Catalina Basin, located within the southern California Inner Continental Borderland (ICB), United States, is traversed by two active submerged fault systems that are part of the broader North America–Pacific plate boundary: the San Clemente fault (along with a prominent splay, the Kimki fault) and the Catalina fault. Previous studies have suggested that the San Clemente fault (SCF) may be accommodating up to half of the ∼8 mm/yr right-lateral slip distributed across the ICB between San Clemente Island and the mainland coast, and that the Catalina fault (CF) acts as a significant restraining bend in the larger transform system. Here, we provide new high-resolution geophysical constraints on the seabed morphology, deformation history, and kinematics of the active faults in and on the margins of Catalina Basin. We significantly revise SCF mapping and describe a discrete releasing bend that corresponds with lows in gravity and magnetic anomalies, as well as a connection between the SCF and the Santa Cruz fault to the north. Subsurface seismic-reflection data show evidence for a vertical SCF with significant lateral offsets, while the CF exhibits lesser cumulative deformation with a vertical component indicated by folding adjacent to the CF. Geodetic data are consistent with SCF right-lateral slip rates as high as ∼3.6 mm/yr and transpressional convergence of <1.5 mm/yr accommodated along the CF. The Quaternary strands of the SCF and CF consistently cut across Miocene and Pliocene structures, suggesting generation of basin and ridge morphology in a previous tectonic environment that has been overprinted by Quaternary transpression. Some inherited crustal fabrics, especially thinned crust and localized, relatively hard crustal blocks, appear to have had a strong influence on the geometry of the main trace of the SCF, whereas inherited faults and other structures (e.g., the Catalina Ridge) appear to have minimal influence on the geometry of active faults in the ICB.

Strike-slip faults are characteristic of continental transform plate boundaries worldwide (e.g., San Andreas fault of southern California [USA], Queen Charlotte fault of southeastern Alaska [USA], North Anatolian fault of Turkey, Alpine fault of New Zealand), and are capable of generating large (M6+) earthquakes (e.g., Stein et al., 1997; Hauksson et al., 2012; Howarth et al., 2012; Yue et al., 2013). Despite their proximity to large population centers, particularly near coasts, there is much we do not yet understand about the way strike-slip systems form and deform. For example, the recent 2016 Mw 7.8 Kaikoura earthquake of New Zealand (e.g., Hollingsworth et al., 2017) highlights the complexity involved during a major strike-slip fault rupture and the need for improved understanding of strike-slip systems. Ongoing research aims to better characterize how, why, and where strike-slip faults form, generate earthquakes, partition slip in oblique environments, and behave at stepovers and endpoints, and ultimately, how to use tectonic geomorphology to quantify deformation and potential geohazards. High-resolution constraints on fault geometry are particularly important for understanding Quaternary fault deformation history (e.g., Brothers et al., 2015) and for characterizing active fault systems, because even a subtle geometry change can inhibit or promote earthquake rupture propagation (e.g., Wesnousky, 2006).

The California Inner Continental Borderland (ICB) offshore of southern California (United States) and northern Baja California (Mexico) (Fig. 1) offers an opportunity to examine a set of active strike-slip faults that accommodate as much as 8 mm/yr of right-lateral shear, or ∼15% of the total Pacific–North America plate boundary slip budget of 48–50 mm/yr (Platt and Becker, 2010; DeMets and Merkouriev, 2016). Several significant earthquakes have occurred along offshore faults in the ICB, including the 1981 Mw 6.0 Santa Barbara Island earthquake, the 1986 Mw 5.8 Oceanside sequence, the recent 2018 Mw 5.3 Santa Cruz Island event, the 1951 Ms 5.9 San Clemente Island earthquake, and the largest recorded ICB earthquake to date, the Ms 6.2 offshore Ensenada earthquake of 1964 (Richter, 1958; Allen et al., 1960; Hauksson and Jones, 1988; Pacheco and Nábělek, 1988; Bent and Helmberger, 1991; Astiz and Shearer, 2000, Legg et al., 2015) (Fig. 1). Shaking from earthquake ruptures can also enhance the risk of local tsunamis via uplift at restraining bends or coseismic slope failure (e.g., Legg and Borrero, 2001; Legg et al., 2004a). Several large submarine landslides have been documented in the offshore California borderland (Bohannon and Gardner, 2004; Locat et al., 2004; Normark et al., 2004a; Lee et al., 2009; Legg and Kamerling, 2003; Brothers et al., 2018).

Understanding how one structure can lead to the formation of another is critical for accurate interpretation of tectonic geomorphology and geohazard assessments. It is well understood that pre-existing crustal fabrics can influence strike-slip fault propagation or fault reactivation (e.g., Christie-Blick and Biddle, 1985; Scholz, 2002; Cunningham and Mann, 2007). Numerous studies have documented reactivation of pre-existing fault structures, including in our study area offshore of southern California (e.g., Crouch and Suppe, 1993; Fisher et al., 2009; Sorlien et al., 2015), and others have analyzed the effects of crustal structures on strike-slip fault geometry (e.g., Johnson and Watt, 2012; Johnson et al., 2018). Christie-Blick and Biddle (1985) differentiated between “essential” and “incidental” pre-existing structures, i.e., structures that significantly influence strike-slip fault geometry and propagation and those that are inherited and do not affect strike-slip deformation, respectively. The study of essential versus incidental pre-existing structures and their relative influence on the development of active fault systems is an important topic in crustal deformation research.

This study focuses on characterizing active structures and deformation in and on the margins of Catalina Basin, an understudied region of the ICB (Figs. 1, 2A), which we interpret to contain two Holocene-active offshore fault zones—the San Clemente and Catalina fault zones. Four of the aforementioned >M5 earthquakes in the ICB occurred near Catalina Basin (Astiz and Shearer, 2000), yet neither the San Clemente fault (SCF) nor the Catalina fault (CF) have been examined systematically with modern high-resolution marine geophysical data within Catalina Basin. Additionally, the SCF alone may be accommodating as much as 4–6 mm/yr of right-lateral slip based on geologic data and GPS models, about half of the total slip taken up within the ICB (Legg, 1985, 1991a, 2005; Larson, 1993; Bennett et al., 1996; Humphreys and Weldon, 1994; Goldfinger et al., 2000), and the CF has been thought to be convergent and thus tsunamigenic (e.g., Legg and Borrero, 2001).

Numerous important studies over the past few decades have described a first-order tectonic and geologic framework for the ICB and Catalina Basin; our study provides, for the first time, a systematic, comprehensive, high-resolution, detailed, and high-quality geophysical data set in Catalina Basin with which to assess prior hypotheses and first-order results. We utilize a suite of new high-resolution multichannel seismic (MCS) data, CHIRP (compressed high-intensity radar pulse) sub-bottom profiles, and high-resolution multibeam bathymetry data in conjunction with legacy crustal-scale and other regional data to examine the relationships between physiography, crustal fabric, and Quaternary and Holocene deformation in and on the margins of Catalina Basin. We revise the geometry and better define the kinematics of the Holocene-active SCF and CF and the Quaternary-active Kimki fault (KF) and discuss the implications for geohazards. We find that Quaternary faults commonly overprint pre-existing structures, and thus that modern physiography does not necessarily indicate the presence of active faults. We also find that inherited crustal blocks (as defined by bulk physical properties of the deep crust), more so than inherited faults, have likely affected Quaternary fault configuration, kinematics, and perhaps even fault formation. In addition to better defining post–late Miocene fault history, geometry, kinematics, and associated geohazards in the Catalina Basin area, this study emphasizes the importance of integrated high-resolution surface and subsurface imaging.

Tectonic Evolution of the Inner Continental Borderland

The ICB extends from the southern California mainland coast to ∼100 km offshore just past the SCF, where the basins of the Outer Continental Borderland neighbor the ICB to the west (Fig. 1). The Western Transverse Ranges province bounds the ICB to the north (Fig. 1). Morphologically, the ICB is characterized by a network of submarine basins, faults, ridges, and islands, with bedrock composed of thinned, extended continental crust, and outcrops featuring Miocene and older volcanic and metamorphic rocks (Barron, 1986; Vedder et al., 1986; Vedder, 1990; ten Brink et al., 2000). The ICB began to form in an extensional tectonic regime beginning ca. 20 Ma (Crouch and Suppe, 1993; Nicholson et al., 1994; Bohannon and Geist, 1998; ten Brink et al., 2000; Miller, 2002; Fisher et al., 2009). The rotation of the Western Transverse Ranges to the north caused rapid opening of the ICB at ca. 19 Ma, shortly after which (ca. 17.5 Ma) subduction of the Arguello microplate ceased, leading to oblique rifting of the Outer Continental Borderland (Lonsdale, 1991; Nicholson et al., 1994; ten Brink et al., 2000). This event was followed by regional transrotation from ca. 18 Ma to ca. 12 Ma, and then transtension from ca. 12 Ma to ca. 6 Ma (Luyendyk et al., 1980; Ingersoll and Rumelhart, 1999; ten Brink et al., 2000). Miocene extension led to the exhumation of the Cretaceous Catalina Schist in the ICB, a metamorphic core associated with past subduction along the margin and correlatable with the Franciscan Complex (e.g., Crouch and Suppe, 1993; Bohannon and Geist, 1998; ten Brink et al., 2000; Miller, 2002).

The basin-and-ridge morphology and some faulting still present in the modern ICB evolved following the ca. 19 Ma rotation of the Western Transverse Ranges. The precise timing of subsequent ICB basin evolution is not well understood; Bohannon et al. (2004) interpreted Pliocene initiation of the Los Angeles Basin, attributing changes in seismic facies to a ca. 6 Ma reversal in physiography that led to flooding of previously subaerial Miocene rocks. The mechanism of local basin subsidence also has not been well defined, perhaps partly due to the complex stress environments in the ICB (Bohannon et al., 2004); Ingersoll and Rumelhart (1999) concluded that basin subsidence may have even occurred independently of plate-motion changes. Proposed models for ICB basin opening include nonuniform crustal thinning (Bohannon et al., 2004) and localized extension along strike-slip releasing bends (Namson and Davis, 1990; Legg et al., 1999, 2007; Legg and Borrero, 2001). Sometime after ca. 6 Ma, the ICB underwent a tectonic transition from transtension to regional transpression, which is ongoing today (e.g., Ingersoll and Rumelhart, 1999; Schindler, 2010).

Age constraints from basin sampling provide regional context for seismic stratigraphy in the ICB basins (e.g., Emery, 1960; Moore, 1969; Gorsline et al., 1984; Gorsline and Douglas, 1987; Teng and Gorsline, 1989; Bohannon et al., 2004; Brothers et al., 2018). Stratigraphic control in the ICB basins comes from Ocean Drilling Program (ODP) sites in the Santa Monica Basin (ODP Leg 167, Site 1015; e.g., Normark and McGann, 2004; Normark et al., 2004b; Romans et al., 2009) and in the San Nicolas Basin (ODP Leg 167, Site 1013; e.g., Shipboard Scientific Party, 1997; Janik et al., 2004; De Hoogh, 2012). Stratigraphic age constraints also come from industry wells in some nearshore areas (e.g., Sorlien et al., 2015) and from shallow sediment cores and grab samples (Nardin et al., 1979; Barron, 1986; Vedder et al., 1986; Vedder, 1990; Normark et al., 2004b). Acoustic basement throughout the ICB (which we refer to herein as simply “basement”) consists of the Catalina Schist, as well as intrusive granitic and volcanic rocks in some areas (e.g., Crouch and Suppe, 1993; Bohannon and Geist, 1998). Atop the acoustic basement, ICB basins contain Miocene sediments characterized by subparallel, highly deformed acoustic reflections that drape the underlying bedrock. A regional unconformity separates deformed Miocene strata from a section of Pliocene-Quaternary sediment as much as 1 km thick, which exhibits more flat-lying, subparallel, coherent reflections and contains a greater proportion of terrigenous material deposited in marine basins following Miocene subsidence and associated sea-level rise (Bohannon et al., 2004).

Active faults that traverse the ICB include: the Newport-Inglewood fault, which carries at least 0.3–0.6 mm/yr of dextral slip at its northernmost extent near Huntington Beach (Grant et al., 1997), with slip rates south of Dana Point increasing to 1–2 mm/yr (Fischer and Mills, 1991); the Rose Canyon fault, which also accommodates 1–2 mm/yr of right-lateral slip (Lindvall and Rockwell, 1995); the Palos Verdes fault, which has a dextral slip rate of 1.6–1.9 mm/yr offshore (Brothers et al., 2015) and 2–4 mm/yr onshore (McNeilan et al., 1996; Ward and Valensise, 1994); and the San Diego Trough fault, which carries 1.2–1.8 mm/yr of right-lateral slip (Ryan et al., 2012). There is no clear consensus on how slip is partitioned amongst individual faults throughout the ICB, owing in large part to uncertainty in mapping of active fault traces and to sparse robust slip-rate estimates along offshore fault segments. Only four studies have provided direct evidence for Late Pleistocene–Holocene lateral offset along faults located offshore (McNeilan et al., 1996; Ryan et al., 2012; Brothers et al., 2015; Conrad et al., 2017). Brothers et al. (2015) noted a 3–5 mm/yr deficit between summed slip rates of the other significant faults in the ICB (i.e., Newport–Inglewood–Rose Canyon, Palos Verdes, San Diego Trough) and the total available slip across the region based on GPS data (6–8 mm/yr); they proposed that the majority of the 3–5 mm/yr deficit is accommodated by the SCF.

Catalina Basin

Catalina Basin is bounded by Santa Catalina Island to the northeast, Santa Barbara Island to the northwest, and San Clemente Island to the southwest (Fig. 2A). These subaerial islands and submerged ridges (e.g., San Clemente Ridge, Catalina Ridge) generally feature exposed Miocene and older deposits including volcanic, intrusive granitic, and metamorphic rocks like the Catalina Schist (Vedder et al., 1986; Vedder, 1990). Although these ridges and islands are composed of largely Miocene and older rocks, the linear, southwestern edge of Catalina Ridge (Fig. 2A) has been previously interpreted as a Quaternary-active fault (e.g., Chaytor et al., 2008; Legg et al., 2015). Emery Knoll, a large (∼10-km-diameter) submerged circular uplift located in southern Catalina Basin (Fig. 2A), has been variably interpreted as a resurgent caldera (Legg et al., 2004b) and a magmatic diapir (Junger and Sylvester, 1979; Ridgway and Zumberge, 2002).

The SCF marks the southwestern boundary of Catalina Basin, and we now know it extends ∼80 km north (this study) and at least 80 km south (e.g., Legg et al., 2007) of the basin (Figs. 2B, 3). The SCF strikes northwest, roughly in line with the Pacific–North America plate motion vector of ∼321° through the region (DeMets et al., 2010); however, several local bends along the fault have been associated with pull-apart basins and popup structures (e.g., Legg et al., 1999, 2007). The SCF has long been considered the dominant fault in the region (e.g., Ridlon, 1968) and has been previously interpreted as a right-lateral strike-slip fault accommodating ∼4–6 mm/yr of dextral motion, with rates estimated from several potential (though unverified) geologic piercing points and GPS models (Legg, 1985, 1991a, 2005; Larson, 1993; Bennett et al., 1996; Humphreys and Weldon, 1994; Goldfinger et al., 2000).

The CF, bounding Catalina Basin to the northeast, has had a number of interpretations, ranging from right lateral to convergent (Kier and Mueller, 1999; Legg et al., 2004a; Chaytor et al., 2008). Previous mapping of the CF within Catalina Basin was done without comprehensive modern high-resolution bathymetry or seismic-reflection data, leading to debate about the fault geometry and relationship with neighboring faults (e.g., Legg et al., 2004a, 2007, 2015; Ryan et al., 2012). The CF has been previously mapped as connecting or interacting with either the Catalina Ridge (Fig. 2A; e.g., Legg et al., 2015) or the Santa Cruz fault zone to the north (Fig. 2B; e.g., Legg et al., 2004a), and/or the San Diego Trough fault to the south (Fig. 2B; e.g., Legg et al., 2007). Each of these fault configurations would serve to extend the length of the CF and therefore its seismogenic potential (e.g., Wells and Coppersmith, 1994; Legg et al., 2004a, 2015). The variable strike of the CF is generally northwest, similar to the SCF, but is obliquely convergent with the Pacific–North America plate motion vector in some areas (DeMets et al., 2010). Because of this obliquity, the Catalina fault has been thought to be transpressional and potentially tsunamigenic (Legg and Borrero, 2001; Legg et al., 2004a, 2007, 2015; Chaytor et al., 2008). Slip rates on the CF are poorly defined, but have been estimated at ∼2.5 mm/yr based on an inferred restoration of Catalina Ridge back to an embayment on western Santa Catalina Island (Chaytor et al., 2008).

In order to better understand the recency of deformation along Catalina Basin fault zones, we consider the age, source, and type of sediment in the basin. The San Gabriel Channel system dominated Late Pleistocene Catalina Basin fill with terrestrial-sourced sediments (Maier et al., 2018), and in the latest Pleistocene and Holocene, hemipelagic sedimentation draped San Gabriel deep-sea fan sediments (e.g., Maier et al., 2018). The San Gabriel depositional system has been interpreted as primarily receiving terrestrially derived sediment during sea-level lowstands due to the ca. 15 ka extrapolated age of the base of the hemipelagic drape layer (Normark et al., 2009; Ryan et al., 2012; Maier et al., 2018) and the 8–18 km distance between the shoreline and San Gabriel canyon heads (Fig. 2A) (Alexander and Lee, 2009; Sommerfield et al., 2009; Maier et al., 2018). Catalina Canyon (Fig. 4), on the northern basin margin, has provided lesser sediment input to Catalina Basin from Santa Catalina Island (Maier et al., 2018). Currently, the best sedimentary age control in the Catalina Basin region consists of paleontological dating of seafloor dart cores (Barron, 1986; Vedder et al., 1986; Vedder, 1990). Although the subsurface fill of Catalina Basin lacks age control, stratigraphic relationships and seismic facies analysis can be used to infer correlative allostratigraphic units across ICB basins due to their similar genetic histories (e.g., Emery, 1960; Moore, 1969; Gorsline et al., 1984; Gorsline and Douglas, 1987; Teng and Gorsline, 1989; Bohannon et al., 2004).

Marine Geophysical Data

The primary data sets used to examine the seafloor and subsurface were acquired in 2014 and 2016 aboard the R/V Robert Gordon Sproul (U.S. Geological Survey [USGS] cruise ID 2014-645-FA) and R/V Thomas Thompson (cruise ID 2016-616-FA), respectively (see Dartnell et al., 2017, and Balster-Gee et al., 2017, 2020], for data releases and detailed survey information). High-resolution MCS reflection and sub-bottom CHIRP data were collected during both the 2014 and 2016 geophysical surveys (Fig. 2B). The 2014 R/V Robert Gordon Sproul data set includes ∼1238 km of coincident 48-channel sparker MCS and 3.5 kHz Knudsen CHIRP data, ∼906 km of which is located within Catalina Basin, the remainder located to the north in Santa Cruz Basin (e.g., Brothers et al., 2018). The 2016 survey aboard the R/V Thomas Thompson acquired 24-channel minisparker MCS reflection data and coincident 3.5 kHz Knudsen CHIRP data totaling ∼726 line kilometers (spacing generally 3–4 km), all within Catalina Basin. High-resolution multibeam bathymetry data were also acquired using the R/V Thomas Thompson’s Kongsberg EM302 hull-mounted multibeam echosounder (MBES) system.

All MCS data were processed to poststack time migration using commercial software (see Balster-Gee et al., 2017, 2020, for details). The MBES data were edited and processed using Caris HIPS and SIPS (; associated backscatter data were processed using the SonarWiz software package. Seafloor data were then gridded at 10 and 15 m resolution throughout Catalina Basin depending on water depth. The 2016 R/V Thomas Thompson MBES data were merged with other high-resolution MBES data sets collected in the ICB since 2010 (Dartnell et al., 2015, 2017). A regional mosaic was generated and gridded at 25 m resolution. Lastly, the Southern California Coastal Relief Model (CRM, version 2; Calsbeek et al., 2013) was used for regions that do not presently contain MBES coverage.

Additional Geophysical Data Sets

Legacy academic and industry MCS data are also available in the Catalina Basin (e.g., Fig. 2B) and were used to examine the entire sedimentary basin fill and character of the acoustic basement, and also to extend fault mapping beyond the edges of the high-resolution data coverage. All legacy data sets crossing our survey area are deeper-penetrating, lower-resolution, crustal-scale MCS data. Two notable legacy MCS surveys crossing Catalina Basin include EW9415 (the Los Angeles Region Seismic Experiment [LARSE] study; e.g., ten Brink et al., 2000) and L490SC (Triezenberg et al., 2016), collected aboard the USGS R/V Lee in 1990. Other relevant exploration industry surveys are publicly available through the USGS National Archive of Marine Seismic Surveys (NAMSS; Triezenberg et al., 2016), including B685SC, W385SC, and W582SC.

We integrate several other regional geophysical data sets with the high-resolution marine geophysical observations, including an earthquake relocation catalog from Hauksson et al. (2012), focal mechanism calculations for a subset of these events (Yang et al., 2012), and significant earthquakes (>M4) occurring in the ICB prior to 1981 from the USGS earthquake catalog ( We also include airborne magnetic anomaly (Langenheim et al., 1993) and Bouguer gravity anomaly (Beyer, 1980; Bankey et al., 2002) data, which have been gridded for the ICB at ∼1 km. Finally, we use GPS velocity data from the Southern California Crustal Motion Map, version 4.0 (Shen et al., 2011), to examine the regional strain field.

Data Analysis

Multibeam bathymetric grid processing (including slope and shaded-relief map generation), analysis of morphological features (e.g., slope failure, ridges), and vector data interpretations (i.e., faults) were done largely using ESRI ArcGIS software. Subsurface (MCS and CHIRP) data interpretation was completed using IHS Kingdom Suite seismic reflection interpretation software. Seismostratigraphic horizons presented in this study represent unconformity surfaces. Isochron sediment thickness maps were generated for key marker horizons, with thickness reported in two-way travel time.

New fault maps were generated using a combination of MBES, high-resolution MCS, and CHIRP data through iterative verification of features mapped at the seafloor and in the subsurface in ArcGIS and Kingdom Suite, respectively. The 2014 R/V Robert Gordon Sproul and 2016 R/V Thomas Thompson data sets provide dense high-resolution multibeam and seismic-reflection data within and on the margins of Catalina Basin, where we map the Quaternary-active SCF, CF, and KF, as well as secondary faults interpreted to be Quaternary active. Holocene (active-fault) offsets were interpreted from offset of the Holocene drape layer (Maier et al., 2018) using CHIRP sub-bottom profile data.

North of Catalina Basin, we slightly modify the geometry of the primary strand of the SCF and the Santa Cruz fault using MBES data and legacy seismic-reflection data sets. We also extend the geometry of the Catalina fault north into Santa Monica Basin using MBES data, high-resolution MCS data across Catalina Ridge, and legacy MCS data within Santa Monica Basin. South of Catalina Basin, we slightly update the primary strand of the SCF from maps provided by Ryan et al. (2009) using MBES data. We do not map or remap any Quaternary fault strands (solid lines in our figures) where we do not explicitly image clear seafloor offset in one or more high-resolution data sets (MBES, MCS, and/or CHIRP). Dashed faults in our maps indicate unverified (likely pre-Quaternary or inactive) fault strands. All fault interpretations from this study have been incorporated into the new USGS Quaternary Faults Offshore of California (QFO) database (Walton et al., 2020) and are publicly accessible.

Here, we present observations of new, high-resolution seafloor and subsurface geophysical data in the Catalina Basin region. We focus on the SCF and CF zones, other related structures, and the stratigraphy of Catalina Basin, and compare our results to plate-motion models, seismicity, and potential-field data sets.

Seafloor Morphology and Surficial Geology

Catalina Basin narrows from ∼40 km wide in the southeast to ∼10 km wide in the northwest (Fig. 2A) and is partially flanked by San Clemente and Santa Catalina Islands. There is significant relief between both islands and the basin floor (as much as ∼1800 m and ∼1700 m, respectively), as well as exceptionally steep submarine slopes (reaching ∼35°) on the basin margins. Conversely, much of the Catalina Basin floor is relatively flat, with slopes generally less than 2°–3° except along channels and scours (e.g., Maier et al., 2018). The depth of the basin increases from ∼1000 m in the southeast where the San Gabriel Channel enters the basin (Fig. 2A) to ∼1300 m in the northwest extent of the basin at the southeastern flank of Santa Barbara Island.

Catalina Basin also contains and is bounded by numerous submerged, elongate ridges. San Clemente Island and Santa Catalina Island both continue as submerged ridges to the north (known as San Clemente Ridge and Catalina Ridge, respectively), both trending NW-SE (Fig. 2A). These linear ridges, particularly Catalina Ridge, have been previously interpreted as being structurally controlled by Quaternary faults (e.g., Legg et al., 2015). In the northwestern Catalina Basin, Kimki Ridge is an elongate, northwest-trending, ∼20-km-long and ∼2-km-wide seafloor ridge system. Kimki Ridge has a maximum relief of ∼400 m, with the ridge peak lying at water depths of ∼900–1000 m. Kimki Ridge is bounded by steep, scarp-like slopes and is also obliquely cut by a sharp lineament that divides the ridge into discrete north and south segments (Fig. 3), and it forms the eastern margin of the ∼10-km-long and ∼5-km-wide Kimki sub-basin (Figs. 2A, 3). Another intrabasin ridge is located downslope of northernmost Santa Catalina Island and ∼3 km southwest of the mapped CF (Fig. 4). This ∼10-km-long ridge has a maximum relief of ∼130 m.

The San Gabriel Channel (Fig. 5), the primary channel system delivering terrestrial sediment to Catalina Basin (Maier et al., 2018), enters Catalina Basin between two bedrock highs at the basin’s southeastern corner (Fig. 5). Maier et al. (2018) describe channels and scours likely associated with the San Gabriel channel-lobe transition zone on the southeastern basin floor (Fig. 5). Sediment supply to Catalina Basin also comes through several canyons and gullies on the steep basin margins, including the ∼1-km-wide Catalina Canyon (Fig. 4) along the northern basin margin, which incises the shelf edge <2 km from the coastline. We also image a number of smaller, ∼100–400-m-wide gullies and smaller canyons along northern Kimki Ridge (Fig. 3) and downslope of each of Santa Barbara, San Clemente, and Santa Catalina Islands (Figs. 2A, 4).

Seafloor backscatter data (Figs. 3B, 4B, and 5B) highlight a number of sedimentary features throughout Catalina Basin, including relatively low backscatter of the San Gabriel Channel as it enters Catalina Basin (Fig. 5B) and relatively high reflectivity at narrower canyon systems (e.g., Catalina Canyon; Fig. 4B). Higher backscatter values are located along ridge systems (e.g., southern Kimki Ridge; Fig. 3B), at basin edges where there are older sediments (e.g., Fig. 4B), and at bedrock highs (e.g., Fig. 5B). Northern Kimki Ridge has three notable bright spots located along the ridge crest (Fig. 3B), which have been sampled and determined to be active methane seeps (Conrad et al., 2018), and a fourth seep was recently discovered on the crest of southern Kimki Ridge (J. Conrad, 2019, personal commun.; Fig. 3B). Another similar bright spot is apparent on the crest of the smaller ridge in the northeastern basin (Figs. 3B, 4B).

The Catalina Basin seafloor is composed of exposed Quaternary sediment with a Holocene drape layer. Uplifted pre-Quaternary sediment and structure crop out at the basin margins; for example, the Kimki Ridge system (Fig. 3) is a fault-bisected anticline composed of Pliocene and older sedimentary rocks (Vedder et al., 1986; Vedder, 1990). Similarly, the ∼10-km-long ridge downslope of Santa Catalina Island (Fig. 4) is composed of Pliocene and older sedimentary deposits (Vedder et al., 1986; Vedder, 1990). Pliocene stratigraphy is commonly also overlain by hemipelagic drape.

Catalina Basin Stratigraphy

The new 2014 and 2016 seismic-reflection surveys (Fig. 2B) reveal basin architecture that we interpret within the general allostratigraphic and seismic facies framework for ICB basins summarized by Teng and Gorsline (1989; Fig. 6). We identify the same units and unconformities in the Catalina Basin, including a notable regional unconformity (Miocene surface, MS; see Fig. 7), which we define as the upper boundary of deformed, tilted, and acoustically chaotic seismic reflections and/or reflection-free zones. The chaotic and semitransparent facies are the acoustic basement here, and the MS unconformity is atop layered, deformed seismic reflections that conform to the underlying basement topography. The MS unconformity commonly truncates the underlying reflections and is likely late Miocene in age based on comparison to ages and seismic facies in adjacent ICB basins (e.g., Teng and Gorsline, 1989; Schindler, 2010). We map another surface similar to the MS unconformity locally through northwestern Catalina Basin (basement or bedrock surface, BB; e.g., Fig. 8). Similar to the MS unconformity, the BB surface caps chaotic and semitransparent seismic facies of the acoustic basement (e.g., Fig. 8). The BB surface may be contemporaneous with the MS unconformity, but we cannot directly link the two surfaces using the available data.

Post–late Miocene strata both downlap onto and onlap the MS unconformity (Fig. 7) and are generally less deformed and more flat lying than sediments below the MS unconformity. These overlying strata fill several local paleo-lows (previous sedimentary depocenters), creating localized sub-basins (Fig. 7). Sub-basins contain localized growth strata in their centers; in other words, with increasing depth, reflection dip increases toward sub-basin edges, layer thicknesses increase toward the sub-basin center, and sediment wedges pinch out at sub-basin edges (Fig. 7).

A relatively transparent package above the MS unconformity is capped by a second notable regional unconformity (Pliocene surface, PS; see Fig. 7). The PS unconformity is overlain by stratified, wedge-shaped strata that downlap onto it. This uppermost stratigraphic unit has variable thickness, generally thickest in the region of the San Gabriel Channel (e.g., Fig. 7) and in northwestern Catalina Basin, and thinnest in the center of Catalina Basin (Fig. 9). The age of the PS unconformity is likely Pliocene based on ages from seafloor samples of the surface where it crops out on the basin margins (Vedder et al., 1986; Vedder, 1990; also see, e.g., Pliocene structure map, Fig. 9). Sub-basins exhibit evidence for growth strata above and below the mapped PS unconformity, and the most recently deposited sediments are parallel, flat lying, and undeformed, except near the CF and SCF (e.g., Fig. 7).

Two-way travel time structure contour and isochron sediment thickness maps were generated for the stratigraphic interval between the MS and PS regional unconformities, as well as between each of the MS and PS unconformities and the seafloor (Fig. 9). The structure contour maps of the MS and PS unconformities reveal the changing geometry of the localized paleo-lows (or depocenters; Fig. 9) through time, which do not mimic the modern basin morphology. Isochron thickness maps indicate areas of thick sediment within these localized sub-basins. The seafloor-Pliocene isochron thickness map shows the most uniform distribution of sediment throughout the basin (Fig. 9), with the thickest sediment in the package located in a persistent sub-basin along southwestern Catalina Basin and near the entrance of the San Gabriel Channel into the basin. Maier et al. (2018) noted that the shallowest depositional lobes of the San Gabriel depositional system appear in this same southwestern region due to a broad topographic low that persists in the basin today.

In Figure 6, we directly compare the sedimentary deposits of Catalina Basin to those in another ICB basin, the Santa Monica Basin. The unconformity between Miocene and Pliocene deposits is similar in both basins, but apart from the deep localized sub-basins in Catalina Basin, Pliocene–Quaternary sediment thickness is generally greater in Santa Monica Basin than in Catalina Basin. This difference is due to the proximity of Santa Monica Basin to active terrestrial sediment sources, leading to relatively thick Quaternary sediment cover there (Romans et al., 2009).

Catalina Basin Faults

San Clemente Fault

Faults in the Catalina Basin generally exhibit linear expressions in the new ICB bathymetric compilation of Dartnell et al. (2017) due to seafloor scarps (Fig. 3A). The SCF extends along the entire ∼90 km length of Catalina Basin with an average strike of ∼327°. For most of this stretch, the SCF is located along the westernmost Catalina Basin floor and characterized by an ∼2–4-m-high, southwest-up scarp on the seafloor. A few kilometers south of the Kimki Ridge system, the SCF makes a broad ∼20° releasing bend and traverses across Kimki Ridge and the flat floor of the Catalina Basin. In the releasing bend area, the sense of offset across the SCF changes to be northeast-side-up, as defined by an ∼2-m-high seafloor scarp.

Bathymetric data show the main trace of the SCF making a broad ∼30° left bend north of the ∼20° releasing right bend in northernmost Catalina Basin. The SCF thus trends toward and connects farther north with the Santa Cruz fault, an ∼80-km-long structure with an average strike of ∼311° that runs along a ridge between Santa Cruz and Santa Monica Basins (Figs. 1, 2B). South of Catalina Basin, the SCF continues for at least another 80 km with an ∼318° strike, nearly parallel with the average Pacific plate motion (321°) through the region (DeMets et al., 2010). The total mapped length of the main strand of the SCF thus exceeds 250 km. Although small (300–400-m-wide) gullies along northern Kimki Ridge appear to have been offset by the SCF, correlative features on either side of the fault cannot be identified with certainty (Fig. 3), and we could not identify any other horizontal SCF piercing points on the seafloor in the new high-resolution seafloor data.

In the subsurface, the SCF exhibits a near-vertical dip with offset seismic reflections to the seafloor. Vertical offset is southwest-up through the southern basin adjacent to San Clemente Island (Figs. 7A, 7C) and northeast-up through northern Catalina Basin (Figs. 10A, 10B). CHIRP data image offsets of the base of the drape layer along the SCF (Figs. 10B, 10C), which was likely deposited starting ca. 15 ka (Normark et al., 2009; Ryan et al., 2012; Maier et al., 2018). The shallowest, 0–200-m-deep seismic reflections are traceable across the SCF, but deeper layers generally cannot be correlated across the fault (e.g., Fig. 10A).

Kimki Fault and Other Splays of the San Clemente Fault

In northern Catalina Basin, the SCF bisects and divides the Kimki Ridge system into north and south ridge segments (Fig. 3). Here, the fault zone becomes more complicated, and we map a number of secondary faults and subparallel SCF splays, including the Kimki fault (KF), a fault originally discovered and informally named by Ford and Normark (1980). The KF runs along and north of the western side of northern Kimki Ridge, and an unnamed splay marks the east side of the southern ridge (Fig. 3). Both the KF and the unnamed splay connect with the SCF at their southern tips (Fig. 3). West of southern Kimki Ridge, we map three additional, subparallel secondary faults on San Clemente Ridge (Fig. 3). These secondary faults feature <2-m-high scarps that are generally smoother than the SCF scarp, but still resolvable in the 25 m high-resolution bathymetry, and do not explicitly connect with the SCF.

At 33 km long, the KF is the third-longest fault in the Catalina Basin (after the SCF and the CF). It strikes subparallel to the SCF through northernmost Catalina Basin with the same average strike of 327°. We consider the KF a splay of the SCF because its southern tip connects with the SCF. Where the KF bounds northern Kimki Ridge on its west side, vertical offset of the KF is northeast-side-up, but the fault transitions to show southwest-side-up north of the Kimki Ridge system through northern Catalina Basin. In contrast to the sharper scarps along the SCF, the KF is characterized by smooth seafloor and an inflection in seafloor slope. The northernmost reach of the KF traverses up a canyon system on the steep slope flanking southern Santa Barbara Island. Here the KF becomes unresolved in existing geophysical data, and may die out as shown in Figure 2B. We were not able to identify piercing points across the KF; several small-scale gullies located west of the KF may cross the KF, but we cannot resolve or correlate the gullies across the fault with the available high-resolution bathymetric data (e.g., Fig. 3).

Like the SCF, MCS data reveal that the KF is a subvertical fault in the shallow subsurface (Fig. 11A), but the offset at the base of the seafloor drape does not appear to be offset in CHIRP data crossing the KF (Figs. 11B, 11C). The SCF splay bounding the eastern edge of the southern ridge in the Kimki Ridge system (Fig. 3) exhibits small offsets in the subsurface that diminish moving north. The secondary faults on San Clemente Ridge (Fig. 3) generally exhibit small normal offsets, and three of these faults offset the seafloor (Figs. 3, 7B).

Catalina Fault

The CF is confined to northeastern Catalina Basin and has a mapped length of ∼63 km. From south to north, the CF bends from a strike of ∼295° to ∼335° and back to ∼310°. It has a 315° average strike as mapped, slightly convergent with Pacific plate motion of 321° (Fig. 4). Southeast of Catalina Canyon, the CF is expressed at the seafloor as an inflection in seabed slope that decreases toward Santa Catalina Island (Fig. 4). Along the southern margin of Santa Catalina Island, we lack high-resolution bathymetry data along much of the steep, nearshore slope; however, the CF may continue along and be represented by the escarpment itself at the base of Santa Catalina Island (Fig. 4), making the total fault length longer than the explicitly mapped 63 km. Ryan et al. (2012) interpreted that the CF does not connect with the San Diego Trough fault southeast of Santa Catalina Island where the San Gabriel Channel enters Catalina Basin, as has been suggested by other previous studies (e.g., Hauksson and Jones, 1988; Legg et al., 2004a, 2004b, 2007), though we also note that Maier et al. (2018) mapped several discontinuous fault structures in this area that may be related to either the Catalina or San Diego Trough fault zone.

In northeastern Catalina Basin, the CF crosses Catalina Canyon, which makes a northwestern bend just south of the CF, but we cannot resolve any clear offsets at the canyon walls (Fig. 4). Just north of Catalina Canyon, the CF exhibits a significant, northeast-side-up scarp with relief of ∼10 m. The height of this scarp lessens moving northward as the CF traverses through Pliocene sediment east of a small ridge and west of the downslope limit of two discrete submarine landslide scarps and debris aprons (Fig. 4). Near these landslides, the CF scarp transitions from northeast-up to southwest-up, and scarp height increases from ∼5 m near the landslides to ∼10 m where the fault exits Catalina Basin to the north (Fig. 4A). North of Catalina Basin, the CF has no obvious geomorphic expression.

In MCS data, the CF exhibits a subvertical dip with some folding and offsets of adjacent seismic reflections from the seafloor to the base of seismic imaging (Fig. 8A). The CF cuts through a sub-basin between basement highs exhibiting wedge-shaped growth strata in Pliocene and older sediments that thicken toward the CF (Figs. 8A, 8B). This sub-basin is similar to the prominent southwestern sub-basin discussed previously (Fig. 7A). The base of the seafloor drape layer at the CF is visibly offset in the CHIRP data (Figs. 8B, 8C), but vertical offset is less than the vertical offset of the drape across the SCF.

While there are no resolvable piercing points along the CF north of Catalina Basin or across Catalina Ridge, there is a subtle offset across the ridge in line with the trend of the mapped CF in Catalina Basin (Fig. 4A). In conjunction with bathymetric data, we use relatively densely spaced (3–4 km) legacy and new high-resolution MCS data to map the CF across Catalina Ridge into Santa Monica Basin (Figs. 4A, 6B, 10A), and note that this is a significant departure from previous mapping (e.g., Legg et al., 2015). Subsurface offset across the CF is similar immediately north and south of Catalina Ridge (Figs. 8A, 10A), but lessens moving northward into Santa Monica Basin (Fig. 6B), where we interpret the northern termination of the CF (Fig. 2B). The southern termination of the CF lacks new definition from MCS data due to absence of nearshore high-resolution geophysical coverage and difficulty of seismic imaging along the steep escarpment at the base of Santa Catalina Island.

Fault Geometry, Seismicity, and Potential-Field Data

Geometric Analysis

New, high-resolution geometric constraints on faults allow us to evaluate the relationship between plate motion and fault geometry. We plot the difference between the fault strike of the SCF, KF, and CF and Pacific–North America plate motion from MORVEL (DeMets et al., 2010) at the midpoints of discrete, 5 km segments along the faults (Fig. 12). The Santa Cruz fault and secondary faults local to Catalina Basin (Figs. 2B, 3, 4) are included in the plot and considered to be a part of their respective primary fault zones (the SCF, KF, or CF). South of ∼33.2°N, the SCF is approximately parallel with average Pacific–North America plate motion of ∼321° (DeMets et al., 2010; Fig. 12). Between ∼33.2°N and ∼33.6°N, the SCF takes a broad ∼20° right bend through Catalina Basin, diverging as much as 25° from Pacific plate motion (Fig. 12). Through this same region, the CF is as much as 25° convergent with Pacific plate motion and the KF strike varies from 16° convergent to 12° divergent (Fig. 12). We also note that the CF and the KF exist at the same latitudes as a broad right step in the SCF, though the average convergence angle of these three fault zones is still in line with Pacific plate motion (Fig. 12). North of 33.6°N, the CF and KF reach their terminations and the SCF becomes convergent (∼10°–15°) relative to Pacific plate motion, with a strike closer to ∼307° as it approaches the Channel Islands in the southern Western Transverse Ranges province (Fig. 12).

Seismicity Analysis

Using the Hauksson et al. (2012) earthquake catalog, we observe quantifiably heightened seismicity on the SCF along the northern and southern thirds of the mapped SCF (Fig. 2B; Fig. S1 in the Supplemental Material1). The earthquake catalog includes the 1981 Mw 6.0 Santa Barbara Island earthquake (e.g., Astiz and Shearer, 2000), which accounts for much of the seismicity cluster along the Santa Cruz fault (Fig. 2B). Although a large earthquake also occurred along the southern segment of the SCF (the 1951 Ms 5.9 San Clemente Island earthquake, one of the largest offshore earthquakes recorded in the ICB; e.g., Astiz and Shearer, 2000), the earthquake catalog shown here does not include the 1951 mainshock or its aftershocks within 30 yr of the event.

Using the Hauksson et al. (2012) catalog, we also quantify the cumulative moment release along the main trace of the SCF between 1981 and 2011 (see Fig. S1 in the Supplemental Material [footnote 1] for detailed methods, figures, and analysis). During that 30 yr period, the Hauksson et al. (2012) catalog indicates several orders of magnitude less seismic moment release over ∼50 km of the central SCF through northern Catalina Basin between San Clemente Island and Santa Barbara Island (Fig. 2B; Fig. S1). The low-seismicity area also correlates with the broad 20° right bend in the SCF and is at similar latitudes to the CF, KF, splays, and secondary faults of the SCF. Although we do not quantify seismicity along the CF, KF, and other secondary faults, we note there is qualitatively very little to no seismicity along these fault zones, with the exception of a cluster of events along the central CF adjacent to north-central Santa Catalina Island (Fig. 2B; Fig. S1). Seismicity decreases outward from this central cluster along the CF to the north and south, with only around three events located along the escarpment offshore of the southern margin of Santa Catalina Island (Fig. 2B; Fig. S1).

Focal mechanisms determined for some earthquakes in this region (Yang et al., 2012) are generally the most uncertain offshore—most events are of lowest “D” quality, meaning they have a large azimuthal gap between observations or very few VP/VS amplitude ratios—but can still provide some important insights for kinematic interpretations. SCF focal mechanisms are generally strike-slip, particularly in the areas with higher seismicity rates along the southern and northern fault (Fig. S1 [footnote 1]). Reported focal mechanisms along the CF all exhibit thrust mechanisms (Fig. S1).

Potential-Field Data

Gridded magnetics (Bankey et al., 2002; Fig. 13A) and gravity (Beyer, 1980; Fig. 13B) data sets in the offshore California borderland have coarse ∼1 km resolution, but illuminate regional patterns in crustal structure through our study area (Fig. 13). Notably, both the gravity and magnetics data show broad, regional lows along and east of the SCF; this area has been interpreted as thinned crust formed during extension and exhumation in the late Miocene (Bohannon and Geist, 1998; ten Brink et al., 2000; Miller, 2002). The main trace of the SCF, and even its splay, the KF, traverse across lows in the gravity data, which correspond with the weaker, thinned crust (Fig. 13B). West of San Clemente Island in the Outer Continental Borderland, gravity values are generally higher but regionally smooth, whereas magnetics data show localized, concentrated highs (asperities) indicating relatively hard crustal blocks with more abundant magnetic minerals. Fault pathways also correlate with magnetic lows, and the SCF appears to divert around a prominent magnetic asperity west of the northernmost Catalina Basin (Fig. 13A).

Here, we utilize the above observations to address the following hypotheses: (1) the SCF accommodates significant (as much as 4–6 mm/yr) right-lateral slip, and the CF accommodates primarily convergent stress; (2) modern physiography does not necessarily indicate where active faults exist, and Quaternary faults can overprint pre-existing crustal structures; and (3) pre-existing crustal fabric is an important control on Quaternary fault geometry, with pre-existing crustal blocks being more important than pre-existing faults. We begin by examining the geometry and structure of shallow active faults and then discuss their relationship to crustal-scale structures and development since the late Miocene.

Importantly, throughout our interpretations, we differentiate between different types of crustal fabric, with “crustal fabric” being the umbrella term for all crustal features. We define crustal fabric as including two broad categories of crustal features: “crustal blocks” and “structures.” Crustal blocks can be differentiated from each other using physical characteristics such as thickness, rock type, and density of the crystalline crustal rock underlying sediment; potential-field data (gravity, magnetics) are commonly diagnostic of these properties, and can thus differentiate between crustal blocks. Crustal structures, the other broad category under crustal fabric, include faults, folds, and other strain structures. All crustal fabrics can be either “pre-existing” (in our study, pre-Quaternary or “inherited”) or “active” (Quaternary, unless otherwise indicated). For the time frame described by our study, crustal blocks are always pre-existing, whereas crustal structures can be pre-existing, active, or both, if the structure has persisted since the pre-Quaternary or has been reactivated in the Quaternary. On the surface of the crust, it is important to be able to differentiate between inherited and active structures, both of which are manifested in the physiography and tectonic geomorphology of the seafloor.

Active Deformation and Hazards

San Clemente Fault

Using direct seafloor and subsurface observations of the SCF, we consistently observe offset of the seafloor and Holocene drape layer (Maier et al., 2018) along the SCF (e.g., Fig. 10), and thus, we interpret the SCF as Holocene active. Moreover, the SCF has both the sharpest and largest seafloor offset (Figs. 3, 7, 10) and the most advanced subsurface deformation (Figs. 7C, 8A, 10) of the faults we map in the Catalina Basin region. Subsurface deformation, including difficulty correlating seismic reflections across the SCF, especially deeper reflections (Figs. 8A, 10A), indicates significant lateral slip and/or a long-lived fault. These subsurface observations, along with the continuity, length (>250 km), and linear seafloor expression of the SCF, are all consistent with significant offsets and Holocene slip on the SCF. The SCF bisects and offsets the Kimki Ridge system, which is composed of Pliocene and older deposits (Vedder et al., 1986; Vedder, 1990; Figs. 3, 11). This relationship suggests that Kimki Ridge is a pre-Quaternary structure, that folding along the ridge is not active (which is also supported by MCS data; Fig. 8A), and that significant lateral SCF slip is younger than Kimki Ridge.

Mapped splays and secondary faults of the SCF generally appear to be Quaternary faults with smaller offsets than the SCF. Like the SCF, the clear seabed expression of the KF (Fig. 3A) suggests Quaternary activity, though ambiguous subsurface offset of the hemipelagic drape layer (Figs. 11B, 11C) suggests that the KF is accommodating much smaller amounts or slower rates of deformation, and perhaps has not been deforming in the Holocene. Other faults within the SCF zone, including the splay bounding the east flank of southern Kimki Ridge and subparallel, subvertical faults on the San Clemente Ridge (Figs. 3, 7B) are short and discontinuous, hence considered inactive or accommodating small, localized offsets.

Our seafloor mapping shows clearly that the SCF connects with the Santa Cruz fault to the north. This is a substantial departure from some published mapping of the SCF, particularly through the northern Catalina Basin (e.g., Fig. 2B), where previous mapping has commonly placed the SCF ∼10 km west of our interpretation and/or as ending in the northern Catalina Basin (e.g., Ford and Normark, 1980; Vedder et al., 1986; U.S. Geological Survey and California Geological Survey, Quaternary Fault and Fold Database for the United States,; additionally, the Santa Cruz fault has previously been interpreted as connecting with the Catalina Ridge fault, with the combined structure previously referred to as the Santa Cruz–Catalina Ridge fault zone (e.g., Legg et al., 2004a). Recent seismicity along the Santa Cruz fault, largely a result of the 1981 Mw 6.0 Santa Barbara Island earthquake (Astiz and Shearer, 2000; Hauksson et al., 2012; Fig. 2B; Fig. S1 [footnote 1]), indicates that the fault is probably active, and focal mechanisms (Yang et al., 2012) are consistent with a strike-slip interpretation, making the kinematics likely similar between the SCF and Santa Cruz fault zones. The SCF and Santa Cruz fault have not previously been connected in fault nomenclature, but we suggest that the combined fault zone might be more accurately described as the Santa Cruz–San Clemente fault zone (SC-SCF; shown in Fig. 12), given the clear connection between the fault zones evident in our data.

The increased length (>250 km) of a combined SC-SCF has implications for earthquake hazards. The SCF has been considered capable of rupturing in infrequent >M7 events (Legg and Borrero, 2001), and the Santa Cruz fault likely ruptured in the 1981 Mw 6.0 Santa Barbara Island earthquake (Astiz and Shearer, 2000; Fig. 2B; Fig. S1 [footnote 1]). A 250 km fault length is capable of generating earthquakes up to magnitude ∼8 (assuming a 10-km-wide rupture patch and 15–20 m of slip), the theoretical maximum for terrestrial strike-slip earthquakes (Wells and Coppersmith, 1994). We speculate, however, that a magnitude 8 event on the SCF is unlikely given the extended crust in the ICB, which has resulted in relatively high heat flow, and thus, potentially a relatively shallow brittle-ductile transition (Lee and Henyey, 1975; ten Brink et al., 2000); relocated earthquake depths of generally <10 km along the Santa Cruz fault and SCF support this idea (Hauksson et al., 2012; Figs. 2B, 10). A number of restraining and releasing bends along the ∼250 km fault (Fig. 12; e.g., Legg et al., 1999, 2007; Legg and Borrero, 2001) could also potentially inhibit rupture propagation (King and Nábělek, 1985; Wesnousky, 2006). However, we also note that strike-slip earthquakes can rupture through large bends (e.g., the 1857 M7.9 Fort Tejon earthquake in south-central California; Sieh, 1978), and that large, complex ruptures have been known to occur elsewhere along the California plate boundary (e.g., the 1992 M7.3 Landers earthquake in southern California; Hauksson et al., 2012) and along other strike-slip plate boundaries (e.g., the recent Kaikoura event in New Zealand, Hollingsworth et al., 2017; the Haida Gwaii event along the Queen Charlotte fault offshore of British Columbia, Canada, Lay et al., 2013).

Fault geometry also illuminates the kinematics of Catalina Basin fault zones (Fig. 12). The strike of the SCF south of ∼33.2°N is consistent with an interpretation of dominantly right-lateral strike-slip fault (Fig. 12), and deformation and seismicity along the SCF are consistent with strike-slip behavior (e.g., Fig. 7C; Fig. S1 [footnote 1]). Between ∼33.2°N and ∼33.5°N, a broad right step in the SCF results in an ∼50-km-long releasing bend along the fault (Fig. 12), though limited evidence for Quaternary subsidence or normal faulting exists in the Catalina Basin region. Late Quaternary subsidence of Pilgrim Banks just north of Catalina Basin on the west side of the Santa Cruz fault (Castillo et al., 2018) is potentially consistent with subsidence along the SCF releasing bend, and the sub-basin we map just south of the Kimki Ridge system (Fig. 7A) has localized growth strata that might suggest local sub-basin subsidence and/or uplift along the San Clemente Ridge through the Pliocene–Quaternary. However, tectonic geomorphology near the SCF releasing bend also seems to indicate post-Pliocene transpression (Kimki Ridge), and the only normal faults we image in or on the margins of Catalina Basin are the short Quaternary faults on San Clemente Ridge (Fig. 7B), which could be nontectonic gravitational faults. Because pervasive transtensional deformation associated with the SCF releasing bend is limited, our observations suggest that the right bend along the SCF may be a relatively recent (Quaternary) geometric adjustment.

Catalina Fault

There is clear offset of the Holocene drape layer along the CF (e.g., Fig. 8), and we thus consider it a Holocene-active fault (Maier et al., 2018). Our kinematic analysis based on fault geometry supports convergent deformation across the CF, particularly between ∼33.2°N and ∼33.5°N along the northeastern margin of Catalina Basin (Fig. 12) where the CF becomes as much as 25° convergent with Pacific plate motion. Strain along the CF indicates the influence of both convergence and transform motion. Gentle folding in the shallow subsurface is consistent with a component of convergence (Fig. 8), and strata diverging into the CF (Fig. 8A) could indicate growth along the fault or basin growth offset by the fault. Seafloor scarps as high as 10 m (Fig. 4A) with changing sense of offset along strike are consistent with lateral motion; in one spot, the fault scarp relief appears to have limited the downslope extent of a submarine landslide debris apron (Fig. 4). We do not observe any piercing points along the CF, but its proximity to insular sediment routed through Catalina Canyon may have resulted in burial or erosion of evidence of transform motion. Relatively small subsurface vertical offsets and correlatable unconformities across the CF in MCS data (Fig. 8A) indicate (1) relatively small amounts of cumulative compressional strain along the CF; (2) a relatively young or short-lived fault, which would not exhibit large amounts of cumulative strain; (3) both (1) and (2); or (4) predominantly strike-slip deformation, which seems unlikely given the convergent CF geometry (Fig. 12). Secondary faults in the CF zone (Fig. 4), like the secondary faults in the SCF zone (Fig. 3), are short and discontinuous and exhibit little to no offset of the hemipelagic drape layer; therefore, the most recent slip on these minor faults is likely pre-Holocene.

The CF exhibits notable changes from previously inferred geometry. We use the trend of the CF as observed in merged bathymetry and in the 2014–2016 high-resolution MCS data to map the active CF across Catalina Ridge into southern Santa Monica Basin, where we image the fault in legacy industry seismic-reflection data (Figs. 2B, 6B, 10A, 12). The surficial bedrock geology of the submerged Catalina Ridge is Miocene in age (Vedder et al., 1986; Vedder, 1990), suggesting a Miocene (or later) generation of Catalina Ridge followed by initiation of the CF. Due to the cross-cutting relationship of the Holocene CF traversing across Catalina Ridge, we suggest that Catalina Ridge is an inherited structure formed during an earlier episode of deformation that is no longer active, although it is still possible that it is an early Quaternary structure or at least may become reactivated. Our mapping is also inconsistent with the interpretation of the Catalina–Catalina Ridge fault zone as a right stepover, as has been suggested previously (Legg et al., 2007). Additionally, the CF as mapped in our study does not appear to connect with the Santa Cruz fault as has commonly been previously interpreted (e.g., Legg et al., 2004a). Because of the likely kinematic and age differences between the Santa Cruz and Catalina Ridge faults, we argue against connecting these two faults in nomenclature (i.e., Santa Cruz–Catalina Ridge fault) going forward.

The kinematic activity along the southern end of the CF cannot be better characterized with our new data due to lack of coverage along the escarpment (Fig. 2B). Thus, our observations could be consistent with either (1) an active southern CF along the escarpment, consistent with demonstrable Holocene activity observed along its western end; or (2) an inactive fault, consistent with waning seismicity along the escarpment (Fig. 2B). An inactive (early Quaternary or pre-Quaternary) fault would also be consistent with the observations of Normark et al. (2004b) and Ryan et al. (2012) that the fault does not deform the youngest sediments or connect with the San Diego Trough fault. Additionally, sparker seismic reflection data collected in 2008 (Sliter et al., 2017) show onlap onto Catalina Ridge and no deformation of the upper ∼60 m of sediment. A sedimentation rate of ∼4 cm/k.y. from radiocarbon dating in Catalina Basin (McGann and Conrad, 2018) suggests that the 60 m sequence of undeformed sediment could represent as much as 1.5 m.y. of fault inactivity.

Late Quaternary asymmetric subsidence of Santa Catalina Island has been attributed to paired normal motion across the CF on the southern flank of Santa Catalina Island and across the San Pedro Basin fault (Fig. 6) on the island’s northern margin (Castillo et al., 2018). Our imaging instead suggests convergence or transpression across the CF, requiring a different mechanism for Santa Catalina Island subsidence. We suggest that the relatively thick Pliocene–Quaternary sediment in Santa Monica Basin (Fig. 6), along with continuous deposition in Santa Monica Basin through the Quaternary (Romans et al., 2009), might have led to flexural subsidence in the Santa Catalina Island region, accounting for both the observed 0.08–0.27 mm/yr subsidence since at least 1.15 Ma and the 1.5° northward asymmetric tilt of Santa Catalina Island.

Slip Rates

Despite the generally clear expression of the SCF, KF, and CF in high-resolution bathymetric data and numerous fault crossings of seabed landforms (e.g., ridges, Fig. 3; gullies and canyons, Fig. 4), we do not observe any clearly offset piercing points that can be used to verify lateral slip rates. It is tempting to reconstruct the offset Kimki Ridge along the SCF (e.g., Fig. 3), but the precise age of Kimki Ridge is uncertain, and the Pliocene deposits of the northern Kimki Ridge and the Miocene rocks of the southern ridge (Vedder et al., 1986; Vedder, 1990; Fig. 3) also suggest that this is not a valid reconstruction. Similarly, gullies along the eastern flank of northern Kimki Ridge (Fig. 3) do not correlate across the SCF and thus are not viable piercing points.

Previous studies have estimated SCF slip rates as high as ∼4–6 mm/yr (Legg, 1985, 1991a, 2005; Larson, 1993; Bennett et al., 1996; Humphreys and Weldon, 1994; Goldfinger et al., 2000). Our work supports a maximum of ∼3.6 mm/yr of right-lateral strain accommodated within the Catalina Basin fault zones, with the majority of this slip likely taken up by the SCF, because it exhibits more cumulative deformation than the CF on the seabed and in the subsurface. GPS data help to verify hypothesized slip rates in the absence of viable piercing points. Offshore GPS data from the Southern California Crustal Motion Map, version 4.0 (CMM4; Shen et al., 2011; Tables S1 and S2 [footnote 1]; Fig. 12), show that between 2.2 mm/yr and 3.6 mm/yr of differential lateral slip is accommodated between San Clemente Island and Santa Catalina Island (Tables S1 and S2; Fig. 12). The aggregate of geologic ICB slip rates suggests a similar slip deficit of 2.7–3.7 mm/yr accommodated along Catalina Basin faults (see text accompanying Tables S1 and S2 for analysis). Thus, results from previous studies and the morphology of the SCF leads us to prefer (1) a slip rate on the higher end of the range suggested by GPS results (∼3.6 mm/yr) and (2) that this slip is largely accommodated along the SCF in Catalina Basin. We note that slip rates for the SCF and all other faults can vary along their length (e.g., the Newport-Inglewood fault; Fischer and Mills, 1991; Grant et al., 1997), and that the rates we report here are representative maximum slip rates throughout the Catalina Basin region.

Kinematics (Fig. 12) and subsurface deformation (Fig. 8) suggest that the CF is accommodating convergent or transpressive stress across the northern Catalina Basin. The CF may be accommodating a small part of the ∼3.6 mm/yr of right-lateral motion taken up across Catalina Basin, but the geometry and subsurface deformation along the CF imply that the CF is a small part of the lateral slip budget and that it is taking up primarily convergence, so we prefer lateral slip rates smaller than the 2.5 mm/yr found by Chaytor et al. (2008) for the CF. The convergent component of slip along the CF could potentially be as high as 1.5 mm/yr where the restraining angle reaches a peak of 25°, if we assume a high-end right-lateral slip rate of 3.6 mm/yr along Catalina Basin faults (namely the SCF).


While a 30 yr window of seismicity (Hauksson et al., 2012) is not indicative of long-term fault behavior, it can tell us where faults are actively deforming and provide insights into crustal rheology. Qualitatively, recent seismicity (Hauksson et al., 2012; Yang et al., 2012) has occurred primarily along the northern and southern SCF (Fig. 2B; Fig. S1 [footnote 1]). The 1981 Mw 6.0 Santa Barbara Island earthquake, which is included in the earthquake catalog we use here, likely occurred along the Santa Cruz fault and thus partly accounts for elevated seismicity in that region. In general, high seismicity rates along the northern and southern SCF are consistent with our subsurface observations that the SCF is actively deforming and that deformation is accommodated across a relatively narrow deformation zone in these areas. Fewer events have occurred along the CF than along the SCF, but their presence supports our interpretation of active deformation along this fault zone as well.

One of the more striking observations of seismicity distribution is the quantifiable lack of seismicity through the central segment of the SCF between northern San Clemente Island and just south of Santa Barbara Island, where the SCF has a broad releasing bend (Fig. 2B; Fig. S1 [footnote 1]). We speculate that the lower seismicity rates could imply more diffuse or distributed slip distribution here, which could potentially be supported by the existence of multiple fault zones (KF, CF) and thus a wider deformation zone at these latitudes. At the latitudes of the SCF releasing bend, we also note an apparent shoreward shift in ICB seismicity indicated by heightened earthquake activity along the San Diego Trough, Palos Verdes, and Newport-Inglewood fault zones, possibly suggesting more slip accommodation closer to shore here as well. While the SCF releasing bend may exhibit a more widespread strain distribution due to increased obliquity with plate motion, it is also possible that the reduced seismicity through the northern Catalina Basin is caused by a seismic gap along the fault zone (Fig. S1), and that the SCF is locked and poised to rupture in a relatively large (>M5) event here; the seismicity gap could also indicate fault creep, resulting in similar net slip but lower seismicity rates.

Earthquake depth distribution along the SCF and CF is consistent with deformation of extended continental crust, with most events occurring at ∼10 km or shallower depths (Fig. 2B; Fig. S1 [footnote 1]). Some slightly deeper earthquakes (5–15 km) along the northern and southern SCF and the CF (Fig. 2B; Fig. S1) support crustal-scale deformation of these fault zones. Focal mechanism observations (Fig. S1; Yang et al., 2012) are consistent with our interpretations of subsurface deformation (e.g., Figs. 8, 10) and kinematics (Fig. 12) along these fault zones, supporting dominantly strike-slip motion along the combined SC-SCF and active convergent thrusting deformation along the CF. Several focal mechanisms potentially located along the KF indicate normal offsets (Fig. S1), which may reflect transtensional strain resulting from the right bend in the SCF being accommodated partly by the KF.

Fault Generation and Basin Evolution

Our results indicate that inherited crustal fabric in the ICB is an important control on fault geometry, that crustal-scale blocks are the most important influence on evolving fault systems, and that pre-existing crustal structures can be overprinted by Quaternary faults. The most important supporting observations are that several demonstrably active faults cut across pre-Quaternary crustal structures—notably, the SCF bisects Kimki Ridge (Fig. 3) and the CF traverses obliquely across Catalina Ridge (Figs. 2B, 10A, 12)—suggesting that evolving faults can cut pre-existing crustal structures in some cases. Potential-field data (Fig. 13) provide some important insight into why this might be occurring in the Catalina Basin region. The gravity and magnetics data (Fig. 13) illuminate physical properties of the deeper crust (i.e., crustal blocks), showing strong correlations between lows in gravity and magnetics data (indicating weak, thin, extended crust) and mapped fault geometry, particularly SCF geometry. This correlation suggests that crustal blocks may be a more important control on fault geometry than other inherited crustal structures, perhaps especially for crustal-scale faults like the SCF and CF. Our study thus supports crustal blocks as “essential” features, and crustal structures like faults and fault-controlled ridges (including the Kimki and Catalina Ridges) as “incidental” structures as described by Christie-Blick and Biddle (1985); in other words, we find that crustal blocks have had more influence on strike-slip fault deformation in the Catalina Basin region than inherited faults and ridges. While we prefer the interpretation that crustal blocks have influenced the geometry of faults, it is also possible that the translation of crustal blocks along the fault strands (perhaps up to tens of kilometers on the SCF alone) has led to the configuration of crustal blocks that we observe today.

If crustal blocks do in fact have some control on fault geometry, it is possible that the SCF has undergone geometric modifications over time in response to changes in the regional stress field or the positions of essential crustal fabrics as they translate along faults. For example, we interpret a prominent magnetic asperity located just northwest of Catalina Basin (Fig. 13A) as a relatively hard crustal block that may have affected SCF fault geometry. We speculate that the geometric releasing bend in the SCF through northern Catalina Basin may be a “sidewall ripout” (as described by Swanson, 1989, 2005, and Mann, 2007), forced by the presence of the hard crustal block (Fig. 13A) as it translated north into this position. If this is the case, we also speculate that it is possible that the KF once accommodated more slip in the SCF zone and was perhaps abandoned once the asperity had translated far enough north to inhibit or deoptimize slip along the KF. Castillo et al. (2018) suggested late Quaternary abandonment of the KF strand, and Conrad et al. (2018) similarly suggested that the KF was active as a restraining bend in the early Pleistocene and involved in the formation of the Kimki Ridge anticline, possibly also accommodating some amount of strike-slip motion that later transferred over to the SCF. The interpretation that the KF was a paleo-SCF and abandoned in the Quaternary is consistent with our observations of waning deformation in the subsurface along the KF (Figs. 11B, 11C). Consequences of the SCF diverting around this crustal block may also include the bisection of the Kimki Ridge system by the SCF and/or generation of secondary fault strands (e.g., CF) in order to best accommodate plate motion through the region containing the forced releasing bend.

With a better understanding of active faults and their relationships to pre-existing topography, we can now revisit our observations of crustal structure and seismic stratigraphy to piece together a loosely defined picture of Catalina Basin evolution and the role of faults and other structures in basin formation through time (Figs. 14, 15). On a crustal scale, tilted blocks evident in legacy seismic-reflection data (Fig. 14) suggest that block rotation, a basin-and-range–style extensional basin growth mechanism, occurred in the ICB. This block rotation (which has been suggested in previous studies; e.g., Legg and Kamerling, 2003) could be responsible for basin opening and the horst-and-graben–style morphology we observe in the ICB today, including steeply sloped, asymmetric islands and the ICB basins themselves. However, there is no current consensus on exactly how the ICB basins developed; other hypotheses include nonuniform crustal thinning (e.g., Bohannon et al., 2004) and basin formation along local strike-slip releasing bends (e.g., Legg et al., 2007). One possibility is that Catalina Basin itself is a “lazy-Z” basin (Mann, 2007) along the broad SCF releasing bend in northern Catalina Basin (Fig. 12). Legg et al. (2007) also noticed the SCF releasing bend and determined that northernmost Catalina Basin was formed as a pull-apart basin along the bend.

New observations of seismic stratigraphy and subsurface deformation help illuminate the viability of the block-rotation and lazy-Z models for Catalina Basin opening. Our Pliocene-seafloor isochron map shows fairly uniform thickness in Catalina Basin (Fig. 9), suggesting that Catalina Basin resembled today’s physiography by the end of the Pliocene. In contrast, the MS unconformity and Miocene-Pliocene and Miocene-seafloor isochron maps (Fig. 9) indicate local sub-basins and depocenters that are much smaller than Catalina Basin and do not match its modern physiography. We thus interpret two phases of Catalina Basin growth and infill, starting with sub-basin development in the Miocene, followed by broader basin development in the Miocene–Pliocene, finally resulting in the Catalina Basin geometry we observe today. The pattern of basin growth (Fig. 9) is not consistent with lazy-Z basin development (Mann, 2007), and we see little to no evidence for Quaternary transtension (e.g., subsidence, normal faults) along the SCF releasing bend segment (Figs. 8A, 10A). A lack of tensile strain might suggest that the SCF releasing bend is a more recent tectonic adjustment and was not involved in Catalina Basin opening in the late Miocene–Pliocene, an idea that is also consistent with the previously discussed sidewall-ripout hypothesis and Quaternary abandonment of the KF.

Although complex fault configuration and stress environments have likely caused tectonic regimes ranging from convergent to divergent to exist in the ICB simultaneously (Bohannon et al., 2004), we prefer a block-rotation mechanism for ICB basin opening to support observations of new (e.g., Figs. 7A, 8A, 10A) and legacy (Fig. 14) geophysical data. The ICB basins may have started to open in the late Miocene–early Pliocene based on general consensus that ICB kinematic history includes a period of Miocene–Pliocene extension or transtension (Crowell, 1974; Legg, 1991b; Crouch and Suppe, 1993; Bohannon and Geist, 1998; ten Brink et al., 2000; Legg and Kamerling, 2003; Sorlien et al., 2015; DeMets and Merkouriev, 2016), and that uplift of Santa Catalina Island began by earliest Pliocene (Castillo et al., 2018). The steep island topography we observe today (i.e., San Clemente Island, San Clemente Ridge, Santa Catalina Island, and Catalina Ridge) was thus likely largely generated in the Miocene–Pliocene via normal faulting (Fig. 15) along structures like the escarpment south of Santa Catalina Island. Some topography may have been preserved from earlier in the Miocene during regional extension and volcanism in the ICB (Crouch and Suppe, 1993; ten Brink et al., 2000; Miller, 2002). For example, Emery Knoll probably represents an eroded and/or subsided volcanic remnant (comparable to Santa Barbara Island to the north) given its volcanic composition (Vedder et al., 1986) and its high acoustic reflectivity, which has been noted previously (e.g., Ridgway and Zumberge, 2002) and is observable in our 2014–2016 MCS data.

Growth of Catalina Basin began with the sub-basins now evident in MCS data (Figs. 7A, 15) in the late Miocene, corresponding with regional sea-level rise and our mapped MS unconformity (e.g., Fig. 7A). Basin development may have continued into the Pliocene, given that subsurface structure and isochron thickness maps (Fig. 9) suggest that Catalina Basin existed in something close to its current form by the end of the Pliocene. Localized growth strata in the centers of the sub-basins persist into Pliocene and possibly Quaternary marine sediments (Figs. 7A, 15), indicating that sedimentation rates did not quite keep up with tectonic growth during this time frame. Beginning in the late Pliocene (ca. 3 Ma), deposition of the San Gabriel Fan accelerated due to opening of the southeastern basin along the San Diego Trough fault (Maier et al., 2018), allowing the San Gabriel Channel to enter Catalina Basin. This event likely correlates with our mapped PS unconformity and the downlapping San Gabriel Fan deposits above it. Late Pliocene–Quaternary regional transpression (Luyendyk et al., 1980; Ingersoll and Rumelhart, 1999; ten Brink et al., 2000; DeMets and Merkouriev, 2016) may have led to uplifted Pliocene sediment at basin margins (e.g., Fig. 8A), as well as a number of local popup structures and folds like Kimki Ridge (Fig. 15), which fits with the proposed Pleistocene development of Kimki Ridge suggested by Conrad et al. (2018). Today, Quaternary and Holocene strike-slip and transpressional structures (SCF, CF) cut across and overprint Miocene–Pleistocene pre-existing structures (Fig. 15).

Using observations of new, high-resolution geophysical data in conjunction with crustal-scale data sets in the Catalina Basin region, we put forth the following conclusions about fault geometry, structure, and evolution:

  1. We map Holocene-active and Quaternary-active traces of the SCF and CF. Based on fault geometry, subsurface deformation, and seismicity patterns, we interpret the SCF to be the primary active fault in the Catalina Basin region, likely a crustal-scale feature accommodating the majority of dextral strike-slip offset (as much as 3.6 mm/yr) in this part of the ICB. The CF is likely accommodating smaller amounts (<1.5 mm/yr) of oblique transpression or convergence.

  2. The SCF connects to the north with the Santa Cruz fault, and the total mapped fault length is >250 km. We map ∼60 km of the active CF cutting across Catalina Ridge to where it terminates in western Santa Monica Basin, a significant departure from previous mapping.

  3. A 30 yr catalog of seismicity in southern California indicates minimal cumulative moment release along an ∼50-km-long SCF releasing bend between northern San Clemente Island and Santa Barbara Island. The relative lack of seismicity may support diffuse slip distributed across a number of secondary faults through this region, a creeping section along this part of the SCF, or a seismic gap and the site of a future earthquake.

  4. Quaternary high-angle faults (namely the SCF and CF) cut across older, Miocene to Pleistocene structures (e.g., Catalina Ridge, Kimki Ridge), indicating that modern physiography does not necessarily reflect active faulting, and that Quaternary faulting overprints pre-existing crustal structures in some cases.

  5. The SCF main trace follows low magnetic and gravity anomalies, and it may preferentially occupy this pathway due to extended, thinner, and/or weaker crust. Likewise, a large right bend in the SCF in the northern Catalina Basin may have formed around a hard crustal block as represented by a substantial magnetic asperity; the releasing bend may also be responsible for the formation of, and/or distribution of, strain across other faults like the CF and KF.

  6. Inherited crustal fabric is an important control on Quaternary fault geometry, although pre-existing crustal blocks are a more important influence on fault geometry than inherited fault structures in this region.

  7. Seismic stratigraphy of Catalina Basin fill, relationships between inherited and active structures, and regional age constraints suggest several stages of basin formation and deformation: (1) late Miocene relative sea-level rise and the onset of marine sedimentary deposition in local deeps formed due to tectonic subsidence and regional extension, (2) early to mid-Pliocene continuing extension and block tilting leading to the basin morphology we observe today, (3) late Pliocene and Pleistocene transpression leading to localized popup structures and continued ridge uplift, and (4) Quaternary strike-slip overprinting.

  8. Our results emphasize the importance of the acquisition and analysis of high-resolution geophysical data in order to identify active deformation structures, contribute to earthquake and hazard analysis, and accurately interpret tectonic geomorphology.

We thank all the scientists and crew who assisted with the 2014 R/V Robert Gordon Sproul survey and the 2016 R/V Thomas Thompson survey for their help with data collection. We specifically thank Alicia Balster-Gee, Pat Hart, and Ray Sliter for their assistance with data processing and dissemination. Thorough reviews from Sam Johnson, Craig Nicholson, an anonymous reviewer, and the Geosphere Associate Editor helped greatly clarify and improve the paper—thank you. Additionally, thanks to John Barron, Scott Bennett, Jayne Bormann, Eileen Evans, Vicki Langenheim, Mark Legg, Tom Parsons, and David Walton for helpful scientific discussion, support, and advice. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

1Supplemental Material. Detailed methods, figures, and analysis. Please visit to access the supplemental material, and contact with any questions.
Science Editor: Andrea Hampel
Associate Editor: James A. Spotila
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.