Radial and linear dike swarms in the eroded roots of volcanoes and along rift zones are sensitive structural indicators of conduit and eruption geometry that can record regional paleostress orientations. Compositionally diverse dikes and larger intrusions that radiate westward from the polycyclic Platoro caldera complex in the Southern Rocky Mountain volcanic field (southwestern United States) merge in structural trend, composition, and age with the enormous but little-studied Dulce swarm of trachybasaltic dikes that continue southwest and south for ∼125 km along the eastern margin of the Colorado Plateau from southern Colorado into northern New Mexico. Some Dulce dikes, though only 1–2 m thick, are traceable for 20 km. More than 200 dikes of the Platoro-Dulce swarm are depicted on regional maps, but only a few compositions and ages have been published previously, and relations to Platoro caldera have not been evaluated. Despite complications from deuteric alteration, bulk compositions of Platoro-Dulce dikes (105 new X-ray fluorescence and inductively coupled plasma mass spectrometry analyses) become more mafic and alkalic with distance from the caldera. Fifty-eight (58) new 40Ar/39Ar ages provide insight into the timing of dike emplacement in relation to evolution of Platoro caldera (source of six regional ignimbrites between 30.3 and 28.8 Ma). The majority of Dulce dikes were emplaced during a brief period (26.5–25.0 Ma) of postcaldera magmatism. Some northeast-trending dikes yield ages as old as 27.5 Ma, and the northernmost north-trending dikes have younger ages (20.1–18.6 Ma). In contrast to high-K lamprophyres farther west on the Colorado Plateau, the Dulce dikes are trachybasalts that contain only anhydrous phenocrysts (clinopyroxene, olivine). Dikes radial to Platoro caldera range from pyroxene- and hornblende-bearing andesite to sanidine dacite, mostly more silicic than trachybasalts of the Dulce swarm. Some distal andesite dikes have ages (31.2–30.4 Ma) similar to those of late precaldera lavas; ages of other proximal dikes (29.2–27.5 Ma) are akin to those of caldera-filling lavas and the oldest Dulce dikes. The largest radial dikes are dacites that have yet younger sanidine 40Ar/39Ar ages (26.5–26.4 Ma), similar to those of the main Dulce swarm.

The older andesitic dikes and precaldera lavas record the inception of a long-lived upper-crustal magmatic locus at Platoro. This system peaked in magmatic output during ignimbrite eruptions but remained intermittently active for at least an additional 9 m.y. Platoro magmatism began to decline at ca. 26 Ma, concurrent with initial basaltic volcanism and regional extension along the Rio Grande rift, but no basalt is known to have erupted proximal to Platoro caldera prior to ca. 20 Ma, just as silicic activity terminated at this magmatic locus. The large numbers and lengths of the radial andesitic-dacitic dikes, in comparison to the absence of similar features at other calderas of the San Juan volcanic locus, may reflect location of the Platoro system peripheral to the main upper-crustal San Juan batholith recorded by gravity data, as well as its proximity to the axis of early rifting. Spatial, temporal, and genetic links between Platoro radial dikes and the linear Dulce swarm suggest that they represent an interconnected regional-scale magmatic suite related to prolonged assembly and solidification of an arc-related subcaldera batholith concurrently with a transition to regional extension. Emplacement of such widespread dikes during the late evolution of a subcaldera batholith could generate earthquakes and trigger dispersed small eruptions. Such events would constitute little-appreciated magmato-tectonic hazards near dormant calderas such as Valles, Long Valley, or Yellowstone (western USA).

Radial and linear dike swarms in the eroded roots of central volcanoes and along rift zones have long been recognized as structures that document the geometry of conduits for eruptions and provide records of paleostress geometry (Nakamura, 1977; Aldrich et al., 1986; Acocella, 2014). In comparison, dike swarms appear to be relatively uncommon at large ignimbrite calderas (Smith and Bailey, 1968; Cole et al., 2005), perhaps because growth of batholithic-scale magma bodies beneath calderas decouples the overlying crust from regional stress geometry (Christiansen et al., 1965; Steven and Lipman, 1976). Even ignimbrite systems that erupted during regional extension, such as Valles, Yellowstone, and Long Valley calderas in the western USA, typically host only sparse fissure-controlled magmatism (the Mono-Inyo chain north of Long Valley being an exception).

In contrast, an enormous system of long-recognized but little-studied dikes in the southwestern USA, which radiate from the Oligocene Platoro caldera in Colorado (Fig. 1) to merge with the Dulce dike swarm that continues ∼125 km into New Mexico (Fig. 2), provides exceptional opportunities to explore magmato-tectonic–temporal links between dike emplacement and a major ignimbrite center. Questions motivating our study include: (1) when are dikes emplaced during a caldera cycle, (2) how far can magma migrate outward from caldera systems, and (3) how do dike compositions vary in relation to age and distance from a caldera locus? The Platoro-Dulce dikes are interpreted here as providing a unique regional-scale record of uplift during prolonged emplacement and solidification of an arc-related subcaldera batholith concurrently with a transition to regional extension.

Although available geologic maps depict distribution of the Platoro-Dulce dikes fairly reliably (Fig. 2), the overall dike swarm has not previously been studied in detail. Several publications include compositions and radiogenic ages of variable precision for a few dikes near the Colorado–New Mexico border, but these were only parts of data sets used primarily to interpret regional paleostress geometry and mafic-magma petrogenesis (Aldrich et al., 1986; Gibson et al., 1993; Gonzales, 2015; Gonzales and Lake, 2017). This paper presents new petrologic data (compositions for 105 widely distributed samples) and 58 new 40Ar/39Ar ages for the overall areal extent of the Platoro-Dulce dike swarm in relation to concurrent regional magmato-tectonic evolution (Tables 12; Supplemental Files 1–51).

Interpretation of the Platoro-Dulce dikes in relation to Cenozoic development of the Southern Rocky Mountains depends on a broad framework of magmato-tectonic events in the Cordilleran USA. As the Mesozoic subduction system between the North American and eastern Pacific (Farallon) plates flattened during the Late Cretaceous and early Cenozoic (Lipman et al., 1972; Coney and Reynolds, 1977), crustal compression generated basement-cored north-trending uplifts that initially defined the Southern Rocky Mountains (Tweto, 1975; Cather, 2004). The eastward migration of compressional tectonics was accompanied by scattered volcanic eruptions and associated intrusions (Mutschler et al., 1987).

A renewed flare-up of continental-arc magmatism in the eastern Cordillera was transgressive in time and space, beginning at ca. 55 Ma in the northern Rockies, migrating southward, and reaching Colorado at ca. 40 Ma (Lipman, 1980; McIntosh and Chapin, 2004). Large ignimbrite eruptions were associated with intermediate-composition lavas (andesite-dacite) and upper-crustal granitoid intrusions, constituting a typical high-K continental-margin arc suite, similar to concurrent magmatism farther west in Nevada and Utah, and comparable to the younger Altiplano Volcanic Complex of the Andes (Best et al., 2016). In all of these areas, the ignimbrite flare-ups occurred during destabilization of the low-angle subduction geometry, just preceding a transition to regional extension.

In the Southern Rocky Mountain volcanic field (SRMVF), initial intermediate-composition lavas (Conejos Formation) erupted from clusters of central volcanoes, many later becoming loci for ignimbrite eruptions (Lipman et al., 1978; Lipman and Bachmann, 2015). At least 28 ignimbrites of high-K calc-alkaline type with individual volumes of 100–5000 km3 erupted between 37 and 27 Ma. Most source calderas are on the western flank of the broad crest of Laramide-age basement uplifts (Fig. 1), perhaps related to westward retreat or rollback of the foundering Farallon slab (Coney and Reynolds, 1977; Ricketts et al., 2016). Shallow granitoid intrusions are exposed at near-roof levels within or adjacent to many of the calderas. The widespread distribution and relatively uniform thickness of the outflow ignimbrites demonstrate that these eruptions predated inception of major extensional faulting along the Rio Grande rift.

Igneous suites associated with the multicylcic Platoro caldera (Lipman, 1975) include precaldera lavas, ignimbrites, postcollapse lavas and granitoid intrusions, dikes of the Platoro-Dulce swarm, and Miocene basaltic lavas (Hinsdale Formation). Notably, lavas as mafic as basalt are nearly absent among Oligocene eruptions of the SRMVF, suggesting that mantle-derived components mingled efficiently with crustal melts (Lipman et al., 1978; Riciputi et al., 1995). Basalts began to erupt only at ca. 26 Ma, concurrent with initial prominent extension along the Rio Grande rift (Lipman and Mehnert, 1975; Turner et al., 2019).

Precaldera Lavas

Early lavas in the vicinity of Platoro caldera (Conejos Formation, 35–30 Ma; Table 2) are basaltic andesite to dacite (>54% SiO2; Colucci et al., 1991), more silicic than most Dulce dikes. A cluster of Conejos volcanoes is inferred to have grown within the area now occupied by Platoro caldera, as documented by outward-dipping flanks preserved along caldera margins (Lipman, 1975). Such an interpretation is consistent with the convergence of newly identified Conejos-age dikes toward a locus within the caldera (Fig. 2). Conejos andesites in the Platoro area vary from aphyric to highly porphyritic, including both anhydrous (plagioclase-pyroxene) and hydrous (hornblende) types, but olivine andesites have not been recognized. Although basalt is rare among precaldera Conejos lavas regionally, informative comparisons with Dulce dikes and Hinsdale basalts are provided by olivine-bearing rocks from the 33–32-Ma Summer Coon volcano just north of Platoro (Fig. 1; Lipman, 1968; Parker et al., 2005; Lake and Farmer, 2015).

Platoro Caldera and Associated Intrusions

Platoro caldera (Fig. 3), the most southerly and oldest ignimbrite center in the San Juan locus of the SRMVF (Lipman, 1975; Dungan et al., 1989; Lipman et al., 1996), is unique in the number of large ignimbrites erupted from a common site: six dacitic tuffs (∼75–1000 km3) between 30.3 and 28.8 Ma. Andesitic lavas continued to erupt between outflow ignimbrites and ponded within the caldera. Shallow plutons of fine-grained equigranular to porphyritic monzonite within and adjacent to Platoro caldera (Lipman, 1975) have crystallization ages close to that of the culminating ignimbrite eruption (Chiquito Peak Tuff), filling of the caldera by ponded lavas, and intrusion-driven resurgent uplift of a triangular hinged block of caldera floor (Table 2; Gilmer et al., 2018). Dikes that radiate westward from Platoro caldera (Fig. 2) vary from aphyric andesites that appear similar to the caldera-filling lavas, to coarsely porphyritic silicic dacites. The most distal andesitic dikes with Platoro affinities extend southwestward beyond the preserved volcanic cover and merge in areal extent with trachybasaltic dikes of the Dulce swarm (Fig. 2).

Platoro caldera is unusual in its record of prolonged postcollapse volcanic and intrusive activity, intermittently from 28.8 to ca. 20 Ma and ranging from andesitic to silicic rhyolitic (Fig. 4; Table 2). In contrast, eruptive activity waned within a few hundred thousand years after ignimbrite eruptions at most other caldera systems in the SRMVF (Lipman, 2007). The postcollapse magmatic history at Platoro (Fig. 5) overlaps in time with the regional transition to extension along the northern Rio Grande rift at ca. 26 Ma (Lipman et al., 1970; Thompson et al., 1991; Gibson et al., 1993). The oldest cooling (uplift) ages along the rift at ca. 25 Ma (Ricketts et al., 2016) coincide with that of a broad magmatic flare-up along the Colorado–New Mexico border (i.e., Navajo volcanic field, Dulce dikes, Questa caldera, Spanish Peaks; Gonzales, 2015; Zimmerer and McIntosh, 2012a; Penn and Lindsey, 1996). Concurrently, the petrologic assemblage in the San Juan region became broadly bimodal (Hinsdale Formation), including trachybasalt lavas and high-silica rhyolites. Space-time-compositional variations among dikes of the Platoro-Dulce system provide additional perspectives on the regional transition from continental-arc to extensional magmatism.

Dulce Dike Swarm

The Dulce dikes are defined here as the arcuate linear swarm of trachybasalt and basaltic trachyandesite dikes that intrude Mesozoic sedimentary strata along the east margin of the Colorado Plateau for ∼100 km from southern Colorado into northern New Mexico (Fig. 2). Some Dulce dikes, though only 1–2 m thick, are traceable for 20 km, and their regional extent was recognized at least as early as the mid–20th century (Dane, 1948; Wood et al. 1948). The Dulce swarm is diffuse, as much as 25 km wide near the Colorado–New Mexico state line where as many as 20 dikes have been depicted along east-west transects. The Dulce dikes form linear rather than radial trends, are thinner but more laterally continuous than the Platoro dikes, and are intrusive into Cretaceous strata along the eastern margin of the Colorado Plateau.

Prior to our work, dikes of the Dulce swarm had been considered distinct in composition, age, geometry, and geologic setting from the Platoro caldera locus. Published regional studies interpreted Dulce dikes as a high-K lamprophyric suite, including minette, vogesite, and kersantite characterized by phenocrysts of amphibole and phlogopite (Gibson et al., 1993; Gonzales and Lake, 2017), but our petrographic and chemical data (Supplemental File 1 [footnote 1]) show that although the Dulce dikes are more mafic than caldera-related San Juan rocks, they are not highly potassic and contain only anhydrous mafic phenocrysts (olivine, clinopyroxene). Mica and amphibole are present in many Dulce samples, but only in late-crystallized groundmass.

Early-Rift Lavas

Important rocks for comparison with Dulce dikes are trachybasaltic to trachyandesitic lavas of the Hinsdale Formation that spread widely in the southeastern San Juan region and adjacent Rio Grande rift (Larsen and Cross, 1956; Lipman and Mehnert, 1975). Recent data suggest pulses at 26–25 Ma and at 21–19 Ma (Turner et al., 2019). The earliest basalts, including subalkaline olivine tholeiites, probably came from vents within the present-day Rio Grande rift (Thompson et al., 1991), but no basalt appears to have erupted within or proximal to Platoro caldera before ca. 21 Ma. Like the Dulce dikes, many of these lavas contain olivine, but they lack hydrous groundmass minerals. They constitute a proximal rift-related magmatic suite, interpreted as counterparts in composition and age to the more distal Dulce dikes across the Continental Divide.

Sampling of the Platoro-Dulce dike system was designed to test for age and compositional variations longitudinally with distance from the caldera, as well as along lateral transects (Supplemental Files 1–5 [footnote 1]). No regional-scale remapping of the lengthy Dulce swarm was attempted; focus was on representative sampling, guided by regional maps (Dane, 1948; Wood et al., 1948; Steven et al., 1974). Many dikes depicted on these maps were readily located, but some sizable ones were not found. We also checked several mapped round or elliptical sites, hoping to locate vents. One site (sample 16L-3; 37°7.66′N, 106°52.11′W) is a hill-capping sill of Dulce-type basaltic trachyandesite; another hill-forming site (37°18.10′N, 106°53.45′W) is an erosional outlier of older Conejos andesitic lava and breccia. One possible vent root is a wide and coarsely crystalized exposure of trachyandesite (∼10 m, 57.6% SiO2) in Taylor Canyon (sample 16L-48; 37°11.41′N, 106°3.22′W) within an 8-km-long dike that is elsewhere only 1–2 m wide and lower in SiO2 (∼53%).

Rock Chemistry

New X-ray fluorescence (XRF) and inductively coupled plasma mass spectrometry (ICP-MS) analyses for the dikes (Table 1; Supplemental Files 1–2 [footnote 1]; 105 total), along with prior data, document sizable variations that are inferred to record tectonic setting and processes of magma generation. Evaluation of primary composition is problematic, however, especially for Dulce dikes. These commonly contain deuterically altered mafic minerals, calcite in vesicles and replacing groundmass, erratic alkali ratios, and high volatile contents compared to lavas. Values for loss on ignition (LOI) are typically >5%, some >10%. Previously unpublished analyses obtained by others in the 1960s have values of CO2 as high as 7.8% (Supplemental Files 3–4). Accordingly, all analyses have been calculated volatile free to facilitate comparisons.

Comparisons with samples collected from the same apparent dike site during prior studies agree fairly well (Supplemental File 1 [footnote 1]), especially in light of diverse methods at different labs and ambiguity about some locations. Agreement is also good for most samples along strike of a single mapped dike, although contrasts between paired samples (15L-39, 16L-52) from one NE-trending dike, depicted as continuous for 6 km, suggest presence of a composite dike or two closely aligned ones. Comparisons between fine-grained margin and coarser interior of dikes were inconsistent. The margin and interior of one dike (samples 15L-40A, 15L-40B) agree closely. In contrast, similarly paired samples from another (16L-9A, 16L-9B) differ substantially in major oxides, probably due to abundant calcite, high LOI values, and other alteration; trace elements agree more closely.

Age Determinations

New 40Ar/39Ar ages for Platoro-Dulce dikes and related rocks of Platoro caldera (58 total; Table 2; Supplemental File 5 [footnote 1]) were determined at the New Mexico Geochronology Research Laboratory (Socorro, New Mexico) by methods similar to our prior work in the SRMVF (e.g., Lipman et al., 2015). The 40Ar/39Ar ages are calculated relative to the 28.201 Ma FC-2 interlaboratory standard (Kuiper et al., 2008), in part because this monitor age seems better intercalibrated with U-Pb zircon dating (e.g., Wotzlaw et al., 2013). Table 2 summarizes ages for the Platoro-Dulce dikes and related rocks, and indicates the highest-quality ages based on Ar systematics. Two-sigma (2σ) uncertainties are reported for individual ages in the text and tables, whereas generalized ages for groups of samples are listed without uncertainties for simplicity. Mineral and groundmass preparation techniques, full analytical methods, data tables, and plots (ideograms, spectra, and inverse isochrons) are in Supplemental File 5.

Accuracy and reproducibility of the 40Ar/39Ar ages were tested by dating mineral pairs in the same sample when available (e.g., sample 15L-32, biotite and hornblende), by comparing the new 40Ar/39Ar ages to published ages for the same site (e.g., sample 11L-27 versus MD-8a), or dating the same sample using different mass spectrometers (e.g., sample 16L-62). In most tests, the compared ages are indistinguishable at 2σ uncertainties. Despite the large number of exposed dikes, no intersecting pairs were found; thus, 40Ar/39Ar ages could not be compared to crosscutting relations. However, the dated dikes and larger intrusions within Platoro caldera or intruded into older Conejos lavas yielded younger ages than their wall rocks.

Areal Distribution of Dike Samples

Sample sites for the Platoro-Dulce swarm have been subdivided into nine semi-equal geographic segments (Fig. 2; Table 3), based largely on dike orientation and access limitations, in order to evaluate variations as a function of distance from the Platoro locus and orthogonally across the dike swarm. Chemical compositions, phenocrysts, and textures of dominant dike types tend to differ from those in adjacent segments, especially for the more proximal and distal areas, although a few dikes within each segment have compositions more typical of neighboring ones. Additionally, some lengthy dikes cross segment boundaries, and assignment becomes arbitrary.

Numerous dikes that range in composition from andesite to silicic dacite and rare rhyolite radiate from several of the granitoid plutons in the vicinity of Platoro caldera (Fig. 2). Especially conspicuous are the dikes that radiate northwest to southwest for 20–25 km from a locus roughly coincident with the Alamosa River pluton, the largest intrusion at Platoro (Fig. 3). Some additional dikes (not discussed here) are crudely radial to the Cat Creek pluton, just east of Platoro caldera. The distribution of these dikes was delineated in conjunction with regional mapping (Steven et al., 1974) and associated study of Platoro caldera (Lipman, 1974, 1975). More than 100 dikes, including andesite and dacite with diverse phenocryst assemblages, were depicted at a scale of 1:48,000, as was their trend toward the Dulce swarm, beyond the preserved extent of volcanic rocks (Lipman, 1975, p. 87). The radial dikes west of Platoro caldera had not been previously studied in any detail; compositions were estimated largely from hand-sample comparisons with chemically analyzed lavas. Only two analyses of dikes (both dacite) had been published (Patton, 1917, p. 31; Lipman, 1975, his table 10); no analyses were obtained for any andesitic dikes. Additional petrologic studies at Platoro in the 1980s were summarized in a field guide (Dungan et al., 1989), but published analytical data are only for precaldera lavas (Colucci et al., 1991).

Dikes analyzed for the present study (Supplemental File 1 [footnote 1]), along with published lava compositions, show that Platoro andesites and dacites form a coherent high-K calc-alkaline suite (54%–61% SiO2), similar to other intermediate-composition rocks of the SRMVF, that straddles the boundary with trachytic compositions (Fig. 6). All of the Platoro rocks are described here as “andesite” and “dacite”, rather than trachyandesite or trachydacite, for brevity and in part to contrast with the more alkalic character of many Dulce dikes. The Platoro-area radial dikes form better-defined linear arrays on alkali-silica and other compositional-variation diagrams than the Dulce dikes (Fig. 6). Other than slightly higher SiO2 contents (58%–61%) of dated hornblende-phyric dikes, no compositional distinction is evident between the relatively few Platoro-area dikes that can be demonstrated by stratigraphy or geochronology to be precaldera (Conejos) versus those of postcaldera or uncertain ages. More detailed field, petrologic, and age data for the Platoro-area dikes would be desirable, especially for less-studied areas adjacent to and west of the Continental Divide and to distinguish more clearly between precaldera versus postcaldera dikes.


Dikes mapped as andesite (Lipman, 1974) include dark-gray aphyric and texturally diverse porphyritic rocks. The porphyritic andesites typically contain phenocrystic plagioclase and augite, or plagioclase and hornblende, but lack biotite. No olivine-bearing dikes (or lavas) of basaltic composition were recognized in the Platoro area other than the younger Miocene basaltic lavas of the Hinsdale Formation. Andesitic dikes are the most numerous and widespread type in the Platoro area, and they are traceable farther outward than more-silicic dikes. Several dikes with Platoro petrologic characteristics (Supplemental File 1 [footnote 1]; samples 16L-1, 16L-22) are present to the southwest beyond the erosionally preserved margins of San Juan volcanic rocks. Most andesitic dikes are narrow, rarely more than a few meters thick, and traceable along strike for only short distances. Additional dikes of this group undoubtedly were missed during mapping, in contrast to the thick dacites, most of which likely have been identified.

The andesitic dikes that radiate westward from the Alamosa River pluton are of several ages that have been challenging to evaluate. Many of the andesitic dikes lack K-bearing phenocrysts or unaltered groundmass suitable for 40Ar/39Ar measurement, and only a few ages have been successfully determined. Some proximal dikes of the western radial group were emplaced concurrently with postcaldera andesitic volcanism, as documented both by stratigraphy and ages, but several large radial dikes are precaldera, associated with Conejos volcanoes.

Confirmed postcollapse dikes at Platoro include south-trending andesites that intrude caldera-fill lavas and volcaniclastic rocks within the southwestern caldera margin. Several have yielded groundmass 40Ar/39Ar ages between 29 and 27 Ma (Table 2), consistent with timing for postcaldera lava eruptions. Without age determinations, however, many andesitic dikes that radiate westward from Platoro are difficult to distinguish confidently as postcaldera versus precaldera. Relatively few dikes of the Platoro suite have been sampled for age or petrologic study, and only reconnaissance mapping is available for large areas west of Platoro, especially west of the Continental Divide. As a result, the westward extent of postcaldera andesite dikes remains poorly constrained.

Dikes containing hornblende or large tabular plagioclase phenocrysts are especially likely to be of Conejos age, because these mineral assemblages are uncommon in postcaldera lavas within and near Platoro. Two proximal hornblende-phyric dikes that intrude volcaniclastic rocks just beyond the southwestern caldera margin yielded variable-quality ages of 32–31 Ma (Table 2, samples 16L-32, 16L-35), clearly recording their relation to precaldera volcanoes. Two SW-trending hornblende-phyric dikes (samples 16L-51, 16KA-02), ∼20 km from the margin of Platoro caldera (Fig. 7A), have similar late-precaldera (Conejos) ages at ca. 31 Ma, as has an even-more-distal aphyric andesite dike of Platoro compositional affinity that intrudes Cretaceous sedimentary strata beyond the preserved volcanic rocks (sample 16L-1). An isolated south-trending dike of mafic dacite along Jim Creek 6 km southeast of the caldera (sample 16L-55), even though atypical for the Conejos Formation in containing biotite and sparse small phenocrysts of sanidine, yielded a sanidine age of 30.36 ± 0.03 Ma, further confirming early development of widespread radial dikes at the Platoro locus prior to ignimbrite eruptions.

In contrast, another SW-trending plagioclase-phyric andesite dike of Conejos petrologic affinity (sample 16L-22), also intrusive into Cretaceous strata within dike-swarm segment D1, yielded a relatively young groundmass age of 29.18 ± 0.06 Ma. This age is indistinguishable from that for the Ra Jadero Tuff from Platoro caldera, suggesting that this distal dike may be syncaldera, fed by deeper and more mafic parts of the magma system than for the dacitic ignimbrites. Alternatively, a few groundmass determinations elsewhere in the SRMVF have documented Ar loss from fine-grained intermediate rocks (e.g., lavas that yielded ages younger than the intruding dikes; Lipman et al., 2015, their supplemental file 1). Thus, this Conejos-like dike may have experienced Ar loss, possibly related to a thermal event such as emplacement of a nearby but unexposed dike.


Two groups of light-gray biotite-bearing dacite dikes, with larger and more abundant phenocrysts than in the andesite suite, were distinguished (Lipman, 1974). Most of these are probably postcaldera, because similar compositions are rare in precaldera lavas near Platoro.

One dike group is low-silica dacite (∼61%–64% SiO2), based on comparison with similar postcaldera lavas (dacite of Park Creek). These dikes (previously rhyodacite; Lipman, 1974, 1975) are thicker than andesitic ones (commonly 5 m or more), better exposed, and more continuous along strike. Several dikes at the head of the Alamosa River are traceable for a kilometer or more. Because these dikes lack sanidine, 40Ar/39Ar dating was not attempted.

The other dacite group (quartz latite in the above-cited publications) is more coarsely porphyritic—many with large phenocrysts of sanidine, some with quartz—and has more silicic compositions, mostly in the range 65%–68% SiO2. Some may be low-silica rhyolite. Carlsbad-twinned sanidine megacrysts are as much as 5 cm across in some dikes. These form the largest and most spectacular dikes and plugs in the region; they are commonly >10 m thick (locally as much as 50 m), and some are continuously exposed for several kilometers. Many of the larger sanidine-dacite intrusions are distant from the caldera complex, but in contrast to some andesite dikes, none extends beyond the preserved volcanic rocks. Many of the sanidine-dacite dikes have marginal vitrophyres, a feature rarely observed in more mafic dikes. A few dacite dikes display marginal grooves along contacts that plunge eastward toward the Alamosa River pluton (Fig. 7D).

Except for the dike of late Conejos age at Jim Creek mentioned previously, all other dated intrusions of sanidine-bearing dacite have yielded ages ∼2.5 m.y. younger than the last ignimbrite erupted from Platoro (Chiquito Peak Tuff at 28.8 Ma). New 40Ar/39Ar sanidine ages for four SW- to NW-trending dacite dikes of the radial system west of Platoro caldera cluster between 26.25 ± 0.04 and 26.49 ± 0.06 Ma (Table 2, PP and PD dike segments; Fig. 5). A dacite plug just beyond the southeastern caldera margin at Ojito Creek yielded a similar age, as did a thick dacite sill in Blanco Basin at the southwestern margin of preserved volcanic rocks (Fig. 2). Scattered dikes of petrographically similar sanidine dacite that intrude postcollapse lavas within Platoro caldera are too altered to date by 40Ar/39Ar methods, but are deemed likely to be similar in age. The ages for these silicic intrusions of the Platoro-Dulce swarm are markedly younger than the caldera-filling andesitic lavas, are not associated with preserved lavas, and even postdate the 27.0 Ma Snowshoe Mountain Tuff, the final major ignimbrite eruption from the central caldera complex (Lipman, 2007).

Other shallow proximal intrusions of silicic dacite and rhyolite have yielded younger ages. A dike and plug of silicic dacite within the northwestern margin of the caldera have sanidine ages of 20.44 ± 0.02 and 20.87 ± 0.02 Ma, and a nearby rhyolite intrusion or lava dome has an age of 21.32 ± 0.02 Ma (Table 2). Emplaced temporally between these and earlier radial dacite dikes at 26.5 Ma is the large mineralized dacite dome at Summitville (ca. 23 Ma by K-Ar: Mehnert et al., 1973). Highly altered rocks of the Summitville dome, in turn, are overlain unconformably by erosional remnants of silicic lava (rhyolite of Cropsy Mountain), ca. 20 Ma by early K-Ar analyses (Steven et al., 1967).

Thus, Platoro caldera has a history of postcollapse intermediate to silicic volcanism and dike emplacement, continuing for at least 9 m.y. after its last ignimbrite eruption. The only other calderas in the SRMVF with exposed records of lengthy postcaldera magmatism (all as granitoid plutons) are Questa (7 m.y.), following eruption of the Amalia Tuff at 25.4 Ma concurrently with early development of the Rio Grande rift (Lipman, 1988; Zimmerer and McIntosh, 2012a), and Aetna (4 m.y.), after the Badger Creek Tuff at 34.3 Ma (Zimmerer and McIntosh, 2012b).

Granitoid Intrusions

Granitoid plutons are aligned roughly east-west across Platoro caldera, including the Alamosa River and Jasper intrusions within the caldera, Cat Creek just to the east, and Bear Creek to the west (Fig. 3). Most are fine- to medium-grained equigranular monzonite, some phases containing plagioclase phenocrysts. In places, near-aphanitic phases are best described as andesite or dacite. These high-level plutons have been interpreted as roots of postcollapse volcanoes that were active soon after the ignimbrite eruptions (Lipman, 1975). The east-west zone of intrusions trends anomalously orthogonal to the basement uplifts of the Southern Rocky Mountains (Figs. 1, 3), perhaps related to location along the south margin of the large Bouguer gravity low that defines a subvolcanic batholith below the central San Juan region (Plouff and Pakiser, 1972; Drenth et al., 2012).

Published K-Ar and new 40Ar/39Ar and U-Pb zircon ages from granitoid intrusions at Platoro suggest two main pulses of crystallization at ca. 29 and 27–28 Ma for texturally contrasting phases (Table 2). The main intracaldera intrusions (Alamosa River, Jasper plutons) of relatively coarse equigranular and porphyritic monzonite that intrude the resurgent block have zircon ages at ca. 29 Ma (Gilmer et al., 2018), within analytical uncertainty of the 28.8 Ma 40Ar/39Ar eruption age of the caldera-related Chiquito Peak Tuff. In contrast, the intrusion at Cat Creek just east of Platoro caldera, which is flanked by andesite and dacite lava that have been interpreted as eroded remnants of a large postcaldera volcano (Lipman, 1975), has a U-Pb age of 28.00 ± 0.19 Ma (Gilmer et al., 2018), distinctly younger than the intracaldera granitoid plutons. Another sizable granitoid body, along Bear and Elwood Creeks west of the caldera, appears to be even younger, with ages at 26.61 ± 0.01 Ma (40Ar/39Ar, biotite, sample SRM-33) and 27.06 ± 0.18 Ma (U-Pb, zircon [same sample]). A previously unpublished hornblende K-Ar age of 26.6 ± 1.0 Ma (Table 2, sample DC-2) for a porphyritic phase of the Alamosa River pluton also suggests delayed cooling or crystallization at this large composite intrusion within Platoro caldera. Available chemical analyses (Supplemental File 1 [footnote 1]; Lipman, 1975, his table 10) show that these intrusions tend to be slightly more silicic than the bulk of the caldera-filling lavas (Summitville Andesite) but have overlapping chemical trends.

Dikes of the Dulce swarm tend to form linear ridges because they are more resistant to erosion than the sedimentary wall rocks. Many dike ridges are conspicuous on aerial images such as in Google Earth (Fig. 8), some traceable for 20 km. Some dikes rise 10 m or more above wall rocks of the surrounding terrain, especially in arid areas of northern New Mexico (Fig. 9). Exposures are less apparent on the more vegetated slopes farther north in Colorado, but individual ridge-forming dikes are mappable for comparable distances (Wood et al., 1948). Some sites include multiple dikes separated by only a few meters of intervening sedimentary rock (Fig. 9C).

Outcrop elevations rise from as low as 2000 m at the southern end of the swarm to 2700 m in Colorado. Emplacement depths appear to have been shallow, perhaps only ∼1 km. This depth estimate is based on the projected thickness of volcanic cover and on levels of the late Oligocene land surface, as defined by ignimbrite sheets in the adjacent San Juan Mountains. Farther south in New Mexico, the thickness of volcanic and other mid-Cenozoic rocks above present exposures may have been even less. Similar estimates for emplacement depths are constrained by paleogeomorphic reconstruction of the Chuska sand sea (erg) and by apatite fission-track dating in the San Juan Basin (Cather et al., 2008, 2012).

Dike outcrops tend to be geometrically simple, but some features provide insight into emplacement processes. Widths of individual dikes are typically no more than a few meters; dike walls tend to be planar. Gently plunging fluting or grooves in rare exposures indicate local flow directions (Fig. 10A), and a collaborative study of magnetic anisotropy indicates that subhorizontal magma flow is widely characteristic (Johnson et al., 2016). Development of spherulitic textures and near-glass interstices at some dike margins (Figs. 10B–10C) indicates rapid quenching, while the lengths of dikes (as much as 20 km) require rapid emplacement during lateral magma flow in thin dikes. Selvages of sedimentary wall rocks are thermally darkened and indurated, but for only tens of centimeters adjacent to contacts. Hydrothermal alteration appears largely absent adjacent to dikes, but small-scale brecciation of dike margins at a few sites may have resulted from local flashing of connate fluids. No large-scale brecciation or volcaniclastic textures that might record roots of a vent were recognized, despite apparently shallow emplacement. Many dikes are dense rock, but some interiors contain 5%–10% small vesicles (typically 2–3 mm).

The first chemical analyses for dikes of the Dulce swarm, obtained in the 1960s for the U.S. Geological Survey by E. Landis and by W.J. Hail, documented trachybasaltic to trachyandesitic compositions characterized by variable alkalis and high carbonate and water contents (Supplemental File 3 [footnote 1]; data not previously published). Major-oxide analyses and K-Ar ages (23.5 ± 0.9, 27.2 ± 1.1 Ma) were published by Aldrich et al. (1986) for two Dulce dikes in northern New Mexico, in conjunction with a regional paleostress survey. As part of a broad petrologic study, Gibson et al. (1993) obtained major- and trace-element analyses, K-Ar ages (23.5 ± 0.6, 24.6 ± 0.5 Ma), and the first radiogenic isotopic data (Sr, Nd) for several dikes near the Colorado–New Mexico border. Additional chemical analyses (five samples), Sr and Nd isotopic data (four samples), and two 40Ar/39Ar ages (24.97 ± 0.06, 20.10 ± 0.06 Ma) were reported for Dulce dikes in southern Colorado (Gonzales, 2015; Gonzales and Lake, 2017). In addition to confirming mafic alkalic compositions, these results suggested a prolonged time span (20–26 Ma) for dike emplacement, and the isotopic data indicate near-mantle magma compositions. Sample sites for these studies were mainly along two east-west road transects near the Colorado–New Mexico state line, extending only ∼20 km north-south, and locations of some samples are uncertain.

Mineralogy and Texture

Throughout their 100 km extent, Dulce dikes have a similar megascopic appearance: medium to dark gray, fine grained, and without conspicuous internal boundaries or structures at outcrop scale. Phenocryst assemblages and textures overlap widely along the trend of the dikes, but modest provincial differences are evident. Many dikes are aphyric, especially in dike-swarm segments D2A and D6 (Supplemental File 1 [footnote 1]). About half contain sparse small equant phenocrysts (as much as ∼5%–10%, to 3 mm across) of variably altered clinopyroxene or dark mineral clots, mainly green mica, that are pseudomorphic replacements of mafic phenocrysts. Inference of former olivine is confirmed chemically by high MgO, Ni, and Cr in samples that contain abundant pseudomorphs (Supplemental File 1). The olivine and clinopyroxene phenocrysts identify these dikes as alkalic basalt (trachybasalt), in contrast to high-K lamprophyres characterized by hydrous mafic phenocrysts as reported from the western San Juan Mountains and the adjacent Navajo province (Lake and Farmer, 2015; Gonzales and Lake, 2017). No trachybasaltic dikes of Dulce type contain phenocrystic plagioclase; two plagioclase-phyric dikes in segment D1 of the Dulce swarm (samples 16L-1, 16L-22) are andesite with chemical and age affinities to proximal rocks of the Platoro locus.

Groundmass textures of the Dulce dikes vary from microcrystalline to fine grained. The most crystalline textures, from dike interiors, are dominated by intergrowth of bladed plagioclase (up to 0.5 mm), with finely granular clinopyroxene, opaque oxides, and alteration products filling interstices. Groundmass plagioclase in well-crystallized samples is relatively fresh, but pyroxene is typically variably altered and primary olivine is completely replaced by secondary minerals. Many samples contain groundmass flakes of black mica (brown in thin section), as coarse as 1.5 mm across in the interiors of some dikes (e.g., samples 15L-40A, 15L-56, 16L-8, 16L-9B, 16L-18, 17L-3A, 17L-9A). Electron microprobe analyses for one trachybasalt (sample 16L-9B) show that the groundmass mica and phenocrystic clinopyroxene have higher MgO/FeO ratios (1.5–2.0, 2.3–2.6, respectively) than these phases in San Juan andesites and dacites (∼1.1–1.5, 1.8–2.1; Colucci et al., 1991; Riciputi, 1991), as expectable for a basaltic composition. Because of its black color and modest MgO/FeO ratios, the mica seems appropriately described as biotite rather than phlogopite. A few north-trending dikes of the D3 segment and farther south contain acicular groundmass crystals of brown amphibole (Mg-rich hornblende?) that are typically <0.5 mm long. The hydrous mafic minerals in the groundmass of Dulce samples contrast conspicuously with their anhydrous phenocrysts.

The proximal northeasterly dikes (segment D1) tend to be fine grained and contain clinopyroxene and olivine as phenocrysts and pseudomorphs. In contrast, the northerly north-trending dikes (segment D2A) have well-crystallized bladed plagioclase in the groundmass but lack phenocrysts. Several dike groups farther south have diverse textures: granular to bladed and variably porphyritic, some aphyric but others containing fresh clinopyroxene and/or olivine pseudomorphs. The southernmost sampled dike (segment D6) has a well-crystallized groundmass of finely bladed plagioclase and acicular amphibole, but lacks phenocrysts. Phenocrystic plagioclase or hornblende are absent throughout the Dulce swarm, in contrast to the andesitic dikes and lavas more proximal to Platoro, while the Platoro dikes lack hydrous groundmass minerals.

Dulce samples contain variable amounts of late-formed calcite, filling vesicles and cracks, partially replacing mafic phenocrysts and groundmass plagioclase, and in interstices between magmatic crystals. No megascopically consistent outcrop-scale variations were recognized in carbonate content or degree of groundmass alteration in relation to dike thickness, distance longitudinally along the Dulce swarm, or location in lateral transects. Some dike rocks yield a distinct hydrocarbon aroma when freshly broken, contain pyrite, and yield detectable SO3 and/or Cl by XRF analysis (Supplemental File 1 [footnote 1]), indicating likely introduction of fluids from wall rocks. These dikes were avoided for 40Ar/39Ar dating. A potentially informative topic for further study could be stable-isotopic analysis (O, H, S) of variations among preserved phenocrysts (mainly clinopyroxene) in comparison to hydrous groundmass minerals (mica, amphibole) to evaluate roles of magmatic versus meteoric fluids during dike solidification and deuteric recrystallization.

Age Determinations

Twenty-six (26) new 40Ar/39Ar ages for representative Dulce dikes (Table 2; Fig. 5) along with two published ages (Gonzales, 2015) indicate emplacement during two pulses: mainly 25–27 Ma, with a smaller pulse at 19–21 Ma. The two oldest ages, 27.51 ± 0.14 (sample 15L-36) and 27.32 ± 0.16 Ma (sample 16L-23), are from the NE-trending proximal end of the Dulce swarm (segment D1) that is also distinctive in the subalkaline chemistry of many samples. The youngest age from this segment (sample 17L-2; rather imprecise, at 25.38 ± 0.52 Ma) is from a dike that continues southwest across the segment D1–segment D2B boundary and is compositionally similar to other D2B dikes. Fourteen (14) dates from more southerly segments cluster at 25–26.5 Ma, similar to the sanidine-dacite dikes that radiate more proximally from Platoro caldera. These ages also coincide with those of initial trachybasaltic to tholeiitic lavas within the Rio Grande rift to the east (Lipman and Mehnert, 1975; Thompson et al., 1991; Turner et al., 2019).

In contrast, four samples from the western segment of north-trending dikes (segment D2A), which form a petrologically distinctive suite in their absence of phenocrysts and relatively coarse groundmass textures, have yielded closely grouped ages at 19.1–18.6 Ma. Attempts to date samples from two additional north-trending dikes of similar petrography within this segment (Table 2; samples 16L-66, 17L-5) were unsuccessful. The southward extent of this younger suite of dikes remains incompletely constrained. To the southwest within the adjoining segment D3, a western dike (sample 15L-48) has a relatively young age of 21.48 ± 0.33 Ma, and another dike sample on trend farther north (sample 16L-63), although characterized by a discordant spectrum with low radiogenic yields and low precision, has an integrated age of ca. 21 Ma that suggests that it is likely part of the younger suite. Along with these results, a whole-rock 40Ar/39Ar age of 20.10 ± 0.06 Ma from a dike farther east, along the Montezuma road within segment D3 (sample PS10; Gonzales, 2015), indicates that scattered dikes of the young group continue south nearly to the state line (Fig. 2). This age, which differs from new ages of 25.7 and 25.9 Ma from dikes only a few kilometers to the east and west in southern Colorado (Table 2; samples 15L-40A, 15L-45), documents southward interfingering of the young dike suite with older parts of the swarm along its western flank. An age of 27.28 ± 0.13 Ma for a north-trending dike within segment D2A (sample 16L-64) suggests similar interfingering farther north.

Overall, we consider that clear age groupings and geographic trends have been documented for the Dulce swarm, despite an incomplete sample suite and variable quality of the dated samples. Ages cluster into distinct groups: 21.5–18.6 Ma, 26.5–25.0 Ma, and a few dikes as old as 27.5 Ma at the northeastern end of the swarm. Groundmass-concentrate ages, in particular, yielded the most complex age spectra, commonly characterized by low radiogenic yields related to the variable alteration of the samples. Ages for coarse groundmass mica from dike interiors (four samples) appear especially robust (25.0–26.4 Ma), and similar groundmass-concentrate ages from nearby dikes, combined with the clustering of ages, add confidence to the trends within the Dulce swarm. However, some determinations, especially for bulk groundmass, are likely to be geologically less significant than implied by the high analytical precision because of variable deuteric alteration, secondary carbonate, and other issues of sample quality.

In addition to proximal granitoid intrusions adjacent to Platoro caldera (Cat Creek to the east, Bear Creek to the west), several fine-grained to porphyritic bodies of substantial size (Jackson Mountain, Blanco Basin, V Mountain, Archuleta Mesa) within or adjacent to the Dulce swarm (Fig. 2) have chemical and age affinities to the Dulce and Platoro dikes. These intrusions appear to be large laccoliths or sills. They are more silicic than most Dulce dikes and compositionally broadly similar to the proximal Platoro plutons, but are variably younger.

The Jackson Mountain intrusion of distinctive mafic monzonite (∼58% SiO2) appears laccolithic, although thus far mapped only at 1:250,000 scale (Steven et al., 1974). Two samples yielded identical U-Pb zircon ages at 25.1 ± 0.5 Ma (Gonzales, 2015), broadly similar in age to the younger suite of granitoids and dacite dikes more proximal to Platoro, but differing chemically in containing higher Ti, P, La, and Ce than otherwise similar intrusions in the southeastern San Juan region (Supplemental File 1 [footnote 1]). In these respects, Jackson Mountain has closer affinities to trachyandesite dikes of the Dulce swarm. Distinctively, this intrusion contains megacrysts of variably resorbed K-feldspar to 4 cm in length and rounded quartz pellets, which are in disequilibrium with the relatively mafic bulk composition. These minerals also form clustered clots, as much as 15 cm across, that suggest mingling with mushy granitic magma. Closely grouped ages of zircons and the absence of xenocrystic zircons from Proterozoic basement (Gonzales, 2015) indicate that any mingled granitic component crystallized at about the same time as the overall intrusion.

A laccolith or thick sill of fine-grained equigranular monzonite (61% SiO2), well exposed along the north side of Blanco Basin, is as much as 300 m thick, emplaced within sedimentary strata just below the basal volcanic rocks. No age is available, although a large sill of sanidine dacite across Blanco Basin to the south has a 40Ar/39Ar age of 26.53 ± 0.02 Ma (Table 2, sample 17L-13).

A thick sill (to 150 m), forming Archuleta Mesa, that straddles the Colorado–New Mexico state line (also home to Dulce Base, the alleged joint human-alien collaborative facility: https://en.wikipedia.org/wiki/Dulce_Base), consists of fine-grained monzonite (58% SiO2) with small phenocrysts of plagioclase and clinopyroxene. It is the youngest known granitoid intrusion in the region, having a U-Pb zircon age of 15.5 ± 0.3 Ma (Gonzales, 2015). Another thick sill (150–175 m), intruded between Cretaceous strata and basal Cenozoic volcanic rocks at V Mountain (Larsen and Cross, 1956, p. 104) consists of finely porphyritic monzonite that is petrographically and chemically similar to the Archuleta Mesa sill 20 km to the southwest (Supplemental File 1 [footnote 1]). Permissively, these two localities could be erosional remnants of a single vast sill. V-Mountain lacks phases suitable for 40Ar/39Ar analysis, but future zircon dating could test relations to Archuleta Mesa and other distal Platoro intrusions.

The new chemical analyses of Dulce dikes scatter substantially on variation diagrams, markedly more than for andesitic rocks of the Platoro region such as precaldera Conejos rocks, caldera-related lavas (Summitville Andesite), and granitoid intrusions, or postcaldera rift-related basalts of the Hinsdale Formation (Figs. 6, 1112). To facilitate interpretation, parallel variation-diagram plots are used: (1) to compare Dulce dikes broadly with older continental-arc suites (Conejos Formation, rocks of the Platoro locus) and younger Hinsdale lavas, and (2) to compare geographic segments of the Dulce swarm as a function of distance from the Platoro locus. The scatter among Dulce samples, involving most major oxides and many trace elements, is interpreted as reflecting substantial variations in magmatic values, variably modified by deuteric and other secondary processes. These include alteration of mafic phenocryst and groundmass minerals, deposition of calcite in vesicles and in altered mafic minerals, and probable alkali exchange.

High volatiles in most dikes (LOI: 3%–14% for Dulce samples, 2.5%–8% for Platoro radial dikes; Supplemental File 1 [footnote 1]) also accompany modification from magmatic values. Comparisons between Platoro andesite dikes and related caldera-filling lavas are pertinent; compositions are closely similar for most elements, but LOI values are much higher for the dikes (3%–6%) than for lavas (1%–1.5%; Supplemental File 1; Lipman, 1975, his table 9). Even higher LOI values characterize many sediment-hosted western San Juan and Navajo province dikes (to 14%; Gonzales and Lake, 2017), suggesting that introduction of external volatiles is especially likely during cooling and deuteric recrystallization of dikes intruded into permeable strata. In contrast, primary basalt magma appears likely to contain no more than a few percent volatiles; for example, alkalic-basalt and tephrite glass quenched at deep-ocean pressure has only ∼1% H2O (Coombs et al., 2006).

Despite the deuteric and other alteration, primary magmatic trends are identified for the Platoro-Dulce swarm by sizable variations in elements inferred to have been relatively immobile during alteration (e.g., TiO2, P2O5, Zr, rare earth elements [REEs]). Notably, the compositions inferred to record magmatic processes become more alkaline with distance from Platoro caldera, as illustrated by averaged compositions for midpoints of progressively more southerly dike segments (Fig. 13). As a whole, the Dulce swarm forms a continuum on these plots with the Platoro radial dikes, with elements such as TiO2, P2O5, Zr, and La increasing with distance from the caldera. Differences are especially striking between dike-swarm segment averages for the NE-trending Dulce dikes (segments D1 to D2B), and between the northernmost and southernmost north-trending dikes (segments D2A to D6). In contrast, compositions overlap among the north-trending dike segments at intermediate distances (segments D3 to D5).

Major Elements

Dulce dikes scatter on an alkali-silica diagram but dominantly are trachybasalts and basaltic trachyandesites (Fig. 6). The Dulce swarm is not a K-rich lamprophyric suite (K2O/Na2O >1), contrary to interpretation by Gonzales and Lake (2017, p. 206). Only 10 of the 75 analyzed Dulce dikes have K2O/Na2O >1, these due more to erratically low Na than to exceptionally high K. Average alkali ratios for dike segments are only 0.2–0.7, tending to increase southward.

Alkali ratios vary widely among samples, many involving high Na2O and low K2O, but total alkalis appear to retain near-magmatic values, as indicated by correlation with other elements and with location. Most samples that have low total alkalis (<5%), plotting within the subalkaline basalt field, also have low values of elements such as Ti, P, Zr, and La. These are from the northernmost dikes (segments D1, D2A) nearest to the San Juan Mountain front and merging geometrically with the dikes radial to Platoro caldera. The highest alkali values at low silica are from the southern parts of the Dulce swarm (segments D4 to D6).

Some Dulce dikes with anomalous alkali contents, accompanied by highly variable Na2O/K2O and Rb/K2O ratios (0.1–46, 11–61 respectively) among samples that are otherwise chemically similar (Supplemental File 1 [footnote 1]), are interpreted as reflecting alkali mobility during deuteric recrystallization. Many samples with anomalous alkali ratios also have high CaO and LOI indicative of abundant calcite, fine-grained groundmass, and otherwise extreme bulk compositions. These include many of the low-SiO2 samples (16L-12, 16L-16, 16L-17, 17L-12) that plot deep in the tephrite field (Fig. 6). Unexpectedly, the apparent alkali exchange in some Dulce samples involves increased Na relative to K, in contrast to the “potassium metasomatism” documented in rhyolites (Ratté and Steven, 1967; Chapin and Lindley, 1986).

In comparison, lavas, proximal dikes, and granitoid intrusions of the Platoro locus are more silicic than most Dulce samples but define alkali-silica trends that project into lower-alkali parts of the Dulce data. Precaldera Conejos lavas and dikes (including those from Summer Coon volcano) and younger rift-related basaltic lavas of the Hinsdale Formation also scatter considerably but follow trends similar to those of the Platoro rocks, extending into the Dulce data (Fig. 11). Many rift-related Hinsdale samples are more alkalic than continental-arc rocks of the Platoro locus, overlapping more fully with the Dulce swarm.

Values of TiO2 and P2O5, which may be less subject to deuteric and secondary alteration, vary substantially for Dulce samples of similar SiO2 contents, many samples having values higher than trends projected from analyses of precaldera Conejos lavas, Platoro intrusions, or the rift-related basaltic lavas of the Hinsdale Formation (Fig. 11). Although alteration effects have likely modified contents of other elements, substantial portions of the scatter in these major-element compositions on silica variation diagrams appear to be geographically related, with dike-swarm segment–averaged values increasing with distance from Platoro caldera (Fig. 13). Low contents of TiO2 and P2O5 are typical for the northernmost north- and NE-trending trachybasalt Dulce dikes closest to Platoro that are also low in total alkalis (segments D1, D2A), while the highest values characterize the long southernmost dike in New Mexico (segment D6). Dikes at intermediate distances in southern Colorado and northern New Mexico (segments D2B to D4) display overlapping compositional scatter for these elements. Dikes containing groundmass amphibole become more common southward from the Colorado–New Mexico border, concurrently with bulk-composition increases in TiO2 and P2O5. Amphibole-bearing dikes differ only modestly in chemistry from dikes in these segments that contain groundmass mica, mainly having higher average values of TiO2, Sr, and Ba.

Variation in SiO2-FeOTOT and SiO2-MgO among Dulce samples (Supplemental File 1, Figs. S1A, S1B [footnote 1]) plot near the semi-linear magmatic trends defined by Hinsdale trachybasalts, especially allowing for the variable former presence of olivine. Values of MgO also correlate closely with those of Ni and Cr, indicating substantial preservation of magmatic compositions despite alteration of olivine. Variations in Al2O3 are more complex (Supplemental File 1, Fig. S1C), the Dulce samples broadly defining a diffuse arcuate trend that increases with SiO2, as is typical for alkalic-basalt suites elsewhere in the region (e.g., Gibson et al., 1993; Leat et al., 1988; Lake and Farmer, 2015; Gonzales and Lake, 2017). High values of Al2O3 at low SiO2 contents, which are especially conspicuous for the northern Dulce dikes (segments D1 and D2A), are interpreted as geographic variations, with lower values to the south. Similarly high Al2O3 also characterizes Platoro postcaldera lavas and dikes, in comparison to basaltic lavas of the Hinsdale Formation that define a nearly linear array at values similar to many of the more southerly Dulce dikes. In addition, Al2O3 in Dulce samples tends to correlate inversely with K2O, TiO2, and P2O5. The high-Al2O3 end of the array is anchored by the northern dikes (segments D1 and D2A), consistent with their relatively low K2O, TiO2, and P2O5 values, along with low trace elements such as Zr, Nb, and the light REEs (LREEs) that suggest affinities toward subalkaline (tholeiitic) compositions. Most of these samples plot low in, or below, the trachybasalt field on an alkali-silica diagram (Fig. 6).

As anticipated by abundant calcite in most Dulce dikes, CaO scatters widely on SiO2 variation diagrams (Supplemental File 4, Figs. S4-1A–S4-1B [footnote 1]), and many samples have substantially higher CaO than trachybasaltic lavas in the San Juan region (Hinsdale Formation) or the most mafic dikes from the precaldera Summer Coon volcano. Alternatively, a crude positive correlation of CaO with LOI in Dulce samples is consistent with presence of much of this element in the secondary carbonate that is abundant in these dikes (Supplemental File 4, Fig. S4-1D). Previously unpublished analyses of 15 Dulce dikes, from samples collected by W. Hail, Jr., and E. Landis (U.S. Geological Survey) in the late 1960s, directly determined as much as 7.8% CO2, with eight samples containing >3% CO2 (Supplemental File 3). The weight percent of calcite in these rocks would be about double that of the CO2, thus indicating as much as 15% calcite in some samples.

A related issue for Dulce dikes concerns the proportion of introduced versus recycled magmatic calcium in the carbonate. Early during the present study, some Dulce samples were leached in dilute HCl (15 mol%) prior to analysis (Supplemental File 4 [footnote 1]) in a failed attempt to optimize comparisons with magmatic compositions. The acid leaching lowered CaO contents by as much as 5 wt% (40% of the total for sample 15L-39). Many of the leached trachybasalts yielded anomalously low CaO values (some <5%), only about two-thirds that expected from projected magmatic trends. Thus, substantial calcium in the carbonate appears to have been derived from magmatic minerals (mainly clinopyroxene?), rather than from external sources. An inverse correlation of LOI with SiO2 in Dulce samples (Supplemental File 4, Fig. S4-1C) also supports interpretation that much of the CaO in secondary carbonate came from alteration of clinopyroxene, a phase which would have been abundant in more mafic dikes. Correlation of high CaO and CO2 with low SiO2 contents in these analyses (Supplemental File 4, Fig. S4-1B) further suggests that much calcium in the carbonate was derived from alteration of pyroxene.

In addition, many leached samples have low values for other elements of interest such as P and the LREE (La, Ce, Nd), in comparison to unleached splits (Supplemental File 4 [footnote 1]). Accordingly, all analyses in Supplemental File 1 and plotted in text figures are for unleached samples. Although these analyses have been recalculated to reported totals with the LOI contents excluded, some samples with high CaO presumably have a component of introduced calcium.

Trace Elements

Trace elements also define compositional variations among the Dulce swarm that in part correlate with distance from Platoro caldera (Fig. 12). As for the major oxides, Dulce samples scatter more on trace-element diagrams than do rocks of the Platoro locus or Hinsdale lavas. Elements such as Zr and La are markedly higher on SiO2 diagrams than for arc-related rocks of the Platoro locus, with Hinsdale lavas tending to plot at more overlapping values. Variations in some elements, such as Ni and Cr, appear to reflect magmatic processes such as presence of olivine as already noted; nonsystematic large variations in elements such as Ba and Sr are probably mainly related to alteration. Despite such data scatter, dike-swarm segment–averaged values for elements such as Zr and the LREEs increase southward (Fig. 13), consistent with trends for TiO2 and P2O5 that suggest increasingly alkalic compositions with distance from the Platoro locus. Segment D6 contains only a single long dike; three sites sampled along 7 km of dike length have similar compositions that are the most alkalic of any segment, based on high Ti, P, Zr, and LREE content.

Chondrite-normalized REEs also show sizable changes southward in the Dulce swarm. Normalized light-heavy fractionation increases (Fig. 14), with La/YbN ratios increasing from 9 for Dulce segment D1 southward to 29 in segment D6. In comparison, proximal caldera-related lavas, dikes, and granitoid plutons at Platoro are tightly grouped, with La/YbN ratios of 9–12. Only the 25 Ma Jackson Mountain pluton, far west of Platoro, has a La/YbN (31) comparable to that of the more southerly Dulce dikes. In contrast, 19–21 Ma Hinsdale lavas close to Platoro caldera (at Red Mountain, Green Ridge) have variable REE compositions, including a greater range in La (23–70 ppm, versus 28–63 ppm for the Dulce swarm) and nearly as large REE fractionations (La/YbN, 9–23).

Additional Areal Variations

While dike-swarm segment–averaged dike compositions change southward along the Dulce swarm in multiple major and trace elements, a few dikes are anomalous relative to the dominant compositions within a segment. Some outliers are likely related to the arbitrary segment boundaries that divide the continuum of the dike swarm. In addition, the great length of some dikes is likely to have produced interfingering of contrasting compositions along traverses orthogonal to the strike of the swarm. For example, sample 16L-52 from segment D1 has atypically high values of TiO2, P2O5, Ba, Zr, and the LREEs. In these respects, this site is more similar to dike compositions in segment D2B, and indeed the 16L-52 site has been mapped as the northeastern end of a dike that continues into this segment, where two additional samples (15L-39, 17L-2) have similar compositions (Supplemental File 1 [footnote 1]). Analogously within segment D2B, two of the 17 analyses (16L-26, 17L-7) have some characteristics (low P, Ti, Zr, LREEs) more closely resembling the dominant compositions of dikes in segment D1. Dikes in other segments, especially at mid distances (segments D3 to D5), display substantial variability but less extreme outlier compositions. Some variations may be due to incorrectly distinguishing the extent of individual dikes or to local fractionation effects. For example, the trachyandesite site 16L-66 is mapped as part of a long dike within segment D2A (Wood et al., 1948); samples farther north at three additional sites (16L-64, 16L-65, 17L-8) yielded more mafic compositions with markedly lower Zr and LREE contents.

Some transverse variations are also present within individual segments. Samples are listed in Supplemental File 1 (footnote 1) in approximate sequence from west to east, orthogonal to dike trends for each segment. Within segment D1, several dikes on the southeastern side of the swarm are higher in Ti and Zr than those to the northwest, and are transitional to compositions in segment D2B; these dikes also have younger isotopic ages (24.5–25.9 Ma) than dikes to the northwest, which are the oldest (27.5 Ma) from anywhere in the Dulce swarm. In segment D2A, the central dikes are higher in Si and lower in Ti, P, Zr, and LREEs than flanking dikes. Dikes in segment D2B have a large range in SiO2 (45%–55%), tending to become more silicic to the southeast. Analyses from segments D3 and D4 also have sizable ranges in SiO2 and other elements, but lack obvious transverse trends. Dikes in segment D5 are especially mafic; seven of the eight analyses have <50% SiO2 calculated volatile-free, and silica content decreases eastward.

Isotopic Data

Sparse published isotope data (εNd, Sr*I [initial Sr isotope ratio]; Gibson et al., 1993; Gonzales and Lake, 2017) show that the Dulce dikes are generally similar to other middle Cenozoic igneous rocks in the Southern Rocky Mountains, all indicative of large mantle-derived components (Fig. 15). Although including more mafic compositions, the Dulce dikes are broadly aligned in isotopic compositions with early-rift lavas of the Hinsdale Formation southeast of Platoro. Precaldera intermediate-composition lavas of the Conejos Formation are more silicic and more radiogenic in εNd (Colucci et al., 1991), probably recording larger lower-crustal components. Mafic alkalic rocks of the Navajo volcanic field, farther west within Colorado, are more mantle like (highest in εNd). The Dulce dikes are thus transitional between Hinsdale lavas and the lamprophyric Navajo rocks. Unfortunately, isotopic data for the Dulce swarm are available only for dikes close to the Colorado–New Mexico line (segments D3, D4); no isotopic data exist for subalkaline NE-trending dikes farther northeast in Colorado, for more-alkalic dikes farther south in New Mexico, or for any andesite dikes of Platoro type. Somewhat surprisingly, the single isotopic analysis (sample DU-10; Gonzales, 2015, for a Miocene (20 Ma) dike is low in εNd, more similar to values for Conejos lavas than for the Hinsdale basalts.

The Platoro-Dulce magmatic suite provides an exceptional space-time-compositional record for the volcanic-intrusive-tectonic evolution of an unusually long-lived large igneous system associated with recurrent ignimbrite eruptions during the transition from continental-arc to extensional tectonics (Fig. 16). Large-volume mantle-derived mafic inputs generated only intermediate-composition to silicic eruptions in the Platoro area during the arc stage (ca. 32–27 Ma) due to interaction with continental crust at this intense magmatic locus (Lipman et al., 1978; Dungan et al., 1989). Basaltic magmas were unable to penetrate the upper crust until input to the Platoro locus waned, starting at ca. 26 Ma, and regional extension overprinted the arc setting.

Growth of the Platoro Locus

As elsewhere in Oligocene SRMVF, initial development of an intense magmatic locus at Platoro was recorded by clustered growth of several intermediate-composition volcanoes, dominantly composed of diverse high-K andesitic lavas of continental-arc type (Conejos Formation), preparatory to the multistage ignimbrite flare-up and associated caldera collapses. Dikes of hornblende and pyroxene andesite, which radiated as much as 30 km outward from the core of Platoro magmatic locus during this early stage, have yielded ages of 32–31 Ma, just prior to the flare-up. The radial distribution of these large dikes, emplaced prior to the initial caldera-related ignimbrite, is interpreted to record stresses generated by initial upward and outward inflation associated with a large-scale upper-crustal locus of magma accumulation and growth of the clustered precaldera volcanoes at Platoro. Such precursor dike emplacement, while rarely documented elsewhere at large caldera systems, is consistent with some recent models for the origin of caldera-forming magma bodies (Jellinek and DePaolo, 2003; Karlstrom et al., 2012).

Recurrent large eruptions of crystal-rich dacite ignimbrite in rapid succession, with a collective volume >3000 km3, repeated concurrent caldera collapse, and associated continued outpouring of intermediate-composition lavas, define the peak stage of magmatism at the Platoro locus from 30.3 to 28.8 Ma. The first major ignimbrite (Black Mountain Tuff, at 30.3 Ma) and inferred collapse of an associated caldera record initial full development of an integrated upper-crustal reservoir, within which voluminous eruptible magma accumulated. All of the large Platoro ignimbrites are crystal-rich dacite, containing hydrous mafic phenocrysts (biotite, sparse hornblende), that are compositionally similar to ignimbrites of continental-arc affinity erupted widely in the SRMVF, elsewhere in the western USA during the middle Cenozoic, and farther north and south along the American Cordillera (Best et al., 2016). Consistent with their arc-type mineralogy and geochemistry, such ignimbrites were able to spread widely and with relatively uniform thickness across areas where later extension generated horsts and grabens in the Southern Rocky Mountains and elsewhere.

Monzonitic plutons (Alamosa River, Jasper), which intruded caldera-filling andesitic lavas and are interpreted to have formed in the cores of postcaldera volcanoes at Platoro, and a porphyry intrusion into the resurgent caldera block yielded U-Pb zircon ages analytically indistinguishable from that of the last-erupted ignimbrite, the 28.8 Ma Chiquito Peak Tuff (Table 2; Gilmer et al., 2018). These plutons are interpreted to be shallow portions of a larger and vertically extensive composite batholith, underlying Platoro caldera, that consisted dominantly of residual near-solidus crystal mush that accumulated during the prolonged duration of large-volume andesitic-dacitic volcanism and associated intrusions of continental-arc affinity. No true basalt reached the surface during this peak period of magmatism at Platoro.

Available age data indicate that the generation, rise, and crystallization of shallow intermediate-composition magma continued episodically at declining rates for several million years after the last ignimbrite eruption at Platoro, probably recording recurrent mafic recharge and enlargement of the underlying subvolcanic batholith. A porphyritic phase of the Alamosa River pluton, the eastern Cat Creek pluton, and the western Bear Creek granitoid intrusion have all yielded 40Ar/39Ar and U-Pb zircon ages of 28–26 Ma, modestly younger than peak caldera magmatism.

Injection of andesite and dacite dikes radiating from a locus approximately coincident with the Alamosa River pluton, which had commenced during construction of precaldera Conejos volcanoes, continued after cessation of ignimbrite eruptions. Radial postcaldera andesite dikes are well documented proximally along the southwestern margin of Platoro caldera; their distal extent remains poorly constrained. Thick sanidine-phyric dikes of silicic dacite, comparable in composition to the Platoro ignimbrites and traceable as much as 20 km to the northwest and southwest from the Platoro locus, have yielded closely clustered ages at 26.5 Ma. These appear to record a discrete late pulse of evolved melt from a long-lived composite body of mushy magma that remained at the Platoro locus after cessation of ignimbrite eruptions. The radial distribution of the dikes is interpreted to reflect the continued influence of stresses generated by intermittent upward and outward inflation of the sub-Platoro batholith, >2 m.y. after the final caldera-related ignimbrite.

Initial Rift-Basalt and Flanking Magmatism

Broadly concurrently with emplacement of the latest granitoids east and west of Platoro and intrusion of radial dacite dikes, basaltic lavas were first erupted in the southeastern San Juan region at ca. 26 Ma. These are now preserved as erosional remnants of the Hinsdale Formation (Lipman and Mehnert, 1975; Thompson et al., 1991; Turner et al., 2019) and as dikes of the Dulce swarm farther southwest in Colorado and northern New Mexico. The early Hinsdale basalts, which include both tholeiitic and trachybasaltic compositions, appear to have erupted as far-traveled lavas, mainly from vents within the present-day Rio Grande rift (Turner et al., 2019); no basaltic lavas erupted proximal to Platoro caldera have been recognized prior to ca. 21 Ma.

Initial intrusion of the Dulce swarm may have commenced as early as 27.5 Ma, as suggested by ages from several subalkaline dikes in dike-swarm segment D1 and trachybasalt dikes within segment D2A, but the majority of dikes have ages of 26.5–25 Ma (Figs. 5, 16). Thus, the bulk of the Dulce swarm is closely concordant with inception of tholeiitic and trachybasaltic lavas of the Hinsdale Formation erupted from vents far to the east, within the Rio Grande rift. This concordance in age and magma compositions, along north-south trends on both the west flank of the Laramide-age uplifts of the Southern Rocky Mountains and axially along the broad uplift crest, is interpreted to mark initial impacts of the transition from magmatic arc to regional crustal extension. Deflection to NE trends along northern segments of the Dulce swarm (segments D1, D2B) and geometric continuity with the radial dacite dikes of similar age shows that inflation at the Platoro locus remained an active influence on a regional stress field that was increasingly affected by westward-directed extension. Variations in dike compositions, becoming generally more alkalic southward, are inferred to record decreasing proportions of mantle melting and/or decreasing interaction with lower crust in response to distance from the Platoro locus. Subhorizontal magma flow for distances of 10 km or more within individual dikes may account for intermingling of subtly differing compositional types when sampled transversely across the Dulce swarm.

Emplacement of the satellitic Jackson Mountain granitoid body at 25 Ma was concurrent with that of the youngest dikes of the main Dulce swarm, although ∼10 km to the northwest. This intrusion is also spatially aligned with the east-west zone of granitoid plutons within and proximal to Platoro caldera (Fig. 3), but it is distinct in petrography and chemistry, having closer chemical affinities with trachyandesite dikes of the Dulce swarm. Presence of several other large laccoliths and sills farther south along the western mountain front at Blanco Basin and V Mountain, emplaced along the contact zone between Cretaceous sedimentary strata and overlying basal volcanic deposits, suggests that waning of magmatic intensity at the Platoro locus and increasing impacts of regional extension may have led to a structural environment conducive to shallow-crustal magma injection laterally outward from the long-lived Platoro locus. Regionally, a broad east-west zone of magmatism was active at 27–25 Ma (Figs. 13), discontinuously from the Navajo province on the Colorado Plateau (Gonzales and Lake, 2017) eastward through Jackson Mountain, Platoro caldera, and Cat Creek volcano and its central intrusion, to the Questa-Latir locus east of the Rio Grande rift (Lipman, 1988; Zimmerer and McIntosh, 2012a), and on to the High Plains at Spanish Peaks (Penn and Lindsey, 1996).

Other than the large dacite lava dome at Summitville at 23 Ma, little igneous activity of any composition, intrusive or extrusive, has been documented in the southeastern San Juan region between 25 and 21 Ma. Then, mafic and silicic activity at Platoro, Hinsdale basaltic lavas, and late Dulce dikes define a final regional episode. Magma as mafic as basalt initially intruded shallow crust or erupted proximally in the Platoro area beginning at ca. 20.5 Ma, concurrently with a late pulse of silicic magmatism at 21.3–20.4 Ma. This pulse included eruption of rhyolite as flows and domes and intrusion of dikes of silicic crystal-rich dacite along the northern margin of the caldera, forming a bimodal suite that contrasts with the prior caldera-related magmatism of dominantly intermediate compositions. Nearly concurrently, a late group of north-south dikes was emplaced along the western margin of the northern Dulce swarm (segments D2A, D4) at 21.4–18.6 Ma. The north-south–striking dikes of segment D2A, in contrast to NE trends of older dikes at the northern end of the Dulce swarm (segments D1, D2B), and the changing proportions of magma compositions that transition to a more bimodal assemblage, are interpreted to record waning influence of the Platoro magmatic locus on the regional stress field, along with increasing impacts of extensional tectonics.

The lengthy magmatic history of the Platoro locus (from >32 to 19 Ma), involving precaldera Conejos lavas, voluminous ignimbrites, postcaldera lavas and intrusion, and later extension-related Dulce dikes, constitutes an exceptionally long-lived and recurrently active magmatic system that involved the entire lithospheric section and probably deeper. Prolonged survival of an upper-crustal body of mushy near-solidus magma beneath Platoro, which could be recurrently reactivated by intermittent arrival of mafic mantle melts to generate surface eruptions and additional intrusions at shallow depth, seems likely to exemplify general processes in formation of large-volume continental silicic systems (Lipman and Bachmann, 2015; Best et al., 2016; Cashman, et al., 2017).

Geometry of Rift Extension, Fault Trends, and Regional Asymmetries

Age relations of the Dulce swarm and the inception of the Rio Grande rift to regional NW-trending normal faults that are the dominant Cenozoic structures along the eastern margin of the San Juan Basin in southern Colorado and northern New Mexico (Fig. 17) remains incompletely understood. The dike swarm is nearly orthogonal to these NW-trending faults, especially along its northeastern segments (D1, D2B), recording contrasting directions of extensional strain. Existing geologic maps (Wood et al., 1948; Dane, 1948) are insufficiently detailed to indicate reliably whether any Dulce dikes are cut by faults of this trend, but aligned faults with similar trends in the mountains to the east displace rocks similar in age to or younger than the dike swarm. In the eastern Tusas Mountains just south of the Colorado–New Mexico state line, basaltic lavas of the Hinsdale Formation at least as young as 25.6 Ma (Turner et al., 2019, their figure 3-23) are cut by a parallel system of NW-trending faults. Another large fault system, which trends NW from Platoro to Wolf Creek Pass in Colorado (Fig. 17), displaces the 23 Ma dacite dome at Summitville and probably the crystal-rich dacite-rhyolite of North Mountain, for which one dated phase is as young as 20.7 Ma (sample 11L-19). And to the east, in the Questa-Latir area, many NW-trending dikes are younger than the 23 Ma Rio Hondo pluton (Lipman, 1988), with at least one as young as 16.6 Ma (Zimmerer and McIntosh, 2012a). Within the eastern San Juan Basin near the state line, a NW-trending fault displaces the 15.5 Ma Archuleta Mesa sill by ∼100 m, yet 20 Ma and 25 Ma dikes nearby to the east and west trend due north. These relations suggest that episodes of SW-directed extension have alternated with periods of west-directed strain during early rifting in the Southern Rocky Mountains. Additional detailed mapping could improve understanding of the temporal and geometric relations between emplacement of Dulce dikes and regional faulting.

The location of the lengthy Dulce swarm, intruded into little-deformed Mesozoic strata along the eastern margin of the Colorado Plateau, across the Continental Divide from the present-day Rio Grande rift, is unique in the region and somewhat puzzling. Perhaps initial extension along Dulce and Rio Grande axes was influenced by geometry of the broad Laramide-age uplift in the Southern Rocky Mountains (Fig. 18). This uplift is well defined by Proterozoic basement rocks exposed along the Sangre de Cristo Range on the east and the Sawatch Range and Tusas Mountains to the west of the axial rift basins (San Luis and upper Arkansas River valleys; Fig. 1). A broadly monoclinal crustal hinge along the western margin of the regional Laramide uplift, as defined in part by the Paleogene depositional axis of the San Juan Basin (Cather et al., 2018), may have become a locus of weak extension, influencing intrusion of the Dulce dikes. In contrast, at least the main initial Hinsdale vents (at ca. 26 Ma: Turner et al., 2019) were along the crest of Laramide uplift, which became the dominant axis of extension for present-day rift geometry.

An additional geometric complexity of the magmato-tectonic transition is the more intense radial diking on the western side of Platoro caldera than to the east. Scattered dikes of andesite and dacite and several small laccoliths radiate outward for a few kilometers from the eastern Cat Creek pluton (Lipman, 1974), but these are modest in number and extent compared to the western dikes that project outward from the locus near the Alamosa River pluton. Basaltic dikes, more comparable to the Dulce swarm, may underlie rift fill axially in the San Luis Valley, but despite less rugged topographic relief, the Cenozoic volcanic rocks east of Platoro are sufficiently exposed to preclude any connecting dike system on a scale comparable to those west of Platoro. More likely, the contrasting scales of dike emplacement reflect influence of the regional tectonic framework: the west flank of the broad Laramide-age uplift may have formed an eastern buttress and barrier to dike emplacement. At the Platoro locus situated roughly along the hinge zone along this flank of the Laramide uplift (Fig. 18), dikes propagated preferentially westward toward the Colorado Plateau. Such a structural interpretation of dike geometry at Platoro would be analogous to that previously advanced for counterpart asymmetry at the Spanish Peaks intrusive cluster (Fig. 18), where central granitoid intrusions and the adjacent asymmetrical dike system were localized along the eastern flank of the Rocky Mountains uplift, mainly from 26 to 21 Ma (Odé, 1957; Penn and Lindsey, 1996).

Another regional asymmetry in the SRMVF is the more areally extensive and voluminous mid-Cenozoic volcanism and intrusion on the western flanks of the Laramide uplift than to the east (Fig. 1). Western centers include the Mount Princeton batholith and ignimbrite calderas along the present-day Sawatch Range, continuing southward into the San Juan region and to the northern Tusas Mountains. Eruptive centers to the east, including the Latir-Questa locus (the only eastern ignimbrite source) and intermediate-composition volcanoes to the north (Rosita, 39 Mile, Buffalo Butte), are notably smaller in volume and extent. The eastern volcanic rocks, especially in the 39 Mile area and farther east, consist largely of ignimbrites from calderas in the Sawatch Range to the west.

No arc-type caldera of the SRMVF other than Platoro, including those closer in age to inception of regional extension at ca. 26 Ma, is closely associated with a comparably developed radial dike system or other prolonged postcaldera mafic volcanism. The numerous dikes associated with granitoid plutons south of the early-rift Questa caldera are rhyolitic, trend NW, and are geometrically related to regional extension, largely unaffected by the adjacent caldera (Lipman, 1988). Erosional remnants of Miocene trachybasalt lavas and rare mafic dikes farther north and west in the San Juan region (Lipman and Mehnert, 1975; Gonzales, 2015) likely represent less-focused responses to late extension, without clear age or geometric links to nearby calderas.

Available models for caldera resurgence do not predict dike emplacement or other deformation beyond the area of caldera subsidence (Marsh, 1984; Roche et al., 2000; Galetto et al., 2017). However, these models do not include possible effects from upper-crustal magma bodies that may be substantially larger than the associated caldera, or possible linkages between magma emplacement and regional tectonic stress. Perhaps the location of the Platoro system, along or just beyond the south margin of the large gravity low that is interpreted to record the presence of a subvolcanic batholith (Plouff and Pakiser, 1972; Drenth et al., 2012), permitted relatively efficient interconnection with the changing regional stress field, in contrast to the calderas more centrally above a vertically and laterally extensive batholith. Whatever the ultimate controls on the voluminous magmatism at the Platoro locus, it provides an exceptional site for evaluation of processes at a long-lived (in this case, >10 m.y.) crustal system of batholithic scale.

The implied link between tectonic and magmatic regimes, and associated impacts of abrupt shifts over geologically short time scales, has implications for continental structural evolution and long-term hazards associated with distal caldera magmatism. Rapid emplacement of dikes comparable in scale to the Platoro-Dulce swarm, in conjunction with late evolution of a caldera-related batholith, could have generated extension-related earthquakes and triggered dispersed volcanism. Such events would constitute previously little-recognized magmato-tectonic hazards that could occur near active and dormant calderas in the western USA and elsewhere. For example, emplacement of long-traveled mafic dikes southward from Valles caldera in New Mexico, along extensional faults of the Rio Grande rift and comparable in scale to the 125 km Dulce swarm, could reach Albuquerque (New Mexico) or farther south. Such dikes would be too young to have been exposed by erosion, but perhaps the basaltic vents that are aligned north-south on San Felipe Mesa and farther south at the Albuquerque volcanoes (a basaltic fissure eruption as young as 220 ka; Peate et al., 1996; Singer et al., 2008) constitute a younger counterpart to the Dulce swarm. These vents extend 80 km or more from Valles caldera, which last erupted a large ignimbrite at 1.26 Ma but has continued to be active in smaller volumes as recently as 68 ka (Wolff et al., 2011; Zimmerer et al., 2016). Zircons from the 68 ka eruption yield a spectrum of U-Th ages back to secular equilibrium (ca. 350 ka), documenting a long-lived system beneath the caldera from which deep mafic magma could have intermittently propagated as dikes.

Similarly, north of Long Valley caldera, which erupted a major ignimbrite at 760 ka, the Inyo-Mono chain of rhyolite domes has been active as recently as 650 yr B.P., in association with northward-propagating dikes (Hildreth, 2004). Any basaltic dikes intruded along this trend for distances comparable to that of the Dulce swarm could come close to the Nevada state capital at Carson City. At the smaller Crater Lake caldera in Oregon, proximal basaltic vents form radial trends for at least 12 km from rims of this 7700 yr B.P. ignimbrite center (Bacon and Lanphere, 2006), while more distal mafic vents are parallel to regional extensional faults that trend north-south. On older, larger scales, could the Miocene dikes associated with basaltic eruptions on the Columbia Plateau in Oregon-Washington and mafic vents along the Northern Nevada rift zone to the south (Camp et al., 2015) constitute distal mafic components, broadly analogous to the Dulce dike swarm, that project at least 200 km north and south from concurrent loci of intense ignimbrite-caldera magmatism at 16–14 Ma during inception of the Yellowstone hot spot? Regardless of whether these calderas are genetically linked to smaller eruptions by long dikes, the Platoro-Dulce dikes demonstrate magma emplacement at great distances from a source caldera. Such relationships suggest that the regional extent of magmato-tectonic hazards associated with calderas may be underestimated.

The sweeping continuity of dikes that radiate westward from the Platoro caldera locus to merge with the arcuate Dulce swarm, in conjunction with overlapping ages and partial convergence in chemical arrays, suggest that they constitute a long-lived integrated dike system of regional extent. Dikes of the Dulce swarm have a broad range of trachybasaltic to trachyandesitic compositions (46%–56% SiO2) and were emplaced during two pulses (mainly 26.5–25.0 Ma, with a smaller pulse at 21.4–18.7 Ma). Proximal dikes that radiate westward from Platoro caldera to merge with the Dulce swarm are more silicic (andesite-dacite; mostly 54%–67% SiO2) and range widely in age, from precaldera (31.2–30.4 Ma) to as young as 20.3 Ma, but many or all of the dacite dikes were emplaced concurrently with the main Dulce swarm.

Dikes of the Platoro-Dulce swarm have large variations in textures, phenocrysts, and elemental compositions that correlate broadly with distance from Platoro caldera, becoming more mafic and alkalic outward. The dike system is associated with protracted postcaldera volcanic and large-volume intrusive activity at Platoro, recording interactions between caldera-related and regional stress fields and appearing to document varying magma sources in cratonic lithosphere as a function of distance from the caldera locus. Proximal radial dikes are inferred to record the effects of magma inflation and uplift associated with prolonged assembly and solidification of a composite batholith beneath Platoro caldera. Concurrently with late Platoro magmatism, linear dikes of the Dulce swarm were emplaced during weak extension along the eastern margin of the Colorado Plateau, satellitic to the early Rio Grande rift and associated basaltic volcanism farther east. More broadly, these features are interpreted to record a regional transition from arc- to rift-related magmatism in a continental setting, with implications for potential hazards near active calderas elsewhere.

We especially thank Kyle Tator and Todd Asmera for arranging access within the Jicarilla Apache Nation Reservation. William McIntosh, Kyle Anderson, and Amy Gilmer assisted with some fieldwork. Also much appreciated are diverse data on southeast San Juan igneous rocks provided by other researchers. Previously unpublished major-oxide chemical analyses of Dulce samples (Supplemental File 3 [footnote 1]), shared with Lipman more than 40 years ago by W.J. Hail, Jr., and E. Landis of the U.S. Geological Survey, constituted an early stimulus for the present study. Amy Gilmer determined U-Pb zircon ages on five samples of granitoid intrusions we had collected in the Platoro area (Gilmer et al., 2018) and also provided electron-microprobe analyses of mafic minerals in a Dulce dike. Samuel Johnson shared preliminary results of his magnetic-anisotropy studies that help constrain dike-emplacement processes. Amy Gilmer, Kathryn Watts, David Gonzales, an anonymous reviewer, and Geosphere associate editor Valerio Acocella provided helpful comments on the manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

1Supplemental Files. Five files of chemical and geochronological analytical data on which the discussion and interpretation of results are based. File 1. Bulk-sample XRF chemical analyses and petrographic summaries: Platoro-Dulce dikes, lavas, and associated rocks. File 2. Inductively coupled plasma mass spectrometry (ICP-MS) analyses: Platoro-Dulce dikes, lavas, and associated rocks. File 3. Previously unpublished chemical analyses of Dulce dikes in southern Colorado and northern New Mexico (E. Landis and W. Hail, Jr., ca. 1972, personal commun.). Analyses by U.S. Geological Survey “rapid-rock” analytical methods (Shapiro, 1967). File 4. Leaching of carbonate with dilute HCl; effects on bulk-sample compositions. A. Dulce Dikes: XRF bulk-sample vs. HCl-leached analyses. B. Dulce dikes: ICP-MS bulk sample versus HCl-leached analyses. File 5. Summary of 40Ar/39Ar age data, analytical methods, and instrumentation. Please visit https://doi.org/10.1130/GES02068.S1 or access the full-text article on www.gsapubs.org to view the Supplemental Files.
Science Editor: Shanaka de Silva
Associate Editor: Valerio Acocella
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.