavas by 40Ar-39Ar.

The basaltic trachyandesite lava sample SR 15-8 yielded reproducible results and well-defined plateau segments in the two experiments performed on groundmass. The plateau age of 9.75 ± 0.14 Ma, calculated with five consecutive fractions representing 76.6% of the 39Ar released, is statistically indistinguishable from the isochron age of 9.69 ± 0.12 Ma calculated with the combined data of the two experiments (Fig. 8A-7). The isochron age of 9.69 ± 0.12 (Fig. 8B-7) is taken as the best estimate for the age of the basaltic trachyandesite lava (Fig. 5), as well as the ignimbrite (tuff of El Morro) that encloses it. This is older than the age Conly et al. (2005) obtained on matrix from his “El Morro Tuff,” likely because the ignimbrite matrix is altered.

In addition, we chiseled out several kg of pumice blocks from the white pumice block concentration described above (sample SR 15-6, Table 1), and we separated the pumice blocks for zircons (Fig. 9). The yield was small, probably because the pumices are nearly aphyric (sample SR 15-6, Table 1). U-Pb analyses were conducted on the CAMECA IMS 1270 ion microprobe at University of California, Los Angeles (Items S1 and S4 [footnote 1]). The sample included two concordant antecrysts (21.19 Ma ± 0.92 and 12.63 Ma ± 1.08; 1σ uncertainty) and two anomalously young grains with low radiogenic 206Pb; these two grains were excluded from the mean age determinations. The sample yielded a concordant mean weighted 206Pb/238U age of 9.00 Ma ± 0.2 (2σ) based on 11 grains (Fig. 9A) in excellent agreement with the Tera-Wasserburg concordia age of 9.04 Ma ± 0.14 (2σ; Fig. 9B). This is slightly younger than the 40Ar-39Ar age on the interstratified lava, perhaps due to differences in techniques and sample types. More interesting is the result from the Hf isotope work (Fig. 9C), described below in Geochemistry.

Arroyo Purgatorio

The Arroyo Purgatorio area is mapped as one unit, basaltic andesite lava, which we designate as rift transitional in chemistry (Fig. 3A, Table 1A, Item S2 [footnote 1]). It is clearly extrusive because it has flow-top and flow-bottom breccias. In addition, we sampled a body that is too small to map individually; therefore, it is included in the basaltic andesite lava map unit (Fig. 4A). This body is a coherent clinopyroxene dacite that may either represent a lava or an intrusion and is calcalkaline (Table 1A, Item S2). The rocks in this area are highly faulted and altered; thus, stratigraphic and intrusive relations are unclear, and no samples were dated.

Cerro Juanita

The Cerro Juanita area has two map units: basalt lava and an overlying basaltic andesite lava (Fig. 3, Item S2), and we designate both as rift transitional in chemistry (Table 1A).

Basalt lava consists of a single, ∼30-m-thick, plagioclase-rich lava (Table 1A, Item S6G). This lava lacks basal and upper flow breccia and rests on purple, thin-bedded tuff and lapilli tuff. The thickness of this basalt is suggestive of ponding in a graben.

Basaltic andesite lava forms a single lava that is >100 m thick, with a basal flow breccia (Item SG) and its top eroded. The upper preserved part, which was quarried to build the harbor at Santa Rosalía, is more aphyric than the lower part (Table 1A). Therefore, there may be more than one flow unit, although the chemistry is the same, and there are no intervening flow breccias or scoriaceous horizons. If it is one flow, its thickness may indicate ponding in a graben.

The phenocryst-poor part of the basaltic andesite lava was the freshest (Item S7L), so it was chosen for geochronology (sample SR 14-1). The data from two experiments yield reproducible results and very well defined plateau segments (Fig. 8A-8). The plateau age of 9.66 ± 0.21 was calculated with the weighted mean of six consecutive fractions that represent 73.41% of the 39Ar released. The plateau age is statistically identical to the isochron age (9.71 ± 0.34 Ma) calculated with the combined data of the two experiments (Fig. 8B-8). The more precise plateau age of 9.66 ± 0.21 is taken as the best estimate for the basaltic andesite lava of Cerro Juanita (Fig. 5).

Cerro Sombrero Montado

The Cerro Sombrero Montado area has two map units—lithic pumice lapilli tuff and an overlying andesite lava that we designate as rift transitional in chemistry (Fig. 3; Table 1A; Item S2 [footnote 1]). These units dip east (Item S6H), as do the nearby lavas on Cerro Juanita (Fig. 3), but the andesite lava that caps Cerro Sombrero Montado is younger than the basaltic andesite lava on Cerro Juanita (Fig. 5). Like the section on Cerro Juanita, the base of the section at Cerro Sombrero Montado is a pyroclastic rock (lithic pumice lapilli tuff,Fig. 3).

Lithic pumice lapilli tuff. The basal deposit on Cerro Sombrero Montado consists mostly of pumice lapilli and smaller pumice shreds, with ∼20% euhedral or broken crystals and 20% volcanic rock fragments (Table 1A). We did not analyze it for geochemistry because of the abundant volcanic rock fragments (Item S7M). It is exposed directly beneath the overlying, resistant andesite lava, for a thickness of ∼10 m. This is the only hornblende tuff we have found. The predominance of pumice lapilli in the deposit is suggestive of Plinian fall, and if it is a fall deposit, its thickness and the presence of lithics may indicate it is proximal to the vent.

Andesite lava. This map unit consists of a single phenocryst-poor lava >80 m thick (top eroded, Fig. 3), with a basal flow breccia (Table 1A). It has <5% plagioclase phenocrysts, most skeletal or sieve textured, and plagioclase microlites with interstitial clinopyroxene (Item S7N).

The two experiments performed with whole-rock sample SR 14-5 (andesite lava) yielded age spectra with some perturbation. However, a very good plateau age of 8.84 ± 0.16 Ma was obtained from the weighted mean of seven consecutive fractions that represent 87.27% of the 39Ar released (Fig. 8A-9). The distribution of the combined data of the two experiments does not constrain the straight line on the (36Ar/40Ar) versus (39Ar/40Ar) correlation diagram (Fig. 8B-9); for this reason, the plateau age is taken as the best estimate for the age of sample SR 14-5 (Fig. 5).

Santa Rosalía South

The Santa Rosalía south area is a contiguous map area dominated by two main map groups: calcalkaline volcanic and volcaniclastic rocks and high-magnesian andesite (Fig. 4; Table 1B). Each of these two main groups is subdivided into map units (Fig. 4), some of which interfinger or repeat. The high-magnesian andesite lavas clearly overlie the calcalkaline lavas, and the high-magnesian andesite intrusion clearly crosscuts the calcalkaline lavas (Fig. 4).

Calcalkaline volcanic and volcaniclastic rocks. Volcanic lithic fluvial sandstones and volcanic debris flow deposits and undifferentiated sedimentary rocks form the base of the section (Fig. 4). We found no primary volcanic rocks to date in this section. Andesitic volcanic lithic sandstones and conglomerates from the Loreto area southward (Fig. 1) occur in the Lower Member of the Comondú Group (ca. 24–19 Ma; Umhoefer et al., 2001; Drake et al., 2017); however, Loreto is nearly 200 km to the south, and the Comondú Group in the intervening area is poorly known. Furthermore, volcanic fluvial sandstones lie in the “East of Curuglu Fault” section, described above, and intervals of volcanic fluvial sandstone are common in arcs. However, if these sandstones are indeed Lower Member of the Comondú Group, this would indicate that there is a ∼10 m.y. unconformity between the basal sedimentary rocks and the andesite lavas above.

The section of andesite lavas has a relatively continuous red hornblende andesite lava and a lens of dacite block-and-ash-flow tuff (Fig. 4). The andesite lavas were not mapped individually but generally dip eastward. The andesite lavas map unit is ∼200 m thick, and the lavas are all very similar in appearance: dark-colored, with thick flow breccias, and commonly flow banded. All have clinopyroxene and no hornblende (Item S7O [footnote 1]), except for the red hornblende andesite lava (Fig. 4), which has hornblende and no pyroxene (Table 1B). Like the andesite lavas, the andesite block-and-ash-flow tuff in the southeast part of the map has abundant clinopyroxene and no hornblende (Item S7P); thus it probably records flow-front collapse of andesite lava. The dacite block-and-ash-flow tuff within the andesite lava section (Fig. 4 and Table 1B) forms a 15-m-thick lens that is visible from the dirt road in the arroyo ∼1 km to the west, because its light-gray color contrasts with the dark color of the andesite lavas above and below, and it weathers recessively. It forms a deposit of angular blocks in a matrix of lapilli and ash of the same composition (Item S6I-1 [footnote 1]). It has abundant plagioclase phenocrysts and lesser hornblende, largely replaced by opaque minerals (Item S7Q). This is the only silicic volcanic unit we have found in the Santa Rosalía south map area (Fig. 4).

Groundmass sample SL 08-4 was prepared from the calcalkaline andesite lava. Two experiments were performed with reproducible and well-defined broad plateau segments. The plateau age of 10.48 ± 0.09 Ma was obtained with eight consecutive fractions that represent 85.03% of the 39Ar released (Fig. 8A-6). The combined data of the two experiments are well aligned in the (36Ar/40Ar) versus (39Ar/40Ar) correlation diagram defining an isochron age of 10.11 ± 0.14 Ma (Fig. 8B-6); that age is taken as the best estimate for the age of andesite lava sample SL 08-4 (Fig. 5).

High-magnesian andesite. This includes trachyandesite lavas and trachyandesite intrusions (Fig. 4). East-dipping, high-magnesian trachyandesite lavas overlie the calcalkaline andesite lavas as well as the undifferentiated sedimentary rocks on the east flank of Cerro San Lucas, in angular unconformity (samples SL 08-5 and SL 08-7, Fig. 4 and Table 1B). Flow breccias 2–4 m thick (Fig. S6I-2) alternate with coherent interiors of similar thickness. These lavas differ from the calcalkaline andesite lavas by having hornblende as well as clinopyroxene phenocrysts, although the hornblende is commonly pseudomorphed by opaque minerals (Item S7R [footnote 1]). The hornblende in the high-magnesian trachyandesite intrusions (Table 1B) is generally much better preserved, and large (∼2 mm), with fine oscillatory zoning (Fig. 7C). These hornblendes are similar in appearance to those of the 1991–1995 Unzen dacite, Japan, which have finely alternating Fe-rich and Mg-rich zones detected by electron microprobe (Sato et al., 2005). That zoning was interpreted to be the result of repeated convective degassing and fluid recharge of the magma. The small high-magnesian trachyandesite intrusion on the west end of the map area intrudes volcanic debris flow deposits of uncertain age, but the larger high-magnesian trachyandesite intrusion clearly crosscuts the calcalkaline andesite lavas, forming the highest peak in the Cerro San Lucas, with a radio tower on top (note that the road to the tower is not always passable by 4WD). Both plugs represent hypabyssal intrusions, with hornblende ± clinopyroxene phenocrysts set in a groundmass of plagioclase microlites and no plagioclase phenocrysts (Table 1B). The larger trachyandesite intrusion is very well exposed along the road to the tower and has miarolitic cavities (Fig. 6F). These cavities commonly occur in shallow-level granites or pegmatites but may also occur in hypabyssal intrusions (Kiddle et al., 2010), providing evidence of a potentially fertile intrusion (Harris et al., 2004).

A hornblende and a groundmass concentrate were analyzed from the larger clinopyroxene hornblende high-magnesian trachyandesite intrusion (sample SL 08-1), collected near the top of the intrusion (Fig. 4). The age for this sample is 6.06 ± 0.27 Ma (Fig. 5). This age is the weighted mean of two fractions of the experiment performed with the hornblende concentrate (Fig. 8A-10); these fractions represent 74% of the 39Ar released. This age is in agreement within 1σ errors to the weighted mean age (5.92 ± 0.29 Ma) of three consecutive fractions of the experiment performed with the groundmass concentrate and with the isochron age of 6.01 ± 0.86 Ma (Fig. 8B-10) calculated with the fractions released at high temperature of the hornblende and groundmass experiments.

Cenozoic volcanic rocks on the Baja California Peninsula have been divided into geochemical groups based on whole-rock major- and trace-element characteristics. Previously recognized geochemical groups include calcalkaline, adakite, high-magnesian andesite (sometimes called bajaites), niobium-enriched basalt, and tholeiite lavas. The calcalkaline, high-magnesian andesite and adakite magmas reflect subduction-influenced melting or modification of mantle sources (Calmus et al., 2003; Conly et al., 2005; Negrete-Aranda and Cañón-Tapia, 2008).

In the Santa Rosalía area, whole-rock geochemistry shows a range of silica contents, with compositions 52–63 wt% SiO2 dominating (Fig. 10A). Trace-element ratios (Figs. 10B–10D), including multi-element plots normalized to primitive mantle (Sun and McDonough, 1989) (Fig. 11), were used to divide samples into calcalkaline, rift-transitional, and high-magnesian andesite geochemical designations based on the relative abundances of light rare-earth elements and/or heavy rare-earth elements (LREEs/HREEs). These designations correlate temporally with the transition between subduction and rift tectonics (Fig. 5). No niobium-enriched basalts or tholeiites were identified in this or previous studies (Calmus et al., 2003; Conly et al., 2005) of the Santa Rosalía area.

The oldest volcanic rocks in the Santa Rosalía area (13.81–10.11 Ma, Fig. 5) are calcalkaline andesite, basaltic andesite, basaltic trachyandesite, trachyandesite, and dacite (Figs. 10 and 11). These occur in both the Santa Rosalía north (Fig. 4) and Santa Rosalía south (Fig. 5) areas. Calcalkaline chemistries are characterized by moderate alkali contents and enrichment in LREE (Fig. 11D). Subduction of the Farallon slab ceased by ca. 12.5 Ma, but two million years of postsubduction calcalkaline eruptive products are not unreasonable in a region that experienced ∼12 million years of subduction magmatism. As noted above, residual arc signatures persisted for 4 m.y. in Sonora (Till et al., 2009), and calcalkaline volcanism recurred in the Quaternary at Tres Virgenes volcanic field (Fig. 2; Sawlan, 1991; Schmitt et al., 2006, 2010). Similarly, Putirka and Platt (2012) showed an up to 17 m.y. lag between the end of subduction and the disappearance of a subduction-related signal in the geochemistry of volcanic rocks in the Walker Lane–Sierra Nevada region of California.

A younger suite of volcanic rocks in the Santa Rosalía area (9.69–8.84 Ma, Fig. 5) shows greater enrichment of HREE relative to calcalkaline samples; the suite is herein designated rift transitional. These samples occur in the Santa Rosalía north area (Fig. 3). They include rhyolite to trachydacite ignimbrites, as well as basalt, basaltic andesite, andesite and trachyandesite lavas (Fig. 10). Although altered, the ignimbrite samples show higher HREE (Fig. 11C), and the rift transitional lavas have higher HREE and lower LREE, relative to calcalkaline rocks (Fig. 11B). This can be seen in the low La/Yb ratio of these lavas (Fig. 10D). The one exception is one of the two core samples through ignimbrite (tuff of El Morro, sample Montado-02, Table 1, Item S2), which has a distinctive peak in Sr (Fig. 11C) that is dissimilar to most rhyolites in the Baja California Peninsula (Ferrari et al., 2018). Because this peak is not associated with an abundance of plagioclase phenocrysts, the significance of this SR peak is not understood.

Five ca. 9 Ma zircon grains (U-Pb ages described above) yielded εHfT values of +10–+12.8 (mean +10.9; Fig. 9C). The 12.63 Ma antecryst and a low 206Pb grain (7.45 Ma ± 0.25 [1σ]) both yielded similar εHfT values. The 21.19 Ma antecryst yielded a more evolved +8.5 εHfT value. The εHfT values suggest that the tuff of El Morro magma had a significant mantle component admixed to crustal material. The ca. 21 Ma antecryst is more evolved but also shows a significant mantle component. Regionally, adjacent ca. 20–25 Ma Sierra Madre Occidental (SMO) magmas have lower εHfT values (+1 to +5); however, SMO ignimbrite flare-up magmas have between +5 and +8, coincident with greater mantle signature (∼50%; Andrews et al., 2012). The high positive εHfT values are consistent with magma genesis in the rift transition as primitive, mantle-derived melts ascended higher into the crust where they melted and then mixed with mid- and upper-crustal rocks having lower εHfT. The Tuff of El Morro was not derived from the melting of crustal rocks.

The youngest suite of volcanic rocks in the Santa Rosalía area (6.06 ± 0.27 Ma, Fig. 5) consists of high-magnesian andesite lavas and intrusions (Fig. 10A and Table 1), which were found in the Santa Rosalía south area (Fig. 4). The high-magnesian andesites have higher abundances of K and Sr (Fig. 10C) and Ba and Nd relative to calcalkaline samples (Fig. 11A). Trace elements of these rocks (normalized to primitive mantle) display the steepest slope of all the geochemical designations, which can be represented by Sr/Y and La/Yb ratios (Figs. 10C and 10D). The high concentrations of Ba and Sr, relative to other incompatible elements such as Rb and Th, are distinctive (Fig. 11A; Calmus et al., 2003; Conly et al., 2005). The K/Rb ratios in the high-magnesian andesites are markedly higher relative to the other chemical groups (up to 1400) and are comparable to K/Rb values of lavas in the Jaraguay and La Purisima volcanic fields (Rogers and Saunders, 1989; Sawlan, 1991).

There is some overlap in the geochemical characteristics of adakites and high-magnesian andesites, and both are recognized by high Sr/Y and La/Yb ratios, with adakites having lower ratios and higher silica contents (Richards and Kerrich, 2007; Moyen, 2009; Castillo, 2012; Item S8 [footnote 1]). Generally, high-magnesian andesite is a broader category, with higher ranges of observed Ba and Sr values than recognized in adakites (Item S8). We concur with Conly et al. (2005) that Cerro San Lucas rocks, which we date at 6.06 ± 0.27, are high-magnesian andesites.

High-magnesian andesites occur throughout central Baja California (Fig. 1), in the Jaraguay (3.00–10.3 Ma), San Borja (0.57–10.93 Ma), San Ignacio (3.08–5.07 Ma), La Purisima (1.01–9.67 Ma), and Vizcaíno (11.13 Ma) volcanic fields (Benoit et al., 2002; Calmus et al., 2003; Bellon et al., 2006; Pallares et al., 2007, 2008; Castillo, 2008; Negrete-Aranda and Cañón-Tapia, 2008; Negrete-Aranda et al., 2010, 2013; Calmus et al., 2011; Di Luccio et al., 2014). These high-magnesian andesites are all postsubduction in age. Adakites in central Baja California in the Santa Clara volcanic field (8.7–11 Ma; Aguillón-Robles et al., 2001) and in the Jaraguay volcanic field (9.4 Ma; Pallares et al., 2008) (Fig. 1) also postdate subduction. Some adakite-like compositions have been found at Tres Virgenes (ca. 100 ka; Sosa Ceballos et al., 2019). However, syn-subduction adakites are present in central Baja California at La Trinidad (LT, Fig. 1) (Busby et al., 2018b) as well as in mainland Mexico (Ferrari et al., 2018). The 5–6 Ma “adakites” of Bonini and Baldwin (1998) on Santa Margarita Island (SM, Fig. 1) may be more appropriately categorized as high-magnesian andesites, using the criteria given in Item S8 (footnote 1). Sr and Pb isotopic signatures of the high-magnesian andesites of Cerro San Lucas (87Sr/86Sr 0.703723–0.704034; 206Pb/204Pb 18.43–18.61; Conly et al., 2005) are similar to other high-magnesian andesites elsewhere in Baja (87Sr/86Sr 0.70355–0.7444, 206Pb/204Pb 18.64–18.764; Rogers and Saunders, 1989; and 87Sr/86Sr 0.70370–0.70399; 206Pb/204Pb 18.42–18.46; Negrete-Aranda and Cañón-Tapia, 2008).

Petrogenic models for high-magnesian andesites require a process in combination with source that preferentially concentrates LREE and Sr and Ba in magmas, producing high LREE and/or HREE but not similar enrichment of Rb and Th (Sawlan, 1991; Calmus et al., 2003). Proposed models include: (1) slab melting related to opening of an asthenospheric window (Calmus et al., 2003); (2) melting of metasomatized mantle (Sawlan, 1991; Conly et al., 2005; Calmus et al., 2011); and (3) crustal contamination (Richards and Kerrich, 2007). Melting of an amphibole-bearing metasomatized mantle source has been proposed to account for the high K/Rb (up to 3000) for high-magnesian andesites in Baja California (Rogers and Saunders, 1989; Sawlan, 1991; Calmus et al., 2011). Because amphibole does not host Rb and Th, this may explain the preferential concentration in melts of particular incompatible elements (Calmus et al., 2003). Although high Ce/Y ratios are frequently interpreted to suggest crustal contamination, the Ce/Y values of Santa Rosalía samples (Fig. 10B) exceed that of the regional crust, and Sr and Pb isotopic values do not support the interpretation of crustal contamination (Conly et al., 2005).

Geophysical evidence shows that there is currently a remnant slab beneath the central portion of Baja California, where the Guadalupe and Magdalena microplates lay offshore (Fig. 1; Wang et al., 2013; Di Luccio et al., 2014). This spatially correlates with high-magnesian andesites in Baja California (Fig. 1), which have all been produced since subduction ceased (<12.5 Ma; Calmus et al., 2003; Negrete-Aranda and Cañón-Tapia, 2008; Calmus et al., 2011). Negrete-Aranda and Cañón-Tapia (2008) suggest that it is unlikely that all the high-magnesian andesites across Baja California were formed from a single source but rather formed by localized processes due to spatial heterogeneities in postsubduction breakup of the slab, which has apparently spanned 12.5 Ma into the Holocene. For these reasons, we prefer the slab window hypothesis over the metasomatized mantle hypothesis for the high-magnesian andesites in the Santa Rosalía area, even though the petrologic data do not allow us to distinguish between them.

In this section, we discuss key stratigraphic, structural, intrusive, and geochronological relations for the Santa Rosalía area, shown in Figure 12, and place them in the larger context of the Guaymas basin and the plate margin (Fig. 1).

How Is the Arc-to-Rift Transition Recorded in Volcanic and Subvolcanic Intrusions in and around the Santa Rosalía Basin, and How Does This Compare to the Transition Elsewhere in the Gulf of California?

We used geochemistry to divide volcanic and intrusive rocks into a calcalkaline (subduction) assemblage, a rift transitional assemblage (subdivided into lavas and ignimbrites), and a high-magnesium andesite assemblage (subdivided into lavas and intrusions) (Fig. 12). In simple terms, relative to the subduction magmas, the rift transitional magmas show lower HREE and/or LREE ratios more similar to primitive mantle than typical subduction. 40Ar-39Ar geochronology was important to determine the ages of these geochemical suites and to determine if those ages were consistent with plate tectonic models for the timing of the arc-rift transition.

The arc-rift transition in the Santa Rosalía area forms a conformable stratigraphic section that spans ca. 14–9 Ma. The subduction phase was dominated by effusive volcanism, with explosive volcanism (tuff of El Morro) marking the onset of rift transitional volcanism (Fig. 12). Although subduction ceased by ca. 12.5 Ma, calcalkaline volcanism persisted until 10.11 ± 0.14 Ma here, and recurred in the Pleistocene at the Tres Vírgenes volcanic center, which overlies the Santa Rosalía basin fill on its north margin (Fig. 2). Volcanic rock compositions in the Walker Lane (the northward extension of the Gulf of California transtensional rift) show that continental mantle lithosphere with subduction signatures has taken up to ∼20 m.y. to be completely degraded by asthenospheric upwelling due to rifting, with full asthenospheric compositions only appearing 20 m.y. after cessation of subduction, accomplished by northward passage of the Mendocino Triple Junction and opening of a slab window (Putirka and Platt, 2012; Putirka et al., 2012). This progressive degradation takes place in a piecemeal fashion at any one locality (see figure 16 of Putirka et al., 2012), resulting in mixed lithosphere-asthenosphere compositions of volcanic rocks and their included xenoliths. We refer to 9.69 ± 0.12 Ma to 8.84 ± 0.16 Ma rocks in the Santa Rosalía area as “rift transitional” because they record this process. However, the persistence of calcalkaline magmatism until 10.11 ± 0.14 Ma could alternatively be interpreted to record slow subduction of the Magdalena plate, interpreted by Michaud et al. (2006) to have occurred between ca. 12 and ca. 8 Ma.

Calcalkaline rocks in the Santa Rosalía area, and the Comondú arc of Baja California in general, are dominantly intermediate in composition and effusive, except for volcanic rocks of the Lower Comondú, which include Early Miocene silicic ignimbrites erupted in mainland Mexico (cf. Ferrari et al., 2018). Rift transitional rocks in the Santa Rosalía area, in contrast with the calcalkaline rocks, include basalt lava and silicic ignimbrite (tuff of El Morro). This is also true in the broader Gulf of California region. Basalts dated at 11.5–10 Ma extend 700 km along the coast of mainland Mexico (Ferrari et al., 2018). In the northern Gulf of California, rift silicic ignimbrites form two age groups. The older group, erupted at the initiation of rifting, includes (1) the tuff of San Felipe (Stock et al., 1999), dated at 12.50 ± 0.08 Ma (Bennett et al., 2013); (2) the tuff of San Ignacio, erupted synchronous with the tuff of San Felipe (Darin et al., 2016); and (3) the underlying Tuff of Cerro Colorado (Darin et al., 2016), which may also be rift related (its U-Pb zircon age, 14.20 ± 1.60 Ma, has large error). The second age group, at ca. 7–6 Ma, appears to correspond with the rapid development of pull-apart basins in the northern Gulf of California (Bennett et al., 2013). This includes seven rhyolite outflow ignimbrites at Puertecitos (40Ar-39Ar ages ranging from 6.7 to 6.1 Ma; Nagy et al., 1999), two outflow ignimbrites on Isla Tiburon (40Ar-39Ar ages of 6.44 ± 0.05 Ma and 6.01 ± 0.20 Ma; Bennett et al., 2015), as well as two airfall tuffs dated at 7.05 ± 0.03 Ma (40Ar-39Ar single crystal sanidine) and 6.53 ± 0.18 Ma (U-Pb zircon, sensitive high resolution ion microprobe–reverse geometry [SHRIMP-RG] spot analysis) by Bennett et al. (2013).

The tuff of El Morro ignimbrite marks the onset of rift transitional volcanism in the Santa Rosalía area, at 9.69 ± 0.12 Ma (Fig. 5), but it does not correspond in time with either the earlier rift initiation pulse or the later transtensional rift pulse in the northern Gulf of California. Furthermore, we have not been able to find reports of any ignimbrites of a similar age or description anywhere in the Gulf of California, and we have asked several active researchers if they know of any unpublished possibilities, and they do not. However, much of the large tract of volcanic rocks between Santa Rosalía and Loreto (Fig. 1) remains undivided; therefore, future work may identify correlative ignimbrites.

When Did Basin and Range–Gulf Extensional Province Extension Switch to the Transtensional Rift Regime at Santa Rosalía and in the Broader Gulf of California?

It remains unclear when the strain regime changed from orthogonal extension (Basin and Range–Gulf Extensional Province) to transtensional in the Gulf of California. In one model, early (ca. 12–6 Ma) strain was partitioned between NNW-trending dextral strike-slip faults west of the modern Baja California Peninsula (Fig. 1) and NNW-striking normal faults within the proto-gulf of California (Stock and Hodges, 1989; Oskin and Stock, 2003a, 2003b, 2003c; Oskin et al., 2001). In a second model, the strike-slip motion west of the Peninsula was small, and transtension dominated the Gulf of California beginning by 12 Ma (Fletcher et al., 2007; Bennett et al., 2013; Sutherland et al., 2012).

Umhoefer et al. (2018) argue that transtension had begun by ca. 7–8 Ma in the Gulf of California. If this is true, then the youngest volcanic rocks in the Santa Rosalía area (the 6.1 Ma high-magnesian trachyandesites, Fig. 12) must have formed under a regional transtensional regime. However, the tectonic setting of the rest of the volcanic rocks in the Santa Rosalía area (ca. 13.81–8.84 Ma, Figs. 5 and 12) remains unclear. A summary of regional relations is given here before addressing the Santa Rosalía area.

In the northern Gulf of California, normal faulting began at ca. 16–14 Ma on the Main Gulf Escarpment (Rossi et al., 2017), between ca. 16 and 11 Ma to the north of it in the Sierra Juarez (Lee et al., 1996), and at ca. 9–8 Ma to the east of it (Seiler et al., 2011, 2013). A pulse of detachment faulting occurred at ca. 10–15 Ma on the Monte Blanco dome, reactivated from Eocene time (Axen et al., 2000). Bennett et al. (2013, 2015, 2016, 2017) and Bennett and Oskin (2014) have shown that the northern 500 km of the 1500-km-long Gulf of California represented a zone of localized strike-slip faulting, clockwise block rotation, and subsiding pull-part basins from ca. 9–6 Ma, before marine incursion occurred at 6.2 Ma.

In the southern Gulf of California, Ferrari et al. (2013, 2018) infer that 11.5–10 Ma basalts along the eastern side of the Gulf of California were erupted on crust thinned by extension. Early Miocene (20.1–18.3 Ma) intermediate to silicic plutons within the southern Gulf of California are inferred to have been rapidly unroofed by extension (Duque-Trujillo et al., 2015). Other intrusions within the southern Gulf of California show evidence for unroofing at 10–9 Ma (Balestrieri et al., 2017). Drake et al. (2017) find no evidence of significant synvolcanic faulting in early to middle Miocene volcanic rocks on the Baja California Peninsula between Loreto and the north end of Bahía de La Paz (Fig. 1). However, south of La Paz, the high-angle San José del Cabo fault was active between 12 and 10 Ma (Fletcher et al., 2000) and is interpreted to be part of a transtensional shear zone (Fletcher et al., 2007). It had an earlier history as a detachment fault at ca. 17–12 Ma (Bot et al., 2016).

On a local scale, strain is commonly partitioned between normal faults and dextral faults in the Gulf of California as well as its northward extension, the Walker Lane (Axen et al., 2000; cf. Busby, 2013), making it difficult to demonstrate transtension without detailed mapping on a more regional scale than we present herein. Furthermore, it is difficult to demonstrate strike-slip displacement in the field, because piercing points can be difficult to find, and kinematic indicators are not commonly well developed in soft volcaniclastic rocks. It took a decade of mapping to determine that many central Sierra Nevada range front faults are oblique, not normal, and were produced by Walker Lane transtension, not Basin and Range extension (Busby et al., 2013a, 2013b, 2016, 2018a). Additionally, paleomagnetic work is needed to determine if dextral shear has been accommodated by vertical axis block rotations, as is observed in the central Walker Lane (Carlson et al., 2013; Carlson, 2017). In many cases, it is not known whether or not normal faults in the Gulf of California had a dextral component of slip (e.g., see discussion in Axen and Fletcher, 1998). Thus, it remains unclear when the transition from extension to transtension began in the Gulf of California and if the transition was synchronous or diachronous along the length of the Gulf.

On the basis of the mapping presented herein, we cannot determine if the subduction-related rocks were faulted during their accumulation. As stated above, we suspect that the rift-transitional lavas were deposited in extensional or transtensional basins because they are unusually thick. We also infer that the rift-transitional tuff of El Morro was disrupted by landslides prior to eruption of the rift transitional lavas, but that may have been triggered by caldera subsidence, rather than by extension or transtension, as described above. The subduction-related rocks and the rift transitional rocks were tilted and juxtaposed by faults and eroded prior to eruption of the 6.1 Ma high-magnesian trachyandesites and deposition of the Boleo Formation (Fig. 12), but there is too much Santa Rosalía sedimentary basin cover (Fig. 2) to be able to study those faults in detail. For this reason, we do not refer to these rocks as an “arc-oblique rift transition” but use the more general term “arc-rift transition.” There are vast tracts of largely undivided volcanic rock in the ∼300-km-long swath between the Santa Rosalía area and the Loreto area (Fig. 1), and they must be studied in detail before we can fully understand the evolution of the Gulf of California rift.

How Does the Santa Rosalía Basin Stratigraphy Fit in with Regional Stratigraphic Models for Transtensional Rifting?

Umhoefer et al. (2018) presented a model for progressive seaway encroachment along en echelon pull-part basins in the Gulf of California. In their model, pre-rift rocks are overlain by nonmarine sedimentary rocks, in turn overlain by marginal marine evaporates, in turn overlain by marine rocks (presumably including limestones). However, this is the reverse of the stratigraphy in the Santa Rosalía basin (Fig. 12). It has long been known that marine limestone forms the base of the Boleo Formation (poorly constrained as late Miocene; Wilson and Rocha, 1955; Ortlieb and Colletta, 1984), and that these in turn are locally overlain by gypsum (Wilson and Rocha, 1955). Large positive values of δ34S in the gypsum units (Ortlieb and Colletta, 1984) have been interpreted as evidence that they formed by evaporation of seawater in isolated basins (Ochoa-Landín et al., 2000). The overlying clastic rocks that dominate the basin fill (Fig. 12) consist largely of alluvial fan conglomerates and sandstones, although there are fine-grained sedimentary rocks interpreted as marginal marine tidal rhythmites (McCann et al., 2017). Thus, the Santa Rosalía basin does not fit the progressive seaway encroachment model of Umhoefer et al. (2018). Perhaps this is due to its unique setting adjacent to the Guaymas basin, which ruptured much earlier (by 3 Ma) than the rest of the basins in the Gulf of California.

Is There a Temporal Relationship between Slab Window Magmatism, the Onset of Guaymas Seafloor Spreading, and Boleo Cu-Zn-Co-Mn Mineralization in the Santa Rosalía Basin?

In this section, we present arguments that the Santa Rosalía basin may have formed and filled with Boleo Formation and its ore beds very quickly (ca. 100 ka), when seafloor spreading began in the Guaymas basin (ca. 6 Ma, Lizarralde et al., 2007) and when the high-magnesian trachyandesite intrusions were emplaced (6.06 ± 0.27 Ma, Fig. 12). This hypothesis is consistent with its position on the margin of the magma-rich Guaymas rift (Lizarralde et al., 2007), as well as its probable origin as a pull-apart basin (McCann et al., 2017). Attempts to constrain the age of the Boleo Formation include the following:

  • (1) The maximum depositional age of the Boleo Formation is constrained by the youngest volcanic rock we have dated that is overlain by Boleo Formation. This is the Cerro Sombrero Montado andesite lava (rift transitional), dated at 8.84 ± 0.16 Ma (Figs. 5 and 8). The contact is an angular unconformity, with the lava dipping 30° east beneath the Boleo Formation, which dips gently east (Figs. 3 and 12).

  • (2) The Boleo Formation has a single dateable tuff (Cinta Colorada, Fig. 12) that forms a decimeter- to 2-m-thick red marker bed in the northwest quadrant of the basin (Wilson and Rocha, 1955), covering an area of ∼7.5 km2 (Henry et al., 2015). It contains plagioclase crystals, andesitic volcanic rock fragments, and andesitic scoria, all altered, in a hematite cement. The location of its vent has not been found, but a small-volume mafic eruption is highly unlikely to provide an adequate heat source for the Cu-Zn-Co-Mn stratiform ore deposit as Conly et al. (2011) suggested it might. The 40Ar-39Ar date on the Cinta Colorada tuff reported by Holt et al. (2000) has large uncertainty (6.76 ± 0.90 Ma [2σ] isochron age). We think this is due to the alteration, based on examination of about a dozen thin sections from widely scattered localities. Preliminary paleomagnetic data (Holt et al., 2000) show that the Boleo Formation contains a geomagnetic polarity reversal (from normal to reversed), but there are seven of these in the 7.66–5.86 Ma interval allowed by the date on the Cinta Colorada (Holt et al., 2000). They used a series of assumptions to suggest an age of 7.1 Ma for the base of the Boleo Formation (Holt et al., 2000). This number is now widely used in the literature as “well constrained” (Miller and Lizarralde, 2013), but it is not. The base could be as young as 6.3 or 5.9 Ma (the age of the youngest two reversals that overlap with the age constraints on the Cinta Colorada). Furthermore, the sedimentation rate used by Holt et al. (2000) to calculate the age of the base is 280 m/m.y., which is normal for a rift basin but too slow for a pull-apart basin, which subsides and fills at over ten times that rate (Christie-Blick and Biddle, 1985; Pitman and Andrews, 1985; Ingersoll and Busby 1995; Nilsen and Sylvester 1995; Xie and Heller 2009; Allen and Allen, 2013). For example, the transtensional Loreto basin to the south by the town of Loreto (shown on Fig. 1) accommodated 615 m of fan delta sediments in only 100 k.y. (Dorsey et al., 1995).

  • (3) The Boleo Formation consists of monotonous andesitic volcanic lithic sandstone and conglomerate with minor siltstone (Ochoa-Landín et al., 2000; and our personal observations of dozens of thin sections made from samples taken at all stratigraphic levels). Although Wilson and Rocha (1955) describe the sandstones and siltstones as tuffs, there is no pyroclastic material (e.g., shards, pumice and/or scoria, or euhedral free crystals) in the Boleo Formation, and no evidence for contemporaneous eruptive products within the basin (e.g., block-and-ash-flow tuffs, pumiceous pyroclastic flow, or fall deposits) that might be suitable for dating. The only exceptions are the Cinta Colorada lapilli tuff, described above, and light-colored very fine-grained tuffs that lie above and below the Cinta Colorada in Arroyo de la Soledad on the western basin margin. We sampled the fine-grained light-colored tuffs and attempted to obtain zircon for dating, but the tuffs yielded none, perhaps because they are too distal. However, the possibility remains that ca. 7–6 Ma pulse of ignimbrite eruptions ∼350–600 km to the north in the northern Gulf of California (described above) delivered coarse, zircon-bearing fallout ashes to the Santa Rosalía Basin, and these ashes were resedimented rather than forming discrete layers of coarse-grained ash suitable for mineral separation for dating. Detrital zircon work was undertaken to test that possibility.

  • Sandstone samples weighing 88 kg each were collected from five stratigraphic levels, separated for detrital zircon, and analyzed for U-Pb ages by laser ablation–inductively coupled plasma mass spectrometry (LA-ICPMS) (Henry et al., 2017; Henry, 2019). Although ∼315 zircons per sample were analyzed (out of a yield of ∼1000 per sample), only 5–22 zircons per sample are <10 Ma, and of those, nearly all are >9 Ma, which does not improve existing depositional age constraints on the Boleo Formation (which overlies 8.84 Ma lavas in angular unconformity, as noted above). The poor yield of young zircons is consistent with the interpretation that the Boleo Formation was derived from erosion of andesite lavas and volcaniclastic rocks, which generally lack zircon. However, the youngest zircons are ca. 6 Ma in age (Henry, 2019), although these are very few and come from high in the section. This agrees well with our age on the high-magnesian andesite intrusions (6.06 ± 0.27 Ma, Fig. 5) and onset of spreading in the Guaymas basin (ca. 6 Ma, calculated from seafloor spreading rates; Lizarralde et al., 2007). The placement of laser pits in our zircons was generated by a program that guarantees representation of the overall detrital zircon population, which was dominated by 12–30 Ma and Cretaceous zircons (Henry, 2019). Future work will target euhedral zircons to see if we can determine the depositional ages of the base and top of the Boleo Formation.

  • (4) The sedimentology of the Boleo Formation suggests it may have accumulated in a very short time. Miller and Lizarralde (2013) infer that the 2 km of salt in the Guaymas basin could have been deposited in as few as ∼57–115 k.y.; therefore, the 100-m-thick gypsum could have formed in an even shorter timespan. The clastic section is dominantly coarse grained and lacks paleosols or deep erosional unconformities or obvious angular unconformities. This suggests rapid subsidence and sedimentation. Its uniformly andesitic provenance and coarse-grained nature indicate a small drainage area undergoing uplift adjacent to the rapidly subsiding basin; although this occurs in orthogonal rift settings, it is more pronounced in strike-slip or oblique rift settings. The ore beds are associated with the fine-grained sedimentary rocks in the clastic section and are inferred to have formed from brine pools in a marginal marine environment, analogous to the Red Sea (Conly et al., 2006, 2011). These beds lack marine fossils entirely, probably due to a high rate of siliciclastic input, or presence of chemical toxins, or both (McCann et al., 2017).

If the Boleo Formation did indeed take <∼100 k.y. to accumulate, the high-magnesian trachyandesite hypabyssal intrusions could have contributed to the heat source and fluids. The high-magnesian trachyandesite hypabyssal intrusions we mapped (Fig. 4) are too small to provide heat and fluids for a timespan of 100 k.y., but they may only represent cupolas on the top of a much larger intrusion that underlies the basin, and that is a reasonable time span for cooling of a larger intrusion. High-magnesian andesites and adakites are associated with Cu porphyry deposits in Japan and the Philippines (Richards and Kerrich, 2007; Castillo 2012), suggesting that these magmas are fertile. This is consistent with the presence of miarolitic cavities in the high-magnesian trachyandesite intrusion, described above (Fig. 6F), which also provides evidence of fertile intrusions (Kiddle et al., 2010). This is also consistent with the presence of finely alternating zones in its hornblende (Fig. 7C), which may be the result of repeated convective degassing and fluid recharge of the magma, by analogy with Unzen volcano (Sato et al., 2005). We therefore tentatively propose that the high-magnesian trachyandesite intrusions exposed along the southern margin of the Santa Rosalía basin may have provided the heat and fluids for coeval sedimentary stratiform Cu-Co-Zn-Mn Boleo mineralization (Fig. 12).

Alternatively, the heat engine and fluids may have been provided by the onset of seafloor spreading in the adjacent Guaymas basin at 6 Ma. It is not possible to separate the heating that the high-magnesian andesite intrusions may have provided from a regional source such as the onset of rifting in the Guaymas basin because they are coeval. In rifts, evaporites typically form under conditions unique to the latest stages of continental rupture and the onset of seafloor spreading (Evans, 1978; Miller and Lizarralde, 2013). If the Boleo Formation accumulated very rapidly, as discussed above, then the ore beds may have formed at the onset of seafloor spreading in the magma-rich Guaymas basin.

What Does the Magmatic and Structural Evolution of the Santa Rosalía Area Tell Us about the Origin of the Guaymas Basin?

The Guaymas basin is a narrow rift segment (terminology of Hopper and Buck, 1996) that has been robustly magmatic since continental breakup (Lizarralde et al., 2007). Magmatism in the Guaymas basin is anomalous, with ∼8 km of igneous product, compared to mid-ocean ridge production (6 km) of basins to the south in the Gulf of California. Lizarralde et al. (2007) pointed out that magmatic activity in the Gulf of California varied over smaller length scales than could be explained by variations in extension rate or mantle temperature and heat flow. They instead suggested that mantle fertility controlled areas of robust rift and postrift magmatism, and that this may be related to the pre-rift magmatic history. They proposed that the mantle became depleted by Miocene ignimbrite volcanism in the southern Gulf of California. Our onshore data show that subduction-related magmatism continued until ca. 10 Ma in the Guaymas basin; so this may have contributed to mantle fertility and/or hydration.

Miller and Lizarralde (2013) considered evaporites to represent “the first deposition in proto-gulf extensional basins” (p. 285); however, we show here that rift transitional lavas (first identified by Conly, 2005) were likely ponded in extensional basins, and that rift transitional lavas and ignimbrites were juxtaposed against calcalkaline lavas by extension prior to deposition of the Boleo Formation evaporites (Fig. 12). Heat supplied by rift-related magmatism aids in evaporation from brines in the Afar triangle (Orszag-Sperber et al., 1998; Jackson and Cramez, 2000), as may have been the case during magma-rich rifting of the Guaymas basin (Lizarralde et al., 2007), and its onshore equivalent, the Santa Rosalía basin. The “igneous basement” below the east Guaymas evaporite is inferred from seismic properties to consist of extended continental crust with rift-related magmatic intrusions (Lizarralde et al., 2007; Miller and Lizarralde, 2013). By analogy with the Santa Rosalía area, some of the rocks there are likely calcalkaline lavas of the Comondú arc.

Nearly all of the volcanic rocks in the Santa Rosalía area dip eastward (Figs. 3 and 4). These rocks are clearly tilted, as shown by the attitudes of interstratified sandstones and tuffs, as well as ignimbrite compaction fabrics. Eastward tilting of the ca. 14–9 Ma volcanic rocks may have been accomplished by normal slip on a fault that now lies below sea level offshore. This is consistent with the fact that the shape of the east Guaymas evaporite indicates that the basin it formed in was bounded by N-S normal faults, which were active before the onset of northwest-southeast transform faulting and seafloor spreading at a northeast-southwest ridge (Miller and Lizarralde, 2013).

Speculation: Could a Slab Tear or Onset of Transtension Have Influenced High-Magnesian Andesite Magmatism in the Santa Rosalía Area and the Guaymas Basin?

The seismic structure beneath Baja California shows a distinct low-velocity (i.e., hot) zone crossing the Baja California Peninsula at ∼28° North (Di Luccio et al., 2014), the latitude of the Santa Rosalía basin. We note that this lies above the boundary between the Guadalupe and Magdalena slab remnants (Fig. 1), and that the Di Luccio et al. (2014) image of the slab appears to bow out toward the paleotrench at this locality, suggesting that there is a slab window that crosses under the width of the Baja California Peninsula. The high-magnesian andesites in the Santa Rosalía area are further inboard from the trench than any others found so far in Baja California (Fig. 1). We speculate that this may possibly be explained by the presence of a large slab tear in this area. Perhaps this also contributed to the anomalously high magma production at the Guaymas spreading center and its early rifting relative to other basins in the Gulf of California. The high-magnesian andesites represent a pulse of trace-element–enriched magmatism at a time that broadly corresponds to the onset of transtension in the Guaymas rift. This may be similar to the onset of transtension in the western-central Walker Lane (the northern extension of the Gulf of California rift), where trace-element–enriched magmas were focused into a pull-apart basin (Putirka and Busby, 2007; Busby et al., 2018a).

We have used detailed field mapping, in conjunction with petrographic, geochemical, and 40Ar-39Ar geochronological analysis, to characterize the evolution of Miocene volcanic rocks surrounding the Santa Rosalía sedimentary rift basin of the Gulf of California, central Baja California Sur, México (Fig. 12). Our results also contribute to an understanding of the origin of the Guaymas oceanic rift basin, because the Santa Rosalía area represents the onland part of the Guaymas rift basin (Fig. 1).

Volcanic rocks in the Santa Rosalía area record the transition from calcalkaline subduction-related volcanism (13.81–10.11 Ma) to rift transitional volcanism (9.69–8.84 Ma). After an apparent volcanic hiatus of ∼3 m.y., high-magnesian trachyandesite lavas and intrusions were emplaced at 6.06 Ma. Calcalkaline rocks consist of largely intermediate-composition lavas, block-and-ash-flow tuffs, and intrusions, with minor mafic ignimbrite and dacite. The onset of rift transitional volcanism is marked by eruption of the tuff of El Morro, a 9.69 Ma aphyric trachydacite to rhyolite ignimbrite that is nonwelded to weakly welded where it contains abundant lithics and strongly welded where lithics are absent. Repeated self-cannibalization of the welded to weakly welded ignimbrite during its eruption indicates remobilization of the earlier erupted ignimbrite from caldera walls or fault scarps. Landsliding immediately after emplacement of the tuff of El Morro also indicates remobilization due to tectonic activity or caldera subsidence. The age of the tuff of El Morro contrasts with the age of ignimbrites erupted at the onset of rifting elsewhere in the Gulf of California (ca. 12 Ma) and a second ignimbrite pulse at the onset of strong transtension (ca. 7–6 Ma). We have found no other reports of a 9.69 Ma ignimbrite in the Gulf of California or of an undated aphyric ignimbrite. Rift transitional lavas include basalt, basaltic trachyandesite, basaltic andesite, and andesite. Our mapping shows that nearly all of the volcanic rocks dip eastward, indicating they were rotated about an offshore fault, probably related to the N-S normal faults that ponded the east Guaymas evaporites.

Subduction-related calcalkaline rocks have the moderate alkali contents and high LREEs and low HREEs that are commonly associated with subduction. Postsubduction calcalkaline eruptions persisted for ∼2 m.y. after subduction ceased, indicating that subduction-enriched mantle persisted in the area, typical of other arc-rift transitions, where degradation of such mantle lithosphere by rifting or opening of a slab window may take millions of years. Rift transitional rocks show more enrichment of HREE than the calcalkaline rocks, as well as less enrichment of LREE. Additionally, εHfT data on zircons from the tuff of El Morro have a significant primitive (i.e., mantle) component admixed to more evolved, although still fairly primitive, crustal material. These data support the interpretation that the tuff of El Morro ignimbrite eruption marks the onset of rift transitional volcanism in the Santa Rosalía area.

The high-magnesian trachyandesites show incompatible element enrichment of K, Sr, Ba, and Nd. The trace-element abundances are depleted in HREE relative to the other groups and have high concentrations of Ba and Sr relative to Rb and Th. Melt generation was likely accomplished by a combination of slab melting related to opening of an asthenospheric window and melting of subduction metasomatized mantle. A remnant slab imaged beneath the central portion of Baja California spatially correlates with high-magnesian andesites. However, the high-magnesian trachyandesites in the Santa Rosalía area are the most inboard examples identified to date. We speculate that these were generated by a hot zone, imaged between the Guadalupe and Magdalena slab remnants, that extends eastward to the Gulf of California, which we interpret as a slab tear. We speculate that this hot zone may also have contributed to early rifting and anomalously high magma production at the Guaymas spreading center.

Sedimentary rocks of the Santa Rosalía basin overlie the eastward-tilted calcalkaline and rift transitional rocks in angular unconformity. The basin fill is dominated by the Boleo Formation, consisting largely of coarse-grained andesitic lithic sandstones and conglomerates, with basal limestone and gypsum, and with stratiform Cu-Co-Zn-Mn ore beds at several stratigraphic levels. The basal Boleo Formation gypsums formed on the margin of the Guaymas basin, where evaporates accumulated to a thickness of 2 km, probably in tens of thousands of years. Previous work has suggested that the ore beds were produced by brine pools. We present the preliminary interpretation that the Boleo Formation accumulated very rapidly (<100 ka) in a pull-part basin, coeval with the onset of seafloor spreading in the “magma-rich” Guaymas basin. This seafloor spreading may have provided the heat engine and fluids for mineralization in the Boleo Formation, perhaps aided by the emplacement of fertile high-magnesian andesite intrusions.

This work was funded largely by the Baja Basins National Science Foundation (NSF) Research Experience for Undergraduates (REU) grant 151120 to principal investigator Tina Niemi and co-principal investigator Cathy Busby. Partial support from NSF grant EAR-1545676 (to Busby and Susan DeBari) is also gratefully acknowledged for providing some of the funds for geochronology and for supporting the participation of Busby’s postdoctoral researcher, Sarah Medynski. A 2017–2018 UC Mexus-Conacyt Collaborative Research Grant (to Busby and López Martínez) funded the final stages of this work. This research was done over three years by REU mentors Busby, Graettinger, López Martínez, Medynski, and Niemi. The six undergraduate students and one M.S. student coauthor are listed in alphabetical order (Andrews, Bowman, Gutierrez, Henry, Lodes, Ojeda, and Rice). The undergraduate students participated for one year each, during their junior years, and the M.S. student (Henry) participated both as an undergraduate and as an M.S. student. The addresses of their undergraduate departmental homes are given here, to acknowledge these departments for giving these students the schedule flexibility needed to participate, and for the mentoring they provided before, during, and after their REU year. We thank the Minera Boleo for housing and feeding us and providing a classroom during the field work each January (2015–2017). Tina Niemi organized all of the logistics of the REU. We also thank Valente Salgado Muñoz, who facilitated our stay at the mine and access to outcrops and provided the drill-core samples of volcanic rocks. Thanks to Dr. Genaro Martinez Gutiérrez for his help with compiling and drafting the maps. We acknowledge the participation of CICESE’s personnel, A. Susana Rosas Montoya and Gabriel Rendón Márquez for sample preparation, and Miguel Angel García García, who performed all the 40Ar-39Ar experiments.

Very helpful formal reviews by Paul Umhoefer and Michael Sawlan, and an informal review by Keith Putirka, are gratefully acknowledged. Discussions with Dan Lizarralde, Joann Stock, Luca Ferrari, Michael Oskin, and Marty Grove were also very helpful.

1Supplemental Items. Item S1: Analytical methods, including XRF and ICP-MS whole-rock geochemical analysis, 40Ar-39Ar geochronological analytical methods, and zircon isotopic methods. Item S2: Excel spreadsheet of whole-rock geochemical data, including XRF and LA-ICPMS data. Item S3: 40Ar-39Ar geochronological data: Summary table, and data tables and plots for each sample. Item S4: U-Pb zircon data. Item S5. Hf-isotope zircon data. Item S6: Outcrop photos. Item 7: Photomicrographs, including examples of all samples dated by 40Ar-39Ar. Item S8: Summary of published definitions of adakites and high magnesium andesites. Item S9: KMZ files for geologic maps. Please visit to access the supplemental material, and contact with any questions.
Science Editor: Shanaka de Silva
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.