Tectonically controlled drainage fragmentation in the southwestern Great Basin, USA

Faulting and volcanism can produce mountain ranges that alter regional hydrological and ecological systems (Craw et al., 2008; Luebert and Muller, 2015; Antonelli et al., 2018; Rahbek et al., 2019; Perrigo et al., 2020). The Great Basin of the western USA represents a prime example of such interplay. Dozens of endorheic (internally drained) basins are separated by fault-bounded mountain ranges. The 300–800-km-wide physiographic province formed from a pre-existing high elevation, low-relief plateau that spanned much of the southwestern USA Cordillera from Paleocene to middle Eocene time (e.g., Coney and Harms, 1984; DeCelles, 2004; Cassel et al., 2014; Chapman et al., 2020; Zhou and Liu, 2019).

This “Nevadaplano” (DeCelles, 2004) was capped by diachronous erosional surfaces crossed by paleo-channels that transported sediment for hundreds of kilometers (e.g., Busby and Putirka, 2009; Jayko, 2009; Henry et al., 2012; Busby et al., 2016; Miller et al., 2022). Extensional collapse from middle Eocene to Holocene time (e.g., McQuarrie and Wernicke, 2005; Zhou and Liu, 2019) eventually fragmented many fluvial links across the plateau, leaving the endorheic basins.

The southwestern Great Basin (Figs. 1 and 2) formed when a final remnant of the over-thickened Nevadaplano crust extended and thinned (e.g., Dickinson, 2006). Most extension of the region was accomplished in late Miocene to Holocene times along regional detachment faults, range-bounding normal faults, and transtensional fault systems (Hamilton and Myers, 1966; Burchfiel and Stewart, 1966; Stewart, 1967, 1983; Wright et al., 1974; Wernicke et al., 1988; Lutz et al., 2021, 2022). The transtensional faults are part of the southern Walker Lane (see inset of Fig. 1). Extension and transtension in the area reduced local crustal thickness and average elevation from 50 km to 60 km and 2500–3500 m to 30–35 km and 1000–1500 m, respectively (e.g., Coney and Harms, 1984; Wernicke et al., 1988; Snow and Wernicke, 2000; Bahadori et al., 2018; Zhou and Liu, 2019; Lutz et al., 2021).

The young age of extension, well-studied faults, and preservation of pre-, syn-, and post-extensional sedimentary and volcanic rocks (e.g., Knott et al., 1999; Wright et al., 1999; Wernicke et al., 1988; Snow and Wernicke, 2000; Niemi et al., 2001; Fridrich and Thompson, 2011; Lutz et al., 2021) allow fairly precise paleogeographic reconstructions in the southwestern Great Basin (e.g., Snow and Wernicke, 2000; Fridrich and Thompson, 2011). However, most paleogeographic studies (Table 1) have focused on middle to late Miocene rocks.

Here, we constrain the age and extent of latest Miocene to Pliocene (ca. 6–3 Ma) drainage systems in the southwestern Great Basin by synthesizing new field observations, geochemistry, and geochronology with existing sedimentological studies. First, we review Miocene paleogeography and fluvio-lacustrine systems of the southwestern Great Basin area (Fig. 1 and Table 1). Then, we present new (1) field observations, (2) geochemical data, (3) paired detrital zircon U/Pb and detrital K-feldspar ⁴⁰Ar/³⁹Ar analyses of alluvial fanglomerates, and (4) ⁴⁰Ar/³⁹Ar dates of volcanic rocks that demonstrate latest Miocene–Pliocene remnants of the Miocene drainages (Fig. 2 and Table 2).

Finally, we discuss how intraplate transtensional faulting and volcanism produced hydrologic barriers, fragmented the fluvial network, and yielded a new system of endorheic basins. Formation of the endorheic basins isolated previously interbreeding springsnail and pupfish populations, leading to the evolution of endemic species in the region’s modern springs.

During middle Miocene times (ca. 16–11 Ma), the area now occupied by the southwestern Great Basin was characterized by a generally south- to southeast-directed paleo-fluvial gradient that originated along a drainage divide in the west (Fig. 1 and Table 1). Oligocene to late Miocene erosional surfaces record relatively high-elevation, moderate-relief (525 ± 175 m; Jayko, 2009) topography that was actively eroding, whereas contemporaneous fluviolacustrine basins in the east and southeast (Fig. 1) received sediment predominately from western and northern highlands and locally derived fanglomerates (Table 1; Jayko, 2009; Fridrich and Thompson, 2011).

Lava flows and pumice clasts sourced in the eastern Sierra Nevada region flowed east and west, respectively (see Figs. 1A and 1B), from the area now occupied by the White Mountains and northern Owens Valley, defining a local middle Miocene paleo-divide (e.g., Huber, 1981; Phillips et al., 2011). If this divide also existed from middle Eocene to early Miocene times (e.g., Zhou and Liu, 2019), then it was likely the southern extension of a province-scale, north- to northeast-trending paleo-divide that ran through western and central Nevada and separated the regional Nevadaplano into east- and west-draining fluvial systems (Busby and Putirka, 2009; Henry et al., 2012; Busby et al., 2016).

The existence of middle Eocene to Miocene (ca. 40–12 Ma) fluvio-lacustrine sedimentary rocks east of Death Valley (Figs. 1 and 2; e.g., Reynolds, 1969; Fridrich and Thompson, 2011; Niemi, 2012; Miller et al., 2022) and contemporaneous marine rocks along the southwestern edge of the Sierra Nevada (e.g., Dibblee, 1962; Dibblee and Minch, 2008) that are separated by the relatively high-elevation composite Lindgren-Inyo erosional surface (Fig. 1; Jayko, 2009) supports the presence of the divide before Miocene time.

Miocene sedimentary and volcanic rocks deposited on the Inyo surface and in the eastern fluvio-lacustrine basins record a south- to southeast-directed fluvial course (Fig. 1 and Table 1). Clasts of fluvial conglomerates and sandstones within eastern basins have unique western and northern provenances as well as predominately east- to south-directed paleocurrent indicators (Fig. 1; Table 1). Taken together, the paths shown in Figure 1 and detailed in Table 1 define an apparent interconnected network of fluvio-lacustrine drainages (Figs. 1 and 2).

The ca. 175 Ma Hunter Mountain Batholith in the southern part of the Cottonwood Mountains (Figs. 1–3) is a distinct lithologic source in the southwestern Great Basin (e.g., Niemi et al., 2001; Dunne and Walker, 2004). The Hunter Mountain Batholith is a multiphase intrusive body with lithology that varies from monzonite to leuco-gabbro and ages spanning ca. 155–180 Ma (see fig. 10 in Niemi et al., 2001). However, ca. 175 Ma is the accepted age of the main quartz monzonite phase of the batholith (Niemi, 2013).

Hunter Mountain Batholith clasts are found within Miocene sedimentary rocks of the Eagle Mountain Formation (Niemi et al., 2001) in southeastern Amargosa Valley (E in Fig. 1 and Table 1) and within the Furnace Creek Basin of Death Valley (Wright et al., 1999). These Hunter Mountain Batholith-bearing deposits preserve a fluvial connection that predates the topographic low of modern Death Valley that now separates the Black Mountains from the Cottonwood Mountains (Figs. 1–3).

The ca. 11.4–13.4 Ma part of the Eagle Mountain Formation (Niemi et al., 2001) contains alluvial fan (Niemi et al., 2001) or distal braided fluvial (Renik et al., 2008) conglomerates with Hunter Mountain Batholith clasts. Conglomerate clasts include >1-m-diameter boulders of Hunter Mountain Batholith (Niemi et al., 2001). Niemi et al. (2001) correlated the boulders to the Hunter Mountain Batholith based on composition and U/Pb zircon dates. Renik et al. (2008) found mostly south- to southeast-directed paleocurrent indicators within the Eagle Mountain Formation conglomerates, which is consistent with the east–southeast fluvial course.

Wright et al. (1999) observed quartz monzonite clasts in the upper part (ca. 12.7–10.6 Ma) of the middle to late Miocene Artist Drive Formation (see D in Fig. 1 and Table 1; Wright et al., 1999). They suspected the clasts were from the Hunter Mountain Batholith. The conglomerates are debris flows with poorly sorted, angular to moderately well rounded, and matrix-supported cobbles to boulders of quartz monzonite and meta-carbonate (Wright et al., 1999). Wright et al. (1999) interpreted the southeast-directed paleocurrent indicators as evidence of northwest–southeast flow from the southern Cottonwood Mountains.

Combined tectonic and sedimentary transport have been proposed to explain the presence of Hunter Mountain Batholith clasts in Miocene rocks east of Death Valley. Presently, the Hunter Mountain Batholith lies ~60–70 km and ~100 km from the Artist Drive and Eagle Mountain formation outcrops, respectively (D and E in Fig. 1), with the Death Valley, the Black Mountains, and Greenwater Range lying between them. Niemi et al. (2001) and Niemi (2013) suggested that Eagle Mountain was <20 km from the Hunter Mountain Batholith at ca. 11.4 Ma, which indicates ~80–100 km of NW-directed tectonic transport since then (Fig. 1). This is consistent with several tectonic reconstructions, based on independent lines of evidence (Stewart, 1983; Wernicke et al., 1988; Snow and Wernicke, 2000; Andrew, 2010; Lutz et al., 2021; Lutz et al., 2022). However, the presence of quartz-monzonite clasts and southeast-directed paleocurrent indicators in ca. 12.7–10.6 Ma rocks at the Desolation Canyon and Billie Mine localities (D in Fig. 1) suggests a maximum of 60–70 km of tectonic transport of the Hunter Mountain Batholith, and 30–40 km of sedimentary transport between the Artist Drive and Eagle Mountain formations.

Conglomerates of the Furnace Creek Formation preserve the latest Miocene to Pliocene (ca. 6.5–3.5 Ma) evolution of fluvial drainages in Death Valley (McAllister, 1970; Wright and Troxel, 1993; Wright et al., 1999; Fridrich et al., 2012). Furnace Creek Formation conglomerates are well-exposed at Gower Gulch (Figs. 3 and 4), the East Coleman Hills (Figs. 3 and 5), and Salt Creek (Figs. 3 and 6). The Gower Gulch and the East Coleman Hills sites are located in the northern Black Mountains, and the Salt Creek site is at the northeastern end of the Panamint Range (Fig. 3).

The Conglomerate of Gower Gulch (Wright et al., 1999; Nfc in Fig. 4), like underlying fluvial arkosic sandstones and debris flows within the Artist Drive Formation, records fluvial connection between the Cottonwood Mountains and Furnace Creek Basin during deposition (see path P to R in Fig. 2 and Table 2). The conglomerate contains Hunter Mountain Batholith-like quartz monzonite clasts and rare southeast-directed paleocurrent indicators (Wright et al., 1999; Fridrich et al., 2012). Wright et al. (1999) described the Conglomerate of Gower Gulch as torrent-related debrites, based on very poor sorting, crude layering, and matrix support of pebble- to small boulder-sized clasts. Fridrich et al. (2012) reported well-rounded to subrounded clasts of similar size and composition. Clasts are predominately Paleozoic carbonate and chert, with minor reworked Miocene volcanic and sedimentary rocks of the underlying Artist Drive Formation, and quartz monzonite (Fig. 4) (Wright et al., 1999; Fridrich et al., 2012).

The age of the Conglomerate of Gower Gulch is poorly constrained. McAllister (1970) mapped the conglomerate as the base of the Furnace Creek Formation (Nfc in Fig. 4), whereas Wright et al. (1999) assigned this same conglomerate to the upper Artist Drive Formation (Na in Fig. 4). The conglomerate is well-exposed at the western front of the Black Mountains and underlies several basalt flows, one of which was dated at 5.87 ± 0.12 Ma (Muessig et al., 2019).

Conglomerates within the upper part of the Furnace Creek Formation are well-exposed in the East Coleman Hills within the Texas Spring Syncline (Fig. 5). McAllister (1970) mapped Tfcu (Nfcu in Fig. 5) in the northeast limb of the syncline in the colemanite-bearing portion of the Furnace Creek Formation. Based on McAllister’s (1970) mapping and Knott et al.’s (2005; 2018) tephrochronology, the Nfcu unit is stratigraphically below the ca. 3.54–3.21 Ma uppermost part of the Furnace Creek Formation (Nfu and Nft). Knott et al. (2018) showed that the top of the main body member of the Furnace Creek Formation (Nf in Fig. 5; Tf of Fridrich et al., 2012), which underlies the Nfcu unit, is 4.187 Ma. These data suggest that unit Nfcu is ca. 4.19–3.54 Ma.

Wright and Troxel (1993) mapped conglomerate beds within the Furnace Creek Formation west of Salt Creek (their QTfc) and noted the prominence of Hunter Mountain Batholith clasts. A Quaternary or Pliocene lava flow (their QTb; Nfand in Fig. 6) overlies the conglomerate. The lava flow was dated at 4.8 ± 0.8 Ma (whole-rock K-Ar; L.A. Wright, 1998, personal commun.).

In an east- to west-trending valley across the southern Last Chance Range, ~5 km north of Dry Mountain, Burchfiel (1969) mapped a Plio-Pleistocene basalt flow filling the valley (Nb and dark green in Fig. 7). Burchfiel (1969) noted flow structures within the “basalt tongue” that showed a west–east flow direction. He inferred that the basalt erupted in the Saline Range, and that Dry Mountain was uplifted relative to Saline Valley after eruption. Several Neogene basalt outcrops are mapped in the eastern Saline Range, across the alluvial piedmont west of the valley, but Burchfiel (1969) did not correlate any of these basalt outcrops to the “basalt tongue.” Sternlof (1988) dated the basalt tongue at 3.17 ± 0.12 Ma (whole-rock, K-Ar). At the eastern end of the valley, Burchfiel (1969) mapped a 3 km² Neogene basalt cone with related flows (Nv or “andesite cone” in Fig. 7).

About 12 km north of Dry Mountain, Wrucke and Corbett (1990) mapped Neogene (Tertiary) volcanic rocks that erupted in the Saline Range, flowed east across what is presently the Last Chance Range, onto the eastern piedmont, and into Death Valley (red arrow north of path M in Fig. 2). These are olivine basalt to trachybasalt dated at ca. 4.2 ± 0.3 Ma to 4.6 ± 0.3 Ma (whole rock K-Ar; Elliott et al., 1984).

The Neogene Nova Formation was deposited in the Nova Basin between the northern Panamint Range and southern Cottonwood Mountains (Hunt and Mabey, 1966; Hall, 1971; Hodges et al., 1989). The Nova Formation (Nnl, m, and u in Fig. 8) consists of conglomerate, sandstone, and megabreccia with intercalated volcanic flows (Hodges et al., 1989). The megabreccia and conglomerates record erosion of the Panamint Mountains to the southeast (Hall, 1971; Hodges et al., 1989; Fig. 3). Neither Hall (1971) nor Hodges et al. (1989) mentioned the presence of clasts from the Hunter Mountain Batholith, which is exposed 15 km to the north–northwest (Figs. 2, 3, and 8), within the Nova Formation conglomerates. Geochemical data indicate that the volcanic flows erupted from the Darwin Plateau volcanic field in the Argus Range west of Panamint Valley (path N in Fig. 2; Coleman and Walker, 1990).

Lava flows in the Nova Basin were dated previously at ca. 3.5–5.8 Ma by K-Ar dating (Hall, 1971; Larson, 1979; Hodges et al., 1989) and re-dated at ca. 3.5–7.2 Ma by ⁴⁰Ar/³⁹Ar dating (Snyder and Hodges, 2000; Fig. 8 and Table 3). Some of the ⁴⁰Ar/³⁹Ar ages, however, were based on imprecise age spectra (e.g., samples NB-2, NB-3, and NB-4) and/or isochron ages that incorporated heating steps with large errors (e.g., NB-5).

Traditional field methods of mapping, observation, and sampling were applied to the specific study areas (see Fig. 2). Samples of volcanic groundmass, granitoid clasts, and/or conglomerate matrix were collected for geochronology and major/trace-element geochemistry.

Clasts and sandy matrix of conglomerates at Salt Creek, East Coleman Hills, and Gower Gulch were dated using zircon U/Pb methods, following the procedures described in Gehrels (2014) and references therein (see Supplemental Material¹). Detrital zircons were separated at California State University, Fullerton, California, USA, using standard mineral separation techniques, which leverage the characteristic high density and low magnetism of zircon. Zircon mounts were prepared and imaged by cathodoluminescence at the Arizona LaserChron Center, Tucson, Arizona, USA; zircon U-Pb geochronology was conducted using high-resolution laser ablation–inductively coupled plasma–mass spectrometry on a Thermo Scientific Element 2 instrument (Gehrels et al., 2006, 2008). U-Pb isotopic data were reduced with 20% discordance and 5% reverse discordance filters using E2agecalc. Approximately 20–30 and 250 zircons were analyzed from clast and matrix samples, respectively.

Volcanic samples and granitoid clasts from the Furnace Creek Formation were collected for whole-rock X-ray fluorescence (XRF) analysis to determine their major, minor, and trace-element compositions. Samples were crushed and powdered at California State University, Fullerton. At Pomona College, Claremont, California, di-lithium tetraborate flux was added to the powders, which were fused twice in graphite crucibles, to maximize homogeneity of the glass beads, which then were analyzed on a Panalytical Axios X-ray fluorescence spectrometer.

Samples of volcanic rocks and conglomerate matrix from the Furnace Creek Formation were collected for ⁴⁰Ar/³⁹Ar dating at the New Mexico Geochronology Research Laboratory (NMGRL), Socorro, New Mexico, USA. We dated volcanic groundmass and detrital-matrix K-feldspar grains.

Volcanic groundmass was prepared for irradiation using standard mineral separation methods. Samples were crushed and sieved. Magnetic grains were separated. Two aliquots of groundmass per sample (~40 g) were hand-picked to eliminate phenocrystic contamination. Groundmass samples were loaded into machined aluminum discs and irradiated in a TRIGA Reactor at either Denver, Colorado, USA (USGS) or Oregon State University, Corvallis, Oregon, USA (see Supplemental Material for a full description of irradiation times and geometries for each sample; see ^{footnote 1}).

Eruption ages of intermediate to mafic volcanic rocks were estimated by incremental step-heating experiments performed on the Felix system at NMGRL (see Supplemental Material for more details). Groundmass samples were heated in typical 45 s steps, followed by 60 s of gas clean up with an SAES GP-50 getter operated at 2 A. Isotopes of argon were measured using a Thermo Scientific HELIX MC Plus multicollector–mass spectrometer. Isotopes ⁴⁰Ar, ³⁹Ar, ³⁸Ar, and ³⁷Ar were measured on Faraday collectors, whereas ³⁶Ar was measured on a compact discrete dynode (CDD) ion counter.

Detrital K-feldspar crystals and neutron flux monitors were dated by single crystal laser fusion (SCLF) utilizing the Jan system at NMGRL (see Supplemental Material for more details). Crystals were fused with a CO₂ laser. Extracted gas was cleaned with a variety of getter configurations. Cleaned gas was then analyzed using a Thermo Scientific Argus VI multicollector-mass spectrometer.

Isotopes of low-concentration samples were collected for 280–400 s followed by 45 s of baseline measurement. High-concentration samples (large unknown grains and FC-2) were measured using 120 s of isotope collection followed by 30–60 s of baseline measurement. Analyses were truncated based on various criteria to facilitate efficient data collection. For instance, relatively old grains that did not contribute significantly to maximum depositional age determination or provenance were analyzed for durations of typically less than 60 s. Basalt samples used 340 s of data collection followed by 60 s of baseline measurement.

Data were collected and reduced with in-house Pychron software and MassSpec version 7.875, respectively (see Supplemental Material for all data tables). Fish Canyon Tuff sanidine (FC-2) with an assigned age of 28.201 Ma (Kuiper et al., 2008) was used as a neutron flux monitor. A ⁴⁰K decay constant of 5.463e-10/a was used (Min et al., 2000) with isotope abundances after Steiger and Jäger (1977).

We found granitoid boulders in the Furnace Creek Formation at Gower Gulch, East Coleman Hills, and Salt Creek (Figs. 3–6). Total alkali versus silica plots from XRF analyses show that all granitoid clasts, including those eroding directly from the Hunter Mountain Batholith at Cottonwood Canyon (Figs. 2 and 3), are monzonite (Fig. 3). The U/Pb zircon ages of monzonite clasts from Furnace Creek Formation Conglomerate at Gower Gulch and Coleman Hills are 177.1 ± 2.4 Ma (Fig. 4) and 173.72 (+5.51 –1.67) Ma (Fig. 5), respectively. These ca. 172–179 Ma dates and monzonite rock types are within error of the Hunter Mountain Batholith (Niemi et al., 2001; Niemi, 2013), which strongly supports that the monzonite clasts in the Furnace Creek Formation conglomerates are from the Hunter Mountain Batholith.

Lithology and deposition. The Conglomerate at Gower Gulch includes subrounded to rounded, gravel- to cobble-sized clasts of Paleozoic carbonate, volcanic rock, and Hunter Mountain Batholith in a sand matrix (Fig. 4). The conglomerate is mostly clast supported and poorly bedded. Granitoid rocks make up ~10% of the clasts, consistent with the data presented by Wright et al. (1999). Our outcrop observations confirm Wright et al.’s (1999) interpretation that these are debris-flow deposits in a braided stream setting. Microscopically, we found rounded to subrounded sanidine crystals in the Conglomerate of Gower Gulch matrix, which supports the fluvial interpretation.

Stratigraphic age. We determined a whole-rock ⁴⁰Ar/³⁹Ar date of 6.197 ± 0.020 Ma for the basalt (Tfb of McAllister, 1970, and Fridrich et al., 2012) intercalated with Furnace Creek Formation sediments overlying the Conglomerate of Gower Gulch (Fig. 9B and Table 3).

Detrital zircon and feldspar. The distribution of detrital K-feldspar and zircon ages from the Conglomerate of Gower Gulch is broad (Fig. 10; also see Supplemental Material), which suggests a mixed-provenance fluvial system. Detrital zircons in the Conglomerate of Gower Gulch matrix ranged in age from 2.9 Ga to 7.1 ± 0.2 Ma. Of the 254 zircons analyzed, 21 crystals (8%) had dates that exceeded 1 Ga.

There is a dominant peak in the zircon probability distribution at ca. 175 Ma (Fig. 10B), which is the accepted age for the main monzonitic phase of the Hunter Mountain Batholith (e.g., Niemi, 2013). Approximately 22% (55/253) of the detrital zircons fall within error of 155–180 Ma (Fig. 10B; also see Supplemental Material). These could be from Hunter Mountain Batholith or other Jurassic intrusive rocks exposed farther west in the White, Inyo, Argus, and Slate Ranges (Fig. 11). The youngest grain dated was 8.1 ± 0.3 Ma and is considered the maximum depositional age.

Eighty-one detrital K-feldspar grains from the Conglomerate of Gower Gulch matrix were dated by the ⁴⁰Ar/³⁹Ar method. Dates ranged from 158 Ma to 8.99 Ma. There is a major peak at ca. 11.4 Ma, which most likely represents eruptions from the southwest Nevada volcanic field. The youngest acceptable grain date was 8.990 ± 0.046 Ma.

Lithology and deposition. The conglomerate in the East Coleman Hills is massive, poorly sorted, poorly bedded, and matrix-supported, which is consistent with deposition as a debris flow (Fig. 5). Subrounded to angular monzonite clasts range in size from gravels to boulders (up to 1 m diameter). The contact between the conglomerate and surrounding, finer-grained member of the Furnace Creek Formation (unit Nf; Fig. 5) is abrupt where exposed, and disconformable.

Stratigraphic Age. Faulting and folding complicate stratigraphic age determination. Our field observations did not improve the ca. 4.19–3.54 Ma stratigraphic age of the upper conglomerate member of the Furnace Creek Formation (Nfcu) unit based on previous mapping by McAllister (1970) and constraints in Knott et al. (2018).

Detrital zircon and feldspar. The detrital K-feldspar and zircon grains from the conglomerate in the East Coleman Hills inform the provenance assessment. Detrital zircon U/Pb ages range from 2.6 Ga to ca. 9.8 Ma with ~68% (167/245) within error of 180–155 Ma (Fig. 10B; see also Supplemental Material). There is a major peak at ca. 175 Ma, with minor peaks at 150–155 Ma, 100 Ma, 85 Ma, and ca. 72 Ma (Fig. 10B). The youngest zircon grain was 9.8 ± 0.1 Ma (Table 4).

The 130 detrital K-feldspar grains analyzed for ⁴⁰Ar/³⁹Ar yielded dates ranging from 199 Ma to 9.76 ± 0.02 Ma (see Supplemental Material). Thirty-seven (28%) K-feldspars yielded dates within error of 180–155 Ma or within the age of the Hunter Mountain Batholith (see Supplemental Material).

Lithology and deposition. The conglomerate in the Furnace Creek Formation of Wright and Troxel (1993), west of Salt Creek, is massive to poorly bedded and matrix-supported, which is consistent with debris-flow deposition. Clasts are subrounded to angular with many >1.5 m in diameter (Fig. 6B); boulder-sized clasts are exclusively Hunter Mountain Batholith monzonite. XRF data show that the reddish-black volcanic rock overlying the conglomerate is andesite (Nfand in Fig. 6). The flow-banding in the andesite is conformable with the northeast dips of the underlying conglomerate (Fig. 6).

Stratigraphic Age. The minimum age of the conglomerate was determined by the overlying andesite. The whole rock ⁴⁰Ar/³⁹Ar date on the andesite is 4.959 ± 0.038 Ma (Fig. 9A and Table 3).

Detrital zircon and feldspar. Most detrital zircon and K-feldspar dates from the Furnace Creek Formation Conglomerate west of Salt Creek have a relatively narrow range (Fig. 10; also see Supplemental Material). Three hundred and thirteen U/Pb dates were measured on zircons from the conglomerate matrix with dates ranging from 2.5 Ga to ca. 6.9 Ma. Ninety percent (228/2533) of the zircons dated were within error of 180–155 Ma. There is a major peak at ca. 175 Ma, with minor peaks at 150–155 Ma, 100 Ma, 85 Ma, and ca. 72 Ma (Fig. 10B). The spectrum is quite similar to that of the East Coleman Hills, which suggests common provenance. Only nine zircons (3%) were younger than 12.1 Ma. The youngest zircon grain dated was 6.9 ± 0.1 Ma.

The 120 detrital K-feldspar grains analyzed for ⁴⁰Ar/³⁹Ar yielded dates between ca. 205 Ma to ca. 7.5 Ma. There is a major peak at ca. 175 Ma. The youngest acceptable grain dated was 7.47 ± 0.10 Ma.

Preferred eruption ages of volcanic rocks collected north of Dry Mountain, at the Darwin Plateau and Nova Basin, are presented in Figures 7, 8, 12, and 13 and Table 3. In many cases, age-spectra and inverse isochron analysis indicate that the dated aliquots contained excess argon and a trapped component of atmospheric argon. Detailed explanations of eruption age interpretations, complete data tables, age-spectra, and isochron diagrams are in the Supplemental Material.

In the Last Chance Range, north of Dry Mountain, the volcanic cone at the mouth of the east-west–trending canyon is an andesite (Fig. S1 and Table 4) with an ⁴⁰Ar/³⁹Ar date of 4.615 ± 0.013 Ma (Fig. 12 and Table 3). Additional flows that were emitted from the cone and accumulated in the canyon thalweg are black, vesicular basalt. A separate, blocky-to-vesicular basalt flow lies up-gradient and to the west. This flow is 3.921 ± 0.026 Ma (Fig. 12 and Table 3). This western basalt flow is the same flow from which Sternlof (1988) obtained a whole rock, K-Ar date of 3.17 ± 0.12 Ma. Based on the lack of a volcanic center and the geomorphic position, we concur with Burchfiel’s (1969) conclusion that the 3.92 Ma basalt flow erupted from the Saline Range.

Seven ⁴⁰Ar/³⁹Ar dates were determined on basalt to trachyandesite flows (see Table 4 for geochemical data and Fig. S1 for the total alkali silica [TAS] plot) along an east–west transect through the Nova Basin (Fig. 8). The ages of these volcanic rocks ranged from 5.13 ± 0.03 Ma to 3.832 ± 0.006 Ma (Fig. 8 and Table 3). Coleman and Walker (1990) determined geochemically that Nova Basin volcanic rocks erupted from the Darwin Plateau (Fig. 8) and that, most critically, the youngest date of 3.832 ± 0.006 Ma at Black Point (Fig. 13) is in the stratigraphically highest position and farthest east.

Our results support the existence of an interconnected, southeast-draining fluvial system in latest Miocene to Pliocene times. We suggest that the system had headwaters in the White-Inyo Range and extended to the Furnace Creek Basin and Amargosa Valley. The system can be broken up into two distinct fluvial courses (Figs. 14 and 15), with each supported by independent provenance, paleocurrent indicators, and age.

New dates on volcanic rocks in the Dry Mountain area, when combined with previous studies, define a Pliocene (ca. 4.5–3.9 Ma) and topographically viable fluvial course that may have connected the modern White-Inyo Range, Saline Range area, and Death Valley (Figs. 11, 14, and 15). This fluvial course may have existed in Miocene time as well (see Fig. 1 and Table 1), but direct evidence is lacking.

The 3.921 ± 0.026 Ma basalt tongue flowed east from the Saline Range through the trending valley north of Dry Mountain (dark green in Figs. 7 and 12). The valley was a local topographic low by at least 4.517 ± 0.013 Ma, when lava from the andesite cone erupted along the northern flank of the valley and flowed south (andesite in Figs. 7 and 12). These data show that the valley was at a lower elevation than the Saline Range to the west from before ca. 4.5 Ma until ca. 3.9 Ma (Figs. 14 and 15).

We infer that the valley north of Dry Mountain extended west into the area now occupied by the modern Saline Range, Saline Valley, and Eureka Valley (Figs. 14 and 15A), although this inference depends upon how a low gravity anomaly beneath the Saline Range is interpreted. The low gravity anomaly beneath the ca. 4.6–1.45 Ma volcanic rocks in the Saline Range (Elliott et al., 1984; Sternlof, 1988; this study) has been interpreted as either (1) an alluvium-filled basin (Figs. 7 and 14), or (2) low-gravity plutons (Blakely and McKee, 1985; Blakely et al., 1999).

If the gravity low beneath the Saline Range represents a non-terminal Miocene basin, a topographic low extended from the eastern edge of the modern White-Inyo Range to the valley north of modern-day Dry Mountain at ca. 4.6 Ma (see paths labeled M in Figs. 2, 14, and 15A). Alternatively, if the gravity low beneath the Saline Range represents a Miocene terminal basin, east-directed flow of lava through the valley north of Dry Mountain (Fig. 7) could indicate that the basin was filled and overtopped. If the gravity low beneath the Saline Range represents subsurface plutons, the east-directed flow of lava through the valley north of Dry Mountain could indicate local high elevation from magma-chamber inflation during the early stages of construction of the Saline Range volcanic center.

We infer that, during the late Pliocene, the valley north of Dry Mountain drained north toward the alluvial-lacustrine Mio-Pliocene Ubehebe Basin at the northern end of the Cottonwood Mountains, which connected to the Furnace Creek Basin via Death Valley.

This inference is based on the fact that: (1) the modern valley (tributary to modern Death Valley) drains north toward the Ubehebe Basin, which was an active depocenter in latest Miocene to Pliocene times (Snow and Lux, 1999) and earlier. We deduce that the canyon would have drained north, not south, because there are no active depocenters southward except across Hunter Mountain and the southern Cottonwood Mountains; and (2) within the Ubehebe Basin, there are lake deposits, which suggests this area is down-fluvial gradient from the valley north of Dry Mountain.

The Pliocene fluvio-lacustrine system connected southeastward, in turn, to the Furnace Creek Basin (Knott et al., 2008, 2018). Together, the above constraints support the presence of a ca. 6–3.9 Ma fluvial drainage that extended from the White-Inyo Range, through the area now occupied by the Saline Range, through the valley at Dry Mountain (path M in Figs. 11, 14, and 15A and Table 2), and toward the Ubehebe Basin. At ca. 4 Ma, this drainage would have reached the widespread fluvio-lacustrine system present in Death Valley (paths Q and R in Figs. 2, 14, and 15A).

Our new age constraints on volcanic rocks in the Nova Basin and on Hunter Mountain Batholith-bearing conglomerates in the Furnace Creek Formation define a southern fluvial course in latest Miocene to Pliocene times (ca. 8.1–3.5 Ma). The southern fluvial course reached westward to at least the area of the modern Darwin Plateau in the southern Inyo Range and possibly to the Sierra Nevada. This is supported by (1) the presence of the east-sloping Inyo and Lindgren surfaces (Fig. 1; Jayko, 2009), and (2) the fact that lavas that erupted on the Darwin Plateau flowed northeast into Death Valley prior to formation of the modern-day northern Panamint Valley (Figs. 8 and 14; Hall, 1971; Burchfiel et al., 1987; Hodges et al., 1989; Coleman and Walker, 1990; Andrew and Walker, 2009; this study). Therefore, the late Miocene–Pliocene western drainage divide of proto-Death Valley was located at least as far west as the modern Darwin Plateau (Figs. 11, 14, and 15A).

The youngest of the Darwin Plateau lava flows (trachyandesite of Black Point) reached western Death Valley northwest of Tucki Mountain at 3.832 ± 0.006 Ma (KAH-BP-1 in Figs. 8 and 13). In general, both the presence of ca. 5.2–3.8 Ma lava flows sourced from the Darwin Plateau (Table 3 and see Supplemental Material) and the absence of Hunter Mountain Batholith clasts in the Nova Basin (Hunt and Mabey, 1966; Hall, 1971; Hodges et al., 1989) support east- to northeast-directed flow from the southern Inyo Range to Death Valley during most of Pliocene time, when the Darwin Plateau lay adjacent to the Nova Basin (Fig. 15; Burchfiel et al., 1987).

The east–northeast-directed fluvial course recorded by Darwin Plateau lava flows in the Nova Basin (path N in Figs. 2, 14, and 15A) was probably connected down gradient to the Mio-Pliocene fluvial system preserved in the Furnace Creek Formation deposits at Salt Creek in Death Valley, northeast of Tucki Mountain. Hunter Mountain Batholith clasts within the middle conglomerate member of the Furnace Creek Formation at Salt Creek (path P in Figs. 11, 14, and 15A) record east- to northeastward flow from the southern Cottonwood Mountains toward Death Valley at ca. 6.9–4.96 Ma. Hunter Mountain Batholith clasts in the debris flows at East Coleman Hills (Nfcm) and Gower Gulch (Nfc) record eastward flow from the southern Cottonwood Mountains to the Furnace Creek Basin at ca. 4.1–3.5 Ma (path Q in Figs. 11, 14, and 15A) and ca. 8.1–6.2 Ma (path Q in Figs. 11, 14, and 15A), respectively.

These flow paths clearly pre-date the formation of modern Death Valley as a regional topographic low. The fluvial system recorded by these rocks certainly drained farther southeast to the modern area of Amargosa Valley during late Miocene time (e.g., Wright et al., 1999). The Miocene system may have drained yet further to the south toward the Pacific Coast, based on the paleo-geographic reconstruction in Figure 1 and Table 1. We add that crustal thickness reconstructions (Bahadori et al., 2018) and numerical simulations of paleo-topography (Zhou and Liu, 2019) support a south-facing slope between our study area (Figs. 1 and 2) and the Mojave Block.

Pliocene (ca. 3–5 Ma) fault-controlled topographic development and volcanism in the southwestern Great Basin fragmented the middle to late Miocene fluvial system that ran from the western highlands (White-Inyo Range) to the Furnace Creek Basin and Amargosa Valley (Figs. 14 and 15A). Destruction of the latest Miocene to Pliocene fluvial links across the southwestern Great Basin migrated east to west with tectonism, which is consistent with more regional tectonic studies (e.g., Wernicke, 1992; Snow and Wernicke, 2000; Fridrich and Thompson, 2011; Lutz et al., 2022). This migratory tectonism produced the modern system of endorheic basins (Fig. 15B).

We infer that the Miocene-aged, southeast-flowing river drainage system that connected the areas now occupied by Death Valley and Amargosa (Wright et al., 1999) was dammed and disconnected by Pliocene volcanism in the Greenwater Range (see lava flow dam in Figs. 2 and 15B). At ca. 4.0–3.9 Ma, Pliocene lavas that erupted from the Greenwater Range flowed north across the Furnace Creek Basin (Tibbetts, 2010) where the southeast-flowing braided stream system is preserved in the lower Furnace Creek Formation (Wright et al., 1999; this study). By ca. 3.5 Ma, the fluvial system was succeeded by lake sedimentation (Nfu in Fig. 14; Knott et al., 2018). The succession suggests that lava flows from the Greenwater Range dammed the pre-existing southeast-flowing river, disconnected the Furnace Creek Basin from Amargosa Valley, and promoted lake formation in the Furnace Creek Basin at ca. 3.5 Ma.

The fluvial system between the southern Cottonwood Mountains and the Furnace Creek Basin (paths Q and R in Figs. 11, 14, and 15) was broken by middle Pliocene–Holocene slip on the Black Mountains fault zone, which down-dropped modern Death Valley relative to the Black Mountains and inverted the Furnace Creek Basin (Fig. 15). Alluvial fan deposition in the northern Black Mountains, along with the angular unconformity between the Furnace Creek Formation and the overlying Funeral Formation, show that the northward expansion of the Black Mountains fault zone began at ca. 3.5 Ma and that the Furnace Creek Basin was uplifted at ca. 1.8 Ma (Knott et al., 1999, 2005). The ca. 3.5–2.8 Ma transitional member of the Furnace Creek Formation (Nft; Fig. 5) contains volcanic clasts derived from the Black Mountains (Knott et al., 2018), which further supports uplift of the range at ca. 3.5 Ma.

Thermochronometric data and modeling from the northern and central Black Mountains (Bidgoli et al., 2015; Sizemore et al., 2019) suggest 3–5 km of exhumation since ca. 4 Ma, which was achieved by slip on the Black Mountains fault zone (Figs. 2 and 15B). This is consistent with multidisciplinary field and geochemical analyses (i.e., oxygen isotope, X-ray diffraction, and scanning electron microscope) of gouges that formed along the Black Mountains fault zone, which suggest at least 3 km of fault-zone exhumation since ca. 3 Ma (Hayman, 2006). The combination of fault-controlled uplift and basalt flows formed a topographic barrier between Death Valley and southern Amargosa Valley.

The fluvial courses that connected the Saline Range and Darwin Plateau areas to Death Valley (paths M and N in Figs. 11, 14, and 15) were fragmented by post-3.9 Ma slip on the Dry Mountain and Hunter Mountain–Panamint Valley fault zones, respectively. Slip on the kinematically linked Hunter Mountain–Panamint Valley fault zone (Fig. 8) has fragmented the southern fluvial course (Figs. 14 and 15A) by producing relief between northern Panamint Valley and the Panamint Range since ca. 3.5–4 Ma. This faulting has generated at least 1 km of relief, based on a combination of cross-sectional (Sternlof, 1988) and map-view (Burchfiel et al., 1987) reconstructions and thermochronometric modeling (Bidgoli et al., 2015). Post-3.9 Ma slip on the Dry Mountain fault zone blocked the northern fluvial course by down-dropping the Saline Range and Saline Valley relative to Dry Mountain and the rest of the southern Last Chance Range.

Slip on the Dry Mountain and Hunter Mountain–Panamint Valley fault zones produced the modern, western drainage divide at Death Valley (Panamint Range to Last Chance Range; Figs. 11 and 15), and therefore fragmented the western parts of the northern and southern fluvial courses. This fault slip took place after ca. 3.9 Ma and was likely linked kinematically with slip on the east Inyo fault zone (e.g., Burchfiel et al., 1987; Lee et al., 2009) via the Hunter Mountain fault zone (Fig. 15). We infer that this fault slip generated the modern endorheic conditions in the Panamint and Saline Valley basins (Fig. 15B), although specific evidence of internal drainage since Pliocene time was not found.

The western part of the northern fluvial course (path L in Figs. 2 and 15A) was fragmented by slip on the White Mountains and East Inyo fault zones (Figs. 1 and 2). This faulting dropped the Owens and Saline valleys down, respectively, relative to the White-Inyo Range in middle Pliocene time (Fig. 15). Sedimentological evidence (Bachman, 1978) suggests that middle Pliocene to Holocene (post-2.5 Ma) slip on these fault zones generated 1–1.5 km of relief between the White-Inyo Range and Owens Valley. This is similar to the magnitude of post-2.8 Ma exhumation in the White-Inyo Range, which was estimated independently from thermochronometry and tilt reconstructions (Stockli et al., 2003; Lee et al., 2009). Slip on the White Mountains and East Inyo fault zones coincided, in part, with west-tilting of the Sierra Nevada during slip on the Sierra Nevada frontal fault zone (Fig. 1), which has generated 1.5–2.1 km of rock uplift since ca. 5 Ma (Huber, 1981; Wakabayashi and Sawyer, 2001; Phillips, 2008; Hildreth et al., 2021).

Slip on the Sierra Nevada, White Mountains, and east Inyo fault zones created the modern Sierran crest and White-Inyo Range, which form major drainage divides in the southwestern Great Basin (Fig. 15B). The formation of Owens Valley by fault slip probably resulted in the modern, endorheic conditions within the greater Owens Valley Basin. In turn, endorheic conditions led to the drying of Pliocene lakes in Owens Valley and the formation of isolated springs (Fig. 15B). Progressive Plio-Pleistocene drying and salinization of the Owens Valley Lake system is preserved in the sedimentary (Lueddecke et al., 1998) and fossil (Oseguera, 2012) records of the ca. 2.63–2.06 Ma Waucoba Lake Beds (De Masi, 2013).

Pliocene fault- and volcanic-controlled fragmentation of the fluvial courses described above (Fig. 15) corresponds temporally with the evolution of several endemic lineages of springsnails and pupfish that live in Owens, Death, and Amargosa valleys (ca. 2–5 Ma; e.g., Echelle, 2008; Hershler and Liu, 2008; Hershler et al., 2011, 2013). Interestingly, the endemic species have Miocene ancestors that lived in the Gila River Basin, coastal California, and northern Mexico. How the ancestral forms entered the southwestern Great Basin and what drove their subsequent divergence and endemism remains unknown. Further understanding is contingent on more robust reconstructions of early to middle Miocene paleogeography of a broader area.

Regardless of how the ancestors entered the area now occupied by the southwestern Great Basin, we hypothesize that subsequent divergent speciation was driven by Pliocene (ca. 3.5–4 Ma) geographic isolation, which was the direct consequence of both fault-controlled topographic development and local volcanism that fragmented the pre-existing fluvial network (Fig. 15). If true, then intraplate tectonics exerted first-order control on the biological evolution of endemic spring-dwelling taxa over relatively short length (<100 km) and time scales (a few millions of years). This type of tectonic control on evolutionary biology has not been demonstrated over such short spatial and temporal scales before.

Regional middle Pliocene (ca. 3.5–4 Ma) topographic development and hydrological isolation in the southwestern Great Basin (Fig. 15) was the result of a westward migration of faulting (Wernicke, 1992; Snow and Wernicke, 2000; Fridrich and Thompson, 2011). The westward migration of intraplate faulting has been explained previously by (1) dynamic buoyancy due to return asthenospheric flow in the wake of a delaminated crustal root in the southern Sierra Nevada (Ducea and Saleeby, 1998; Farmer et al., 2002; Saleeby et al., 2003; Jones et al., 2004; Jones and Saleeby, 2013), (2) localization of range-bounding fault zones due to plate-boundary kinematic changes (e.g., Wernicke et al., 1988; Snow and Wernicke, 2000; Stockli et al., 2003; Unruh et al., 2003; McQuarrie and Wernicke 2005; Fridrich and Thompson, 2011; Walker et al., 2014; Lutz et al., 2022), and (3) internal rheological changes in the eastern Death Valley area that caused a shift in strain localization (e.g., Wernicke, 1992; Harry et al., 1993; Fridrich and Thompson, 2011; Lutz et al., 2021, 2022). Critical evaluation of these three tectonic drivers, which are not mutually exclusive, is beyond the scope of this paper.

Regardless of the driver(s), the result was a rapid change in the regional hydrological framework. Late Pliocene to Holocene tectonics and volcanism fragmented a preexisting Mio-Pliocene drainage system and created new environmental niches over the course of 1–2 m.y. The relatively rapid tectonically controlled fragmentation isolated previously interbreeding populations of spring-dwelling taxa, which led to divergent evolution and endemism.

We documented the existence and subsequent fragmentation of a Mio-Pliocene drainage network in the southwestern Great Basin. Previously connected remnants of this drainage system, now preserved in ranges, are now disrupted and separated by hydrologically isolated, endorheic basins (Fig. 15B).

Our results support the existence of two interconnected drainages (or fluvial courses) that predated the local Basin and Range topography. The northern fluvial course (Figs. 14 and 15A), defined primarily by lava flows north of Dry Mountain (Fig. 7), connected the White-Inyo Range area to Death Valley between ca. 4.5 Ma and 3.8 Ma.

The southern fluvial course (Figs. 14 and 15A), represented by lava flows and alluvial fanglomerates, connected the southern Inyo Range (Darwin Plateau area) to Death Valley and Amargosa Valley in latest Miocene to Pliocene times. A connection between the southern Inyo Range and Death Valley existed before 3.8 Ma, based on lava flows in the Nova Basin.

A pre-Death Valley fluvial connection between the southern Cottonwood Mountains and the Furnace Creek Basin existed at ca. 8.1–6.2 Ma (R in Fig. 15A) based on the Hunter Mountain Batholith-bearing Conglomerate of Gower Gulch, and, speculatively, until ca. 3.5 Ma (Q in Fig. 15A), based on the conglomerate in the east Coleman Hills. Fluvial connection between the southern Cottonwood Mountains and Amargosa Valley before 8.1 Ma, and possibly as long ago as ca. 11.4 Ma, is supported by Hunter Mountain Batholith-bearing conglomerates within the Artist Drive Formation (Wright et al., 1999) and Eagle Mountain Formation (Niemi et al., 2001).

Penultimate fragmentation of the Mio-Pliocene fluvial network was achieved by transtensional, range-front faulting and local volcanism (Figs. 2 and 15). Fault-controlled topographic development was an intraplate response to one or more of three potential geodynamic drivers and formed the modern endorheic basins in the region. This rapid, tectonically controlled environmental change probably led to endemism among spring-dwelling taxa in the region through population isolation and allopatric speciation.

This work was supported by National Science Foundation (NSF) Integrated Earth Systems grant EAR-1516680 (to B.M. Lutz, J.R. Knott, F.M. Phillips, and G.J. Axen) and NSF grant EAR-1649254 (to J.R. Knott). Critical reviews of the manuscript and figures by two anonymous reviewers and the associate editor greatly improved the manuscript. We are grateful to Death Valley National Park for the sampling permit. Nels Iverson, Jake Ross, Matt Zimmerer, and Levi Begay assisted with sample preparation, experimental setup, and/or data reduction at New Mexico Geochronology Laboratories. Conversations with the late Lauren Wright, Angela Jayko, Don Sada, and John Umek helped to develop this paper. We thank Mike Machette and the late Bennie Troxel for directing us toward the East Coleman Hills and Salt Creek localities. Diane Clemens-Knott, Nancy Chen, Matt Pilker, and Priscilla Martinez of California State University, Fullerton, assisted with zircon sample preparation and analysis. We also thank the staff at the University of Arizona Laserchron Center and at Pomona College for their assistance with zircon U/Pb dating and X-ray fluorescence analyses, respectively.