Evolution of the Rodgers Creek–Maacama right-lateral fault system and associated basins east of the northward-migrating Mendocino Triple Junction, northern California

The transform boundary between the Pacific and North American plates in northern California (United States) is a wide zone that reflects eastward migration into the North American plate and lengthening since the late Tertiary (Fig. 1). East of the San Andreas fault (the western boundary of the transform margin) and south of the subducting Gorda–Juan de Fuca plate, this wide transform boundary is composed of mixed structural domains dominated in places by active extensional right-lateral faults associated with releasing bends and strike-slip basins. The Rodgers Creek–Maacama fault system is one such domain of extensional right-lateral faults and releasing bend basins, though the long-term history of faulting in the area appears to have included significant compression. Elsewhere, the transform boundary zone clearly includes mixed compressional and extensional right-lateral faulting. The Bartlett Springs fault zone west of the Sacramento Valley, for example (McLaughlin et al., 1990), is predominantly a steeply east dipping transpressional fault zone that includes right-stepped strike-slip basins (such as the Covelo and Lake Pillsbury basins) along its length. Clear Lake basin, another complex extensional strike-slip basin (Hearn et al., 1988), is bounded by northwest-trending faults that have pre-basin compressional strike-slip histories (Fig. 1).

The mixed histories of transtension and transpression associated with the wide transform boundary east of the San Andreas fault are the consequence of processes operating along the Pacific, Gorda–Juan de Fuca, and North American plate boundaries since the Late Miocene (ca. 10 Ma), and in some instances since much earlier in the Tertiary. These processes include northward-migrating slab window–related volcanism associated with migration of the Mendocino Triple Junction (Dickinson and Snyder, 1979; Fox et al., 1985; Stanley, 1987; McLaughlin et al., 1994, 1996; Graymer et al., 2002); large-scale block rotations and plate motions (Argus and Gordon, 2001; Wells and Simpson, 2001); and northward-migrating restraining and releasing bends in the northern San Andreas fault (Fox, 1976; Wakabayashi et al., 2004; Wilson et al., 2005). Other processes that may have indirectly influenced the long-term evolution of the transform boundary include partial coupling between the North American and Gorda–Juan de Fuca plates and the reactivation of structural wedge–related thrust faults separating the Mesozoic basements of the Coast Ranges and northern Sacramento Valley (Berry, 1973; Wentworth et al., 1984; Wentworth and Zoback, 1990; Jachens et al., 1995).

However, details of how these processes have affected evolution of the transform boundary and associated basins east of the San Andreas fault are poorly known and data on long-term slip history and kinematic evolution of most of the eastern transform boundary zone faults are largely lacking north of San Francisco Bay, beyond paleoseismic investigations of Holocene faulting or geomorphologic studies (e.g., Prentice and Fenton, 2005; Hecker et al., 2005; Lock et al., 2006). The Rodgers Creek–Maacama fault system is well suited for detailed study of this long-term slip history because of its suggested continuity with the creeping Hayward fault zone south of San Pablo Bay, and because the fault system displaces thick sequences of Neogene volcanic and sedimentary layers that are readily datable and correlatable and useful in working out fault slip histories.

Based on the potential for constraining long-term slip rates, we have used framework geologic mapping, new ⁴⁰Ar/³⁹Ar dating, and tephrochronology to establish a detailed chronostratigraphy for interpreting the offset history of the Rodgers Creek–Maacama fault system. The timing of faulting and basin formation is determined from the sedimentologic and structural relations of interbedded sedimentary and volcanic units. Configurations of structural basins that we interpret to have formed during evolution of the Rodgers Creek–Maacama fault system are constrained from recent gravity and aeromagnetic investigations (Langenheim et al., 2006, 2008, 2010; McPhee et al., 2007). The amounts of offset and slip rates for the principal faults of the Rodgers Creek–Maacama fault system are determined from best estimates of the limits of distribution of the displaced volcanic and sedimentary sequences, including Mesozoic bedrock units. We compare the kinematics of fault zone and pull-apart basin evolution with laboratory models and determine the contribution of the Rodgers Creek–Maacama fault system to the total long-term slip budget of the Hayward fault zone and other faults of the East Bay fault system.

Samples of Neogene volcanic rocks used to establish offsets and rates of slip across the Rodgers Creek–Maacama fault system (Table 1) were analyzed by ⁴⁰Ar/³⁹Ar methodology, either by incremental-heating with a tantalum resistance furnace and molybdenum crucible, or by laser fusion analysis with a CO₂ laser. The ⁴⁰Ar/³⁹Ar analyses were done mainly in the Menlo Park Geochronology lab of the U.S. Geological Survey. One sample cited in Table 1 was dated by A. Deino at Berkeley Geochronology Center (Wagner et al., 2011). (For details of dating methodology and mineral separation and sample processing procedures that apply to the samples of this study, see Sarna-Wojcicki et al., 2011; Wagner et al., 2011.)

Samples of volcanic ash used to make stratigraphic correlations (Table 2) were chemically analyzed by electron-microprobe analysis, energy- and wavelength-dispersive X-ray fluorescence, and instrumental neutron activation methods and compared to the compositions of other tephra units in a database of ∼5500 analyses (Sarna-Wojcicki et al., 2011). Correlations in Table 2 were established based on similarity coefficients to tephra units of known ages in the database. The tephra correlations in this study are partly reinforced by ⁴⁰Ar/³⁹Ar ages, but several local tephra layers are correlated primarily on the basis of their geochemical similarities and stratigraphic positions between well-dated widespread ash units in the region. The geochemical correlations are not only useful for age determinations, they also are useful in determining locations of the eruptive sources, especially for far-field volcanic eruptions. For a detailed discussion of the methodology used for tephra correlations in the northern San Francisco Bay region, see Sarna-Wojcicki et al. (2011, and references therein).

Figure 1B is a simplified representation of the hierarchy of fault nomenclature used in this paper. Our focus is on evolution of faulting east of the main boundary between the Pacific and North American plates (the San Andreas fault), recognizing that the plate boundary is broad and that relations between the San Andreas and the strike-slip faults to the east have changed with time due to northward migration of the main plate boundary. Here, we focus on two main fault systems east of the San Andreas fault: the East Bay and Rodgers Creek–Maacama fault systems, which are also considered to be linked by the Hayward and proto-Hayward fault zones of the East Bay fault system (Fig. 1B). Our study concentrates only on the part of the East Bay fault system that extends north of San Pablo Bay and west of Napa Valley. The Rodgers Creek–Maacama and East Bay fault systems include several other fault zones discussed in the text and delineated in more detail in Figures 2 and 3 (and other figures). The Rodgers Creek–Maacama fault system includes both the Rodgers Creek and Maacama fault zones.

The proto-Hayward fault zone is largely transpressional and is located southwest of Sebastopol and Cotati; it predated initiation of the Rodgers Creek–Maacama fault system and contributed to long-term displacement of the Hayward and Calaveras faults (Figs. 1 and 2). The proto-Hayward fault zone, as used here, incorporates several local faults and fault zones, including the Tolay, Bloomfield, Petaluma Valley, and Burdell Mountain, discussed previously (e.g., Wagner et al., 2005; Graymer et al., 2002; McLaughlin et al., 1996). Several recent studies suggest that beginning ca. 12 Ma, the composite proto-Hayward, Hayward, and Calaveras fault zones contributed to a cumulative offset of ∼174 km across the southern Calaveras fault zone (Graymer et al., 2002; McLaughlin et al., 1996), though a somewhat larger total displacement has also been suggested (e.g., Wakabayashi, 1999). The amount of long-term displacement is inferred (1) from correlations of volcanic and sedimentary rocks at Burdell Mountain (BMV in Fig. 1) with the equivalent Miocene Quien Sabe Volcanics and underlying marine strata (QSV in Fig. 1) southeast of Hollister; (2) from distinctive offset Cretaceous rocks (McLaughlin et al., 1996); and (3) from an offset northward-younging trend in ages of hydrothermal mineralization and volcanism across the proto-Hayward, Hayward, and Calaveras fault zones of the East Bay fault system (Graymer et al., 2002; Obradovich et al., 2000; McLaughlin et al., 1996; Van Baalen, 1995).

The Rodgers Creek–Maacama fault system splays northeastward from the proto-Hayward fault zone, and the timing of this splaying is interpreted to represent initiation of the Rodgers Creek fault zone and abandonment of the proto-Hayward as the active extension of the East Bay fault system.

The southwestern side of the Rodgers Creek–Maacama fault system consists of the Rodgers Creek fault zone, which extends into San Pablo Bay, and steps southwest beneath the bay (Fig. 2) to emerge as the right-lateral Hayward fault in the eastern San Francisco Bay region west of Berkeley (Brown, 1970; Wright and Smith, 1992; Parsons et al., 2003). North of Santa Rosa, the Rodgers Creek fault zone consists of the northern Rodgers Creek fault zone, which locally is referred to as the Healdsburg fault segment of the Rodgers Creek fault zone. The part of the Rodgers Creek fault zone extending south of Santa Rosa is here referred to as the southern Rodgers Creek fault zone. Collectively, the southern and northern Rodgers Creek fault zones step right, and also go through a complexly evolved releasing bend in transferring slip to the Maacama fault zone northeast of Santa Rosa (Wong and Bott, 1995; McLaughlin et al., 2005, 2006, 2008; McPhee et al., 2007; Langenheim et al., 2008). Multiple strands and segments composing the separate northern Rodgers Creek and Maacama fault zones form a right step in which the two fault zones are subparallel and overlap along strike for ∼40 km between Santa Rosa and Geyserville. A non-overlapping right-releasing bend and pull-apart structure also links the Rodgers Creek and Maacama fault zones via the Bennett Valley fault zone in the Santa Rosa area. For the purpose of this paper, these right-stepped, overlapping, and bending links along with other segments and strands of the Rodgers Creek and Maacama fault zones define the Rodgers Creek–Maacama fault system (Figs. 1 and 2).

The Rodgers Creek–Maacama fault system is underlain by a composite basement that includes Jurassic to Miocene accretionary rocks of the Franciscan Complex, Jurassic to early Tertiary forearc or marginal basin strata of the Great Valley Sequence, and Jurassic mafic igneous rocks of the Coast Range Ophiolite (Fig. 2). Beneath the Great Valley, the Coast Range Ophiolite is considered to be part of the crystalline basement of the Great Valley Sequence. In the area of this study, however, the ophiolite and lower part of the Great Valley Sequence are structurally attenuated and complexly interleaved with Mesozoic and Tertiary rocks of the Franciscan Complex as a consequence of tectonism that predated the Rodgers Creek–Maacama fault system (McLaughlin et al., 1988).

The distribution of distinctive arc-related rocks and igneous breccias in tectonostratigraphic terranes of the Coast Range Ophiolite west of Sacramento Valley suggests that the ophiolite has undergone at least 240–320 km of dextral translation subparallel to the northern California margin (McLaughlin et al., 1988). At least 66–146 km of this dextral translation is attributable to Miocene and older faulting that predated northward migration of the San Andreas transform to this latitude (McLaughlin et al., 1988, 1996).

The Rodgers Creek–Maacama fault system is developed partly over remnants of a southeastward extension of the Neogene Eel River forearc basin (Nilsen and Clarke, 1989) that existed in most of northern California prior to its disruption by strike-slip faulting. Evidence of a Neogene forearc basin predating 8–9 Ma in the study area, however, has largely been removed by uplift and erosion or is buried in basins, though minor forearc remnants are recognized locally (e.g., older Miocene marine strata of Figs. 2 and 3). The amount and timing of displacement and slip rates for the Rodgers Creek–Maacama fault system are mainly constrained by the stratigraphic relations of the post-forearc marine and nonmarine sedimentary units deposited during development of the faulting, enhanced by the age distribution of intercalated volcanic rocks (Figs. 2, 4, and 5).

From ca. 10 to 4 Ma, the northern San Francisco Bay region, San Pablo Bay, and the area east of the Hayward fault zone were the site of a nonmarine to marine sediment transport system that flowed westward across the northern part of the East Bay fault system and San Andreas fault to the northward-migrating Pacific plate.

North of San Pablo Bay, nonmarine deposits of this paleodepositional system are assigned to the Petaluma Formation, equivalent to parts of the Contra Costa Group south of San Pablo Bay (Graham et al., 1984; Liniecki-Laporte and Andersen, 1988; Sarna-Wojcicki, 1992; Graymer et al., 2002; Allen, 2003; Wagner et al., 2005). These deposits are composed of fluvial gravels and sands exhibiting overall west-directed paleoflow, with local fresh-water diatomaceous (lacustrine) beds. Petaluma fluvial strata interfinger westward with brackish to estuarine deposits of mudstone and siltstone that may have encompassed the coastal outlet of a large river (Starratt et al., 2005; Allen, 2003). In the subsurface west of Santa Rosa, the Petaluma Formation interfingers with winnowed, bioclastic gravel and sandstone and siltstone of the marine Wilson Grove Formation, deposited in a shoreline to offshore open-ocean setting (Figs. 3, 4, 5B, and 5C; also see Sarna-Wojcicki, 1992; Graymer et al., 2002; Allen, 2003; Valin and McLaughlin, 2005; Wagner et al., 2005; Powell et al., 2006; Sweetkind et al., 2010). West of the Santa Rosa Plain (Figs. 2 and 3), Wilson Grove sediment was transported in submarine canyons across the San Andreas fault to the Pacific plate, where offset equivalents of these deposits are now found in the Delgada submarine fan southwest of the Mendocino Triple Junction (Drake et al., 1989; Sarna-Wojcicki, 1992).

A distinctive marker unit locally in the lower part of the Petaluma Formation, with an age between 7.3 and 6.7 Ma, consists predominantly of angular to rounded, unsorted volcaniclastic debris of rhyodacitic composition. The unit formed as a debris flow or talus breccia deposit from underlying volcanic rocks prominently exposed at Cooks Peak along the southwest flank of Taylor Mountain near Santa Rosa (Figs. 3–7). Much of the angular rhyodacitic debris is slickensided and some of the rhyodacitic blocks are >3 m in diameter. The locally cross-bedded breccia includes steep to moderately west-dipping lenses of nonvolcanic rounded fluvial gravel containing Franciscan Complex–derived clasts. The southeastern exposures of the breccia unit in the Taylor Mountain area terminate along strike at the southern Rodgers Creek fault zone (Figs. 3 [offset element R′] and 7). The breccia is exposed on the northeast side of the Rodgers Creek fault zone ∼28 km to the southeast, in the Sears Point area (Fig. 3 [offset element R], 6A, 6B, and 7; see also Wagner et al., 2011).

Pliocene and early Pleistocene fluvial and lacustrine sediments (Figs. 2, 4, and 5) unconformably overlie the Petaluma and Wilson Grove Formations. East of the northern Rodgers Creek fault zone, these deposits are compressed into northwest-trending open folds (Fig. 5, structure sections A and C). West of the northern Rodgers Creek fault zone, they are only mildly warped or undeformed, though water well and seismic data (Williams et al., 2008; Sweetkind et al., 2008) suggest that these strata are increasingly tilted and folded deeper in the subsurface.

Gravels of these deposits contain rare to common obsidian pebbles and generally are referred to as the Glen Ellen Formation, though regionally other names are locally applied. Geochemical fingerprinting of the obsidian pebbles (McLaughlin et al., 2004, 2005, 2008) shows their derivation is mainly from two widely separated obsidian sources of different age: one source area is 2.8 Ma flows and domes in the Napa and Franz Valleys (Figs. 7 and Table 1, location 18); the other source is 4.5 Ma flows in the Annadel area east of Santa Rosa (stratigraphic column 4 in Fig. 4; Table 1, location 26). The youngest folded deposits of the Glen Ellen Formation near Santa Rosa include a 0.8–1.2 Ma ash (stratigraphic section 3 in Fig. 4; location 72 in Table 2 and Fig. 7; see also Figs. 3 and 5), which correlates geochemically with the Bishop ash bed or the chemically similar younger set of the Glass Mountain ash beds from the Long Valley Caldera on the southeast side of the Sierra Nevada Mountains (McLaughlin et al., 2008; Sarna-Wojcicki et al., 2000, 2005; Metz and Mahood, 1991).

Obsidian clast provenance and paleoflow data show that paleodrainage for the Glen Ellen gravels was largely westward across the Rodgers Creek–Maacama fault system, into basins on the northern and southern parts of the Santa Rosa Plain (Sweetkind et al., 2008; see discussion of northern and southern Rodgers Creek fault zone displacement later in this paper). Though the timing and amount of strike-slip displacement of the Glen Ellen gravels along the Rodgers Creek fault zone seem to require it, the paleoflow and clast size distribution data show no clear indication that Glen Ellen deposition was concurrent with and controlled by strike-slip faulting. Perhaps the Rodgers Creek fault zone at that time was too diffuse and did not rupture to the surface often enough to create significant surface fault expression (e.g., basins, topographic barriers, and sediment transport channels) that would influence sedimentation patterns. In contrast, nearby studies (Nilsen and McLaughlin, 1985; McLaughlin and Nilsen, 1982) documented deposition concurrent with strike-slip faulting in basin gravels younger than 3 Ma uplifted along the Maacama fault zone.

Pleistocene deposits that overlie the deformed Pleistocene and older formations are generally flat lying and dissected, and may be mildly tilted locally, for example, as along the west side of the Santa Rosa Plain (Fig. 8). These deposits are broadly regarded as younger than the ca. 1.2–0.8 Ma tephra layer described near Santa Rosa.

Abundant volcanic rocks that range in age from ca. 12 to 1.2 Ma in our study area (Figs. 2, 4, and 5) provide the principal basis for dating faulting and related deformation. Analytical data for the ⁴⁰Ar/³⁹Ar ages determined for the Neogene volcanic units of this study are shown in Table 1. Tephrochronologic correlations of numerous chemically analyzed volcanic ash samples used to constrain stratigraphic relations and complement the radiometric dates are in Table 2. The map distributions of the dated volcanics and correlated tephra layers are in Figure 7 (keyed to Tables 1 and 2).

The volcanic rocks were largely erupted from volcanic centers east of, or dispersed along, the Rodgers Creek–Maacama and East Bay fault systems (Fig. 1), and are divided by age into different eruptive sequences intercalated with the sedimentary units described herein. As with the regional distribution of Neogene volcanics in all of the Coast Ranges, the ages of these volcanics generally young in a northeastward direction, but the volcanics are also displaced right-laterally with associated enclosing and overlying sedimentary units by the Rodgers Creek–Maacama fault system. Volcanics that constrain displacements across the Rodgers Creek–Maacama fault system, from oldest to youngest and from southwest to northeast, include those of Burdell Mountain, and the Tolay, Sonoma, and Clear Lake Volcanics. A tephra layer correlated herein with the Bishop ash bed or younger set of the Glass Mountain ash beds, with a far-field eruptive source in southeastern California, is recognized at one locality. Constraints imposed by the Neogene volcanic rocks on timing and amounts of displacement for specific faults of the stepover fault system are included in the discussion of faulting.

The oldest volcanic fields in the progression of northward-younging volcanism (Fox et al., 1985; Graymer et al., 2002) are the volcanics of Burdell Mountain and the Tolay Volcanics (Fig. 2), which are older than ca. 8 Ma and are southwest of the Rodgers Creek–Maacama fault system. These rocks have been displaced a minimum of many tens of kilometers from their in-place eruptive centers east of the Hayward and southern Calaveras faults. Details of these volcanic units and their displacements along faults predating the Rodgers Creek–Maacama fault system were discussed in detail elsewhere (McLaughlin et al., 1996; Graymer et al., 2002; Ford, 2003, 2007).

The ca. 8.0–2.5 Ma Sonoma Volcanics (Weaver, 1949; Table 1) are intercalated in the middle and upper parts of the Petaluma Formation and in younger Pliocene deposits dispersed between faults of the Rodgers Creek–Maacama fault system (Figs. 4 and 5). The Sonoma Volcanics are informally divided into age groupings associated with spatially separated northward-younging volcanic centers. The younger volcanic sequences in places overlap the older volcanics and pre-Neogene rocks. In the direction of their younging pattern from south to north, these informal age groupings include the San Pablo Bay, Napa Valley, and Mount St. Helena eruptive sequences. Local details of the stratigraphy of these volcanic sequences that were used to constrain displacements and slip rates for faults of the Rodgers Creek–Maacama fault system, are included in the discussion of faulting.

The Rodgers Creek–Maacama fault system consists of the Rodgers Creek and Maacama fault zones. The Rodgers Creek fault zone is divided into the northern Rodgers Creek fault zone north of the floodplain of Santa Rosa Creek (locally referred to as the Healdsburg fault segment) and the southern Rodgers Creek fault zone south of Santa Rosa Creek floodplain. In addition, the seismically active Bennett Valley fault zone, northeast and subparallel to the southern Rodgers Creek fault zone, partitions slip northeastward from the south end of the southern Rodgers Creek fault zone toward the Maacama fault zone. In the Santa Rosa area, this slip transfer is via the Spring Valley fault segment of the Bennett Valley fault zone that forms the eastern boundary of a prominent pull-apart basin beneath Santa Rosa and Rincon and Bennett Valleys (Figs. 3, 9, and 10). Significant transfer of slip between the Maacama and Rodgers Creek fault zones occurs across this pull-apart structure.

As the principal southwestern bounding fault zone of the dextral right-stepped Rodgers Creek–Maacama fault system (Figs. 1–3), the Rodgers Creek fault zone represents the earliest and most complexly evolved part of the stepover system. The fault zone complexity appears to result from at least four fault zone reorganizations that gave rise to separately named faults of different orientations and rates of right-lateral slip with time. Long-term slip rates for the Rodgers Creek fault zone have sequentially changed with fault zone geometries during these four reorganizations.

These faults bound concealed basins beneath the Santa Rosa Plain and are not mapped at the surface everywhere, but are inferred at depth from gravity data and from local normal faults draped by fault breccia (Figs. 3, 5, and 9). Basins buried beneath the Santa Rosa Plain are bounded on their east sides (Figs. 2, 3, and 5) by a 900–1400-m-high west-facing basement escarpment that is evident in gravity data (Langenheim et al., 2008, 2010; McPhee et al., 2007). This west-facing basement surface is inferred to be a west-side-down zone of normal faults bounding the Santa Rosa Plain, as shown in Figure 5 (section C-C′). The extensional character of these early faults is expressed at Cooks Peak south of Santa Rosa (Fig. 5), where a prominent 7.3–8.0 Ma rhyodacite unit (Table 1, locations 10, 27) is intruded along the Cooks Peak fault zone (mapped as a west-dipping normal fault by McLaughlin et al., 2008). The fault here is draped by the Breccia of Warrington Road and Sears Point (see discussion of stratigraphy of Neogene sedimentary rocks) composed of angular, coarse, blocky, slickensided debris (Fig. 6) derived from the rhyodacite of Cooks Peak with minor interbedded fluvial gravel of the lower Petaluma Formation (Fig. 5; see McLaughlin et al., 2008). Overlying Petaluma strata contain two tuffs that are dated as ca. 6.7 Ma (tuff of Lichau Creek; Wagner et al., 2011) and 6.3 Ma (the Roblar Tuff; see Table 2). Fault scarp breccias similar to the breccias of Warrington Road and Sears Point are particularly characteristic of extensional strike-slip basin margins, and are well documented in the Ridge Basin of southern California (Crowell and Link, 1982), the Hornelen Basin of Norway, and along the Maacama fault zone (the Little Sulfur Creek basins of Nilsen and McLaughlin, 1985). Although fault scarp breccia deposits conceivably can form in compressional settings, they are usually associated with faulted extensional strike-slip basin margins, consistent with the strike-slip setting of the breccias of Warrington Road and Sears Point. Breccias that might form along the scarps of thrust faults during basin inversion would probably not be exposed or preserved due to crustal shortening and structural burial. A thrust fault breccia, would be susceptible to entrainment in the fault zone and to being overridden by the hanging wall of the thrust. The breccia-draped extensional scarp along the Cooks Peak fault zone and its suggested westward connection with the escarpment beneath the Santa Rosa Plain is thus inferred to mark the opening of a large pull-apart basin between ca. 7.3 and 6.7 Ma (ca. 7.0 ± 0.3 Ma). The exposures of fault scarp breccias we correlate across the Rodgers Creek fault zone in the Taylor Mountain and Sears Point areas are now separated by a later stage of faulting along the southern Rodgers Creek fault zone.

Discontinuous normal faults are also mapped along the west side of the Santa Rosa Plain, including the Laguna de Santa Rosa fault, faulting uncovered in excavations near Sebastopol, and faulting seen in cuts along Laughlin Road south of Sonoma County Airport (Figs. 2, 3, and 8). These faults displace early Pleistocene and older deposits and are inferred to be linked to the same extension as normal faults seen on the east side of the Santa Rosa Plain. They have been active in the Quaternary, but have relatively minor down-to-the-east displacements of less than a few meters, and are discontinuous at the surface. These faults are also weakly expressed in the subsurface based on gravity data, compared to the major subsurface basement escarpment bounding the east side of the Santa Rosa Plain (Langenheim et al., 2010; McPhee et al., 2007). Based on this structural relief we interpret normal fault displacement to have been focused along the east side of the plain and to reflect earliest slip on the Rodgers Creek fault zone. If this interpretation is correct, the age of the fault scarp breccias of Warrington Road and Sears Point constrains the timing of earliest slip on the Rodgers Creek fault zone to ca. 7.0 ± 0.3 Ma.

Northeast-directed imbricate thrust faults are mapped southwest of the southern Rodgers Creek fault zone south of Santa Rosa (McLaughlin et al., 2008), where they underlie Taylor Mountain and the Cooks Peak fault zone (Figs. 3, 5C, and 9). These poorly exposed thrust faults dip moderately southwest (∼35°) and warp and imbricate the volcanic and sedimentary section. The thrusts generally place 7.3 Ma and older volcanics and Petaluma Formation strata on the southwest side of the Taylor Mountain fault zone, over 6.3 Ma and younger volcanics and strata to the northeast. The strike-slip–related, northeast-directed transpressional motion of these faults is interpreted to have uplifted and exposed the former normal fault–bounded margin of the basins beneath the Santa Rosa Plain. At the surface, the Taylor Mountain fault zone (Figs. 3 and 9) is mapped as dipping southwest beneath the earlier fault scarp breccia-draped extensional faults of the Cooks Peak fault zone. In the subsurface (Fig. 5, section C) this thrust faulting is interpreted to have reactivated faults of the west-facing extensional basement escarpment beneath the Santa Rosa Plain.

Structural repetition of ca. 5.4 Ma and older volcanic and sedimentary rocks by closely spaced faults of the Taylor Mountain fault zone is indicative that the transpressional faulting is ca. 5.4 Ma or younger (McLaughlin et al., 2008). We suggest a similar timing for the onset of transpression associated with blind thrusts beneath the Trenton Ridge structural high (Figs. 3, 5B, and 5C; McLaughlin et al., 2008; McPhee et al., 2007) that divides the Santa Rosa Plain into the Cotati and Windsor basins (Figs. 2 and 3). Well-log stratigraphy (Valin and McLaughlin, 2005; Powell et al., 2006; Sweetkind et al., 2010) and seismic reflection data (Williams et al., 2008; Sweetkind et al., 2008) show that growth of the Trenton Ridge began well before 3 Ma and that its uplift and erosion continued until ca. 1 Ma.

Transpressional deformation over Taylor Mountain, the east side of the Santa Rosa Plain, and beneath Trenton Ridge was followed by a shifting of slip to dominantly right-lateral, geomorphically youthful, steeply dipping faults of the southern Rodgers Creek fault zone. Southeast of Taylor Mountain and Santa Rosa, the transpressional Taylor Mountain fault zone (Figs. 2, 3, 5C, and 9) and basin-bounding extensional faults of the Cooks Peak fault zone splay northwest from a local north-northwest jog in the active southern Rodgers Creek fault zone. The youngest traces of the southern Rodgers Creek fault zone in that area are oriented ∼30° clockwise from the trends of the Taylor Mountain and Cooks Peak fault zones (Figs. 2, 3, 7, and 9). The 7.3–8.0 Ma rhyodacitic volcanics of Cooks Peak and overlying fault scarp breccias of Warrington Road and Sears Point that are bounded by these splaying faults are truncated against the southern Rodgers Creek fault zone (McLaughlin et al., 2008; Figs. 2, 3, and 9). The recently active fault segments and the older transpressional and extensional fault segments, however, are colinear (Figs. 2 and 3) farther to the southeast. Restoration of the rhyodacitic volcanics and fault scarp breccia of the Cooks Peak–Taylor Mountain area across the southern Rodgers Creek fault zone to the Sears Point area, based on their truncation at the Southern Rodgers Creek fault zone, together with an antiformal axis that aligns after restoring offset of the volcanics and breccia (Figs. 3 and 11), suggests that ∼28 ± 0.5 km of right-lateral displacement is taken up by the combined Cooks Peak, Taylor Mountain, and more youthful southern Rodgers Creek fault zones. The 28 ± 0.5 km dextral displacement of the fault scarp breccia, rhyodacitic volcanics, and antiform axis is inferred to have been taken up since ca. 7 Ma, first by transtensional slip along the Cooks Peak fault zone, followed by transpressional displacement along the Taylor Mountain fault zone, and most recently by steeply dipping active faults of the northern and southern Rodgers Creek fault zones. Relative amounts of the total strike slip partitioned to the Cooks Peak and Taylor Mountain fault zones is unknown, except that the extensional and compressional styles of these early faults imply that much pre-Quaternary displacement occurred as dip slip.

The Santa Rosa pull-apart basin (McLaughlin et al., 2008; McPhee et al., 2007) is a structure that defines the most recent stage of Rodgers Creek–Maacama fault system reorganization. This pull-apart structure is a young, ∼3-km-wide extensional depression between the Rodgers Creek and Maacama faults in the Santa Rosa area, filled with a thin cover of undeformed Quaternary sediments (Figs. 3, 5B, 5C, 9, 10, and 12). Faults bounding the east and west sides of this structure as well as the principal bounding faults of the Maacama and Rodgers Creek fault zones to the northeast and southwest are seismically active with prominent microseismicity (Fig. 9) extending to depths of ∼10 km and with focal mechanisms indicating pure and oblique strike slip, with secondary components of extension or compression. The Santa Rosa area was severely shaken by two earthquakes (M5.6 and M5.7) in October 1969 that were located on the northern Rodgers Creek fault zone close to the western margin of the Santa Rosa pull-apart basin (Wong and Bott, 1995; McPhee et al., 2007).

The geometry and timing of the opening of this pull-apart structure affected the long-term partitioning of slip between the Rodgers Creek and Maacama fault zones. Undeformed Quaternary sediments deposited in the north-oriented depression of the pull-apart basin unconformably overlie gravels and Sonoma Volcanics in northern Rincon Valley that are compressed into a northwest-trending synclinal trough. This relation is interpreted to indicate that the pull-apart depression postdates the folding and formed prior to and during deposition of the undeformed sediment fill. From the earlier section on Neogene stratigraphy, the upper part of the folded section includes the 0.8–1.2 Ma Bishop ash bed or an ash correlative with the younger set of the Glass Mountain ash beds, erupted from the Long Valley Caldera on the southeast side of the Sierra Nevada Mountains (McLaughlin et al., 2008; Sarna-Wojcicki et al., 2000, 2005; Metz and Mahood, 1991). This constrains opening of the pull-apart basin to after ca. 1.0 ± 0.2 Ma. The southwestern side of the Santa Rosa pull-apart basin is bounded partly by the southern Rodgers Creek fault zone and also by the Matanzas Creek fault zone, which splays southeast from the southern Rodgers Creek fault zone east of Taylor Mountain and merges with the Bennett Valley fault zone (Fig. 9). The approximate distance along the parallel trends of the Matanzas Creek and southern Rodgers Creek fault zones necessary to close the Santa Rosa pull-apart structure is ∼6.5 ± 0.5 km. We interpret this as the amount of dextral slip taken up by the Matanzas Creek and Bennett Valley fault zones during opening of the Santa Rosa pull-apart basin (Fig. 9).

Similarly, the northeast side of the Santa Rosa pull-apart basin is bounded for ∼6.0 ± 0.5 km (Fig. 9) by faults associated with the south end of the Maacama fault zone (including strands of the Maacama and Mark West fault zones). Like the south side of the pull-apart basin, this length of the Maacama fault zone that bounds the north side of the pull-apart basin is the approximate distance required to close the north side of the basin. As such, it is inferred to represent the approximate amount of dextral strike slip taken up by the Maacama fault zone during opening of the pull-apart basin. Thus, the Santa Rosa pull-apart basin represents a kinematic link between the Bennett Valley and Maacama fault zones that partly explains the partitioning of slip between the southern Rodgers Creek and Maacama fault zones across a prominent releasing bend or step initiated in the past 1 m.y. In the Santa Rosa area, however, the partitioning of slip between the Rodgers Creek fault zone and the Maacama fault zone since 1.0 Ma is accommodated on the Matanzas Creek and Bennett Valley fault zones rather than on the southern Rodgers Creek fault zone. The Bennett Valley fault zone converges with the southern Rodgers Creek fault zone only south of Sears Point, beneath Sonoma Valley or San Pablo Bay. The northern Rodgers Creek fault zone (Healdsburg fault segment), which is seismically active (Fig. 9) and displays evidence of Holocene surface displacement (Hecker and Kelsey, 2006; Crampton et al., 2004; Swan et al., 2003), is seemingly a continuation of the southern Rodgers Creek fault zone, apparently bypassing the Santa Rosa pull-apart structure. The northern and southern Rodgers Creek fault zones, north-trending faults bounding the west side of the Santa Rosa pull-apart basin, and the Matanzas Creek fault zone all merge or intersect each other beneath Santa Rosa Creek floodplain.

The Healdsburg fault segment is north of the Santa Rosa Creek floodplain (Figs. 2, 3, and 9). Gravity and aeromagnetic data suggest that the Healdsburg fault segment and southern Rodgers Creek fault zone are connected at shallow depth (Langenheim et al., 2008, 2010) and thus have overlapping histories and possibly similar long-term rates of slip. Surface geologic mapping (McLaughlin et al., 2008; Figs. 3 and 9) indicates that the fault connection occurs across a covered small right jog beneath the floodplain of Santa Rosa Creek.

Direct partitioning of slip from the Matanzas Creek fault zone to the northern Rodgers Creek fault zone (Healdsburg fault segment) is significantly diverted by faults associated with the eastern and western margins of the Santa Rosa pull-apart basin. North-trending faults bounding the Santa Rosa pull-apart basin (e.g., the Spring Valley fault segment of the Bennett Valley fault zone; Fig. 9) link and clearly transfer slip between these faults and the Maacama fault zone, as indicated by seismicity. Other seismicity and youthful fault geomorphology are dispersed along and between an overlap in the along-strike trends of the northern Rodgers Creek and Maacama fault zones northwest of Santa Rosa (Fig. 9). Faults that may accommodate the partitioning of slip to the northern Rodgers Creek fault zone include the likely link between the southern and northern Rodgers Creek fault zones beneath the Santa Rosa Creek floodplain; the Matanzas Creek fault and earlier extensional and transpressional faults that disrupt Neogene volcanic and sedimentary units north and south of the Santa Rosa Creek floodplain. Several faults mapped east of the northern and southern Rodgers Creek fault zones (McLaughlin et al., 2004, 2008) exhibit strike slip, reverse slip, and normal slip, but their contributions to the Maacama or northern Rodgers Creek fault zones are largely unknown (McLaughlin et al., 2002).

Displacement on the northern Rodgers Creek fault zone (Healdsburg fault segment) and its rate of slip north of Santa Rosa since 1 Ma are currently unconstrained by the bedrock geology. A longer term displacement history can be determined for the Healdsburg fault segment and southern Rodgers Creek fault zone between ca. 3 and 1 Ma, however, assuming that they were continuous prior to opening of the Santa Rosa pull-apart basin.

Correlative gravel remnants of the Glen Ellen Formation that now are separated right-laterally across the northern and southern Rodgers Creek fault zones are dated as younger than 2.8 ± 0.02 Ma from their contained obsidian clasts and a younger than 3.1 Ma basal tuff (Fig. 3, areas H and H′). The gravel remnants were therefore apparently right-laterally separated across the Rodgers Creek fault zone after ca. 3 Ma. The outcrop separation, however, does not provide a well-defined piercing blob for establishing fault displacement because the gravel remnant east of the Rodgers Creek fault zone is now isolated on a ridge top ∼2 km from the main fault zone and because the original gravel distribution has been modified by dissection and erosion. However, the presence in the gravel of obsidian clasts derived from in-place sources in the Annadel State Park area to the northeast (location 26, Table 1) is distinct from other gravels of the Glen Ellen Formation east of the Rodgers Creek fault zone that contain only obsidian clasts sourced from the Napa and Franz Valley areas (locations 18, 20, Table 1). The correlative gravel remnant along the southwest side of the Rodgers Creek fault zone (Figs. 3 and 12, offset points H, H′) is the northwesternmost area of known Annadel-sourced obsidian clasts southwest of the Rodgers Creek fault zone. The distribution of several other Glen Ellen gravel localities containing Annadel-sourced obsidian clasts on the Santa Rosa Plain to the southwest, together with paleoflow data, constrains the aerial configuration of the fluvial system that transported this lithofacies from the Annadel area (McLaughlin et al., 2005; Sweetkind et al., 2008, 2010). The distribution of the Annadel lithofacies on the Santa Rosa Plain suggests that the fluvial transport system may have had a width of ∼2.5–3.0 km where it crossed the Rodgers Creek fault zone (Fig. 12). Using this width to crudely constrain that of the Annadel-sourced fluvial system for our one locality east of the southern Rodgers Creek fault zone, and assuming that the gravel at this exposure was deposited in a 3-km-wide paleochannel, permits a crude restoration of dextral displacement. Based on this restoration (Fig. 12), we estimate that the Annadel-sourced gravel remnants are offset ∼14.8 ± 6.0 km across the northern and southern Rodgers Creek fault zones.

Paleoseismology studies since the 1990s along the Rodgers Creek fault zone south of Santa Rosa provide a Holocene slip rate estimate for the southern part of the Rodgers Creek fault zone of 6.4–10.4 mm/yr, with an average rupture recurrence of 131–370 yr (Hecker et al., 2005; Budding et al., 1991). In addition, recent satellite-based permanent scatterer interferometric synthetic aperture radar (PS-InSAR) studies (Funning et al., 2007) suggest that to the northwest and southeast of the Santa Rosa pull-apart basin, the Rodgers Creek fault zone is undergoing as much as 7.5 ± 2.6 mm/yr of shallow creep above a depth of 6 km.

Long-term slip rates for several time windows during evolution of the Rodgers Creek fault zone between ca. 7 and 0.8 Ma are inferred here, from displacement constraints on the several faults described here (Table 3). As the geometry and style of faulting associated with the Rodgers Creek fault zone evolved, the components of normal and reverse slip on early faulting stages appear to have increasingly been taken up by younger, steeper faults that accommodated larger components of right-lateral strike slip.

The 28 ± 0.5 km of total minimum offset estimated for the Rodgers Creek fault zone (Table 3) is partitioned between the combined active southern Rodgers Creek and northern Rodgers Creek fault zones, thrust faults that partitioned and uplifted the east side of the Santa Rosa Plain between ca. 5 and 3 Ma, and earlier extensional faults bounding the east side of Cotati basin (Fig. 13; Table 3). Approximately 13.2 ± 1.8 km of strike-slip displacement appears to have predated the later than 2.78 ± 0.02 Ma deposition of gravels offset along the combined southern and northern Rodgers Creek fault zones (Table 3). The unconstrained partitioning of this 13 km of slip could be taken up in part by the early transpressional faults (e.g., Taylor Mountain fault zone, active between 5 and 3 Ma) and extensional (probably transtensional) faults (e.g., Cooks Peak fault zone, active between ca. 5 and 7 Ma) that splay northwestward on Taylor Mountain from their junction (Figs. 3, 9, and 13) with the active southern Rodgers Creek fault zone.

The Santa Rosa pull-apart basin that initiated a well-delineated link of partitioned slip between the Bennett Valley and Maacama fault zones via the Spring Valley fault (Fig. 9) is not clearly linked to the Rodgers Creek fault zone at the surface. If the Matanzas Creek fault zone existed prior to ca. 2.8 Ma, unknown additional slip could have transferred between the northern Rodgers Creek and Bennett Valley fault zones via the Matanzas Creek fault zone. The northern Rodgers Creek fault zone (Healdsburg fault segment) may currently take up all southern Rodgers Creek fault zone slip, but complexities along the junction of the Rodgers Creek fault zone with the Santa Rosa pull-apart basin discussed here (Fig. 9) and lack of a recognized offset exclusively along the northern Rodgers Creek fault zone (Healdsburg fault segment) leads us to consider the northern Rodgers Creek fault zone slip rate since ca. 1.0 ± 0.2 Ma as unconstrained.

Based on these offset relations and an assumption that the earliest phase of extensional deformation for the Rodgers Creek fault zone included a component of dextral slip, the composite long-term slip rate of the Rodgers Creek fault zone since opening of the Cotati basin is 28 ± 0.5 km in 7.0 ± 0.3 m.y., or 4.1 ± 0.3 mm/yr (Table 3).

If the early phase of extensional deformation did not accommodate any of the long-term dextral displacement (a permissive but unproven interpretation), it can be argued that all strike slip on the fault zone has occurred since the initiation of transpression ca. 5 Ma. This latter interpretation would yield a higher composite long-term slip rate of ∼5.6 mm/yr, which is similar to rates derived here for the more recent time windows of fault zone evolution and is compelling for that reason. Normal faults bounding the west side of the Santa Rosa Plain, however, have orientations slightly oblique to the direction of regional extension, suggesting a component of transtension during the early extensional basin phase of fault zone evolution that would contribute to and result in a lower composite long-term rate of strike slip. Both options for modeling early slip suggest that significant fault displacement during early stages in the evolution of the Rodgers Creek fault zone was translated into vertical slip, reducing the derived long-term rate of strike slip.

The composite long-term slip on the Rodgers Creek fault zone before 0.8–1.2 Ma is 13.2 ± 1.8 km in 5.2 ± 1.3 m.y., or 2.8 ± 1.1 mm/yr. Though not transferred to the Rodgers Creek fault zone north of Sears Point, displacement on the Bennett Valley and Matanzas Creek fault zones since the 1.0 ± 0.2 Ma opening of the Santa Rosa pull-apart basin appears to have been ∼6.5 ± 0.5 km at a rate of ∼6.8 ± 1.8 mm/yr. This slip rate is significantly higher than for the somewhat earlier faults of the Rodgers Creek fault zone in the Santa Rosa area.

The lithofacies of the Glen Ellen Formation containing obsidian pebbles derived both from Annadel and sources in the Napa and Franz Valleys is offset 14.8 ± 6.0 km across the combined northern (Healdsburg fault segment) and southern Rodgers Creek fault zones, which are inferred to have been more continuous prior to formation of the Santa Rosa pull-apart basin. The displacement of these Glen Ellen Formation gravels yields a slip rate of 5.3 ± 2.2 mm/yr since 2.76–2.80 Ma, which is also higher than for the 3 Ma and older composite faulting on the northern and southern Rodgers Creek fault zones. The composite slip rate for the Rodgers Creek fault zone thus appears to have increased prior to ca. 3 Ma, close to when transpression along the Taylor Mountain fault zone largely ceased and slip splayed eastward onto a newly initiated, dominantly strike-slip Rodgers Creek fault zone (Fig. 13). Faults of the currently active Rodgers Creek fault zone are subvertical in geometry and clearly accommodate dominant dextral strike slip (Wong and Bott, 1995), consistent with a comparatively higher observed rate of strike slip since 3 Ma.

If our proposed 7 Ma timing for initiation of slip on the Rodgers Creek fault zone is not valid, total slip on the Rodgers Creek fault zone could be significantly greater than 28 km, since no constraints on offset of units older than ca. 8 Ma are determined. Restoration of the offset breccia of Warrington Road to the breccias exposed near Sears Point is the minimum displacement needed to restore these rocks to their predisplacement location because the area southeast of Sears Point and Donnell Ranch is covered by alluvium and the San Francisco Bay margin. Although we have correlated the axes of antiformal features across the Rodgers Creek fault zone in addition to the offset fault scarp breccias (Figs. 3 and 11), this alignment is possibly fortuitous and the breccia of Warrington Road might restore farther south, to somewhere along the buried southern margins of Sonoma Valley or San Pablo Bay basins. The gravity expression of closure for Cotati and Windsor basins (Fig. 11; Table 3) suggests that an additional 24 km of slip along the Rodgers Creek fault would bring the northeast side of Windsor basin into alignment with the southwest side of Sonoma and San Pablo Bay basins along the Rodgers Creek fault zone (Fig. 11). This would increase the maximum slip on the Rodgers Creek fault zone to ∼52 km and raise the long-term slip rate to ∼7.7 ± 0.6 mm/yr, assuming the same timing of initiation of the faulting.

This larger displacement based on alignment of gravity-defined basin margins, however, implies that the Cotati and Windsor basins should include thick sections of the Coast Range Ophiolite overlain by Great Valley Sequence rocks as well as Tertiary strata that predate the Petaluma Formation, all of which occur in the Sonoma and San Pablo Bay basins (Wright and Smith, 1992). Significant sections of Great Valley Sequence and Coast Range Ophiolite are exposed along the northwestern margins of the Santa Rosa Plain and project beneath the Windsor and Cotati basins. However, the deepest drilled wells, which are in Cotati basin (∼1530–1820 m), bottomed in sedimentary rocks interpreted to be Franciscan Complex sandstone and argillite, with no intervening Coast Range Ophiolite or Great Valley Sequence. With the exception of oil having a Miocene Monterey Formation geochemical signature (Lillis et al., 2001), no actual pre-Petaluma Tertiary strata are known. An offset substantially greater than 28 km would also misalign the correlative fault scarp breccias of Warrington Road and Sears Point by 24 km along the Rodgers Creek fault zone, with no data from the intervening covered area to corroborate their continuity. Without more information on the subsurface distribution of the fault scarp breccias of Warrington Road and Sears Point beneath Sonoma Valley, the larger displacement and greater slip rate are here considered highly speculative. The lower long-term displacement and slip rate proposed here are therefore favored, but viewed as a minimum.

Displacement across the Maacama fault zone is estimated from several cross-fault correlations of distinctive units of the Sonoma Volcanics, Neogene gravels, and Mesozoic basement rocks. The most definitive estimate of long-term Neogene offset comes from the correlation of exposures of Sonoma Volcanics belonging to the Mount St. Helena eruptive sequence and associated distinctive fluvial strata (Fig. 7). The volcanics were vented from an area of flows, domes, intrusive vents, thick ash, and laharic breccia deposits that are exposed for ∼11 km along the east side of the Maacama fault zone.

The northwesternmost and youngest outcrop areas of Sonoma Volcanics occur between the northern Rodgers Creek and Maacama fault zones (areas M.1 and M′, Figs. 3 and 7). Together, these outcrop areas of Sonoma Volcanics constrain the maximum northwestward extent of Sonoma volcanism and seemingly also limit post–3.2 Ma offset across the Maacama fault zone.

The northwesternmost and youngest of these exposures are ∼2 km southwest of the Maacama fault zone just northwest of Geyserville (M.1, Fig. 3). These dacitic volcanics are apparently the youngest of the Sonoma Volcanics, with a ⁴⁰Ar/³⁹Ar age of 2.5 ± 0.09 Ma (location 1, Table 1; Fig. 7). No Sonoma Volcanics of equivalent age (2.5 Ma) have been mapped on the northeast side of the Maacama fault zone, although they could be present in undated, stratigraphically high parts of the Mount St. Helena volcanic section. Alternatively, the 2.5 Ma dacitic rocks represent a small volcanic center that was not offset by the Maacama fault zone, that erupted separately from and slightly northwest of the Mount St. Helena eruptive sequence. Somewhat younger rhyolitic volcanics of Pine Mountain (location 15, Table 1; Fig. 7), dated at 2.2 ± 0.029 Ma, that occur east of the Maacama fault zone and northwest of Mount St. Helena, are considered part of the younger Clear Lake volcanic field, and this constrains the northwestern extent of Sonoma volcanism east of the Maacama fault zone.

A second area of youngest and most northwestward displaced exposures of Sonoma Volcanics is exposed to the southeast of the Geyserville volcanics for ∼5 km along the southwest side of the Maacama fault zone (Figs. 7 and 14). These rocks are best exposed in the southeastern parts of these exposures, in roadcuts east of Alexander Valley (fault length M′ in Figs. 3 and 14; also locations 16 and 69 in Fig. 7; Tables 1 and 2). At this locality, a west-dipping section of ash flow and air-fall tuff unconformably overlies basaltic andesite. The volcanics in turn overlie steeply dipping or folded Pliocene fluvial siltstone and pebble gravel composed of rounded to subrounded clasts derived entirely from mélange of the Mesozoic Franciscan Central belt and the Coast Range Ophiolite. The gravels contain no clasts derived from Tertiary volcanics, distinguishing them from other 3.2 Ma and older gravel units of the region. The gravel, basaltic andesite, and tuff section abuts the southwest side of the Maacama fault zone along the Geysers-Healdsburg road. The tuff is dated by ⁴⁰Ar/³⁹Ar analysis of plagioclase at 3.17 ± 0.04 Ma (isochron age, location 15, Table 1; location 69, Table 2; Figs. 7 and 14). The lowermost tephra layers in the tuff section correlate geochemically with the Putah Tuff, dated elsewhere at 3.34–3.27 Ma and the uppermost ash-flow tuff of the section is correlated to tephra layers dated elsewhere at 3.25–3.19 Ma (McLaughlin et al., 2005, and this paper). Thus, the ash section was erupted in a relatively short time interval between ∼3.3 and 3.2 Ma.

We correlate the part of the Mount St. Helena eruptive center abutting the Maacama fault zone for ∼5–6 km along the southwest side of Franz Valley (fault length M in Figs. 3, and 14; McLaughlin et al., 2004) with the Geysers-Healdsburg road volcanics and gravels. Correlated tephra units and Ar/Ar ages in this area include the Putah Tuff (∼3.3–3.2 Ma), the tuff of the Petrified Forest (∼3.3–3.4 Ma), and a local tuff (tuff of the Pepperwood Ranch) dated at 3.19 Ma (See Tables 1 and 2 and Figure 7 for detailed age data and uncertainties). Locally, steeply dipping fluvial gravels with the same clast suite as the gravels along the Geysers-Healdsburg road unconformably underlie the volcanic section of Franz Valley. Basaltic andesite occurs sporadically and unconformably beneath the tuffs and gravels of Franz Valley and also higher in the tuff section. The proximal aspect and connection of the Franz Valley volcanic section to the Mount St. Helena eruptive center and its correspondence to the tuff and gravel section along the Geysers-Healdsburg road, suggests a displacement along the Maacama fault zone of between 17.5 and 24 km since ca. 3.2 Ma (Figs. 3, 7, and 14; Ar/Ar age location 3, Table 1; tephra locations 23 and 13–16, Table 2).

A bedrock-offset relation corroborating Neogene displacement of the Sonoma Volcanics across the Maacama fault zone restores 21–22 km of displacement of the Coast Range Ophiolite from the vicinity of Hopland to the Geyser Peak area (Fig. 15; Table 3). This restoration aligns the northwest and southeast contacts of the Geyser Peak section of the Coast Range Ophiolite to the northeast, with northwest and southeast contacts of the ophiolite on the southwest side of the fault near Hopland (Fig. 15). The mapped extent of the Hopland ophiolite belt along the southwest side of the Maacama fault zone corresponds closely with the width of the Geyser Peak ophiolite section along the northeast side of the Maacama fault zone, providing an elongate 3 ± 0.5 km wide body that is offset 21.5 ± 0.5 km along the Maacama fault zone. The Hopland section, recognized in this report as part of the Coast Range Ophiolite, was previously only mapped in reconnaissance as a west-northwest–trending belt of serpentinite enclosed by mélange of the Franciscan Complex (Irwin, 1960).

Reconnaissance of the Hopland area ophiolite section indicates that several aspects of its stratigraphy match that of the upper part of the Geyser Peak ophiolite section (Fig. 16). Criteria for this correlation include the presence of a gabbroic intrusive complex overlying serpentinized peridotite, together overlain locally by a distinctive angular, coarse clastic breccia of mafic plutonic, volcanic, and volcanopelagic debris of Jurassic age shed from the underlying ophiolite. This clastic ophiolitic breccia is overlain in both the Hopland and Geyser Peak areas by turbiditic sandstone and argillite composed predominantly of mafic detritus (Fig. 16). The Geyser Peak and Hopland sections of the Coast Range Ophiolite are typical of a tectonostratigraphic terrane of the Coast Range Ophiolite and lower Great Valley Sequence referred to as the Elder Creek terrane (Blake et al., 1985; McLaughlin et al., 1988; Hopson et al., 2008) that is exposed along the western side of the Sacramento Valley.

Except for the Hopland section of the Elder Creek terrane, this distinctive stratigraphy, including ophiolitic breccia at the base of the Great Valley Sequence, is unknown west of the Maacama fault zone. A very different, well-studied terrane of the Coast Range Ophiolite plus Great Valley Sequence referred to as the Healdsburg terrane (Blake et al., 1984; Hopson et al., 1981, 2008) occurs 25–30 km south of the Hopland ophiolite and west of the Maacama and Healdsburg faults and Alexander and Dry Creek Valleys (Fig. 15). The Healdsburg terrane includes thick volcanospelagic strata and keratophyric volcanic rocks in the ophiolite, overlain by Late Jurassic to Early Cretaceous non-ophiolite–derived conglomerate, sandstone, and shale (Fig. 16; Blake et al., 1984).

Fault offsets from aeromagnetic data are essentially based on the same criteria (ophiolitic or related mafic rocks) used to determine offset surface contacts of the correlated Geyser Peak and Hopland outliers of the Coast Range Ophiolite. Interpreted magnetic offsets, however, are based on the matching of similar maximum intensities and configurations of correlated magnetic (or nonmagnetic) bodies across the fault. Also, magnetic anomalies in general often reflect the geometry of a magnetic body at depth, and as such do not necessarily correspond with mapped surface contacts. In spite of these fundamental differences in how fault displacements are determined, the geologic and aeromagnetic data sets for the Maacama fault zone are complementary and provide similar independent long-term offset and slip rate estimates.

Separate basement offsets of 15 ± 3 km of a weakly magnetic mélange unit in the Franciscan Complex and a 21 ± 5 km offset of parts of the Coast Range Ophiolite were obtained by matching magnetic anomalies across the Maacama fault zone (respectively, anomalies 1–1′ and 2–2′, Fig. 17). Offset anomaly 1–1′ in Figure 17 corresponds to a mélange unit of the Franciscan Complex along the Maacama fault zone, which at the surface contains entrained lenticular bodies of serpentinite, gabbro, and greenstone. Offset anomaly 2–2′ (Fig. 17) matches northwest and southeast limits of an anomaly associated with the Hopland ophiolite section where it abuts the southwest side of the Maacama fault zone, with the projected extent of an anomaly over the Geyser Peak ophiolite northeast of the fault zone. The Geyser Peak anomaly is separated from the main trace of the Maacama fault zone by fault strands bounding the strike-slip basins of Little Sulfur Creek (Fig. 15), and the fill of these basins obscures the magnetic expression of truncation of the Geyser Peak anomaly at the Maacama fault zone.

Results of this study suggest that the Maacama fault zone has maintained a long-term average slip rate of ∼6.7 ± 1.2 mm/yr since ca. 3.17 ± 0.04 Ma, based on 17.5–24.0 km of offset of the Sonoma Volcanics. As discussed herein, the Maacama fault zone appears to have accommodated 6.0 ± 0.5 km of slip since 0.8–1.2 Ma, during Santa Rosa pull-apart basin opening. The average slip rate of 6.3 ± 1.8 mm/yr since 1.0 ± 0.2 Ma (Table 3) is generally comparable to the rate determined for the Maacama fault zone based on offset of the Sonoma Volcanics since 3.2 Ma. Geologic displacements of Jurassic ophiolitic basement across the Maacama fault zone favor a maximum displacement of ∼21.5 ± 0.5 km, which is about the same as the offset of the Sonoma Volcanics (20.9 ± 3.4 km). We therefore suggest an initiation time of ca. 3.17 ± 0.04 Ma for displacement along the Maacama fault zone and a long-term slip rate based on the offset ophiolite sections of 6.9 ± 0.4 mm/yr.

Aeromagnetically determined fault displacements are in reasonably close agreement with the geologically determined displacements, given the uncertainties associated with the different approaches. Maximum displacements of 15 ± 3 and 21 ± 5 km for two separate offset magnetic anomaly sets, associated with the Franciscan Complex and the Coast Range Ophiolite, respectively, suggest a total long-term displacement of 19 ± 7 km for the Maacama fault zone, at a rate of 6.0 ± 2.3 mm/yr (Fig. 17; Table 3).

By comparison, geodetic and paleoseismic data along the Maacama fault zone north of Santa Rosa suggest that its slip rate in the Holocene has fluctuated between 6.5 and 14 mm/yr (Freymueller et al., 1999; Larsen et al., 2005; Sickler et al., 2005; Prentice and Fenton, 2005; Simpson, 2005). Local episodic creep that occurs along the fault both at the surface and at depth is poorly understood in the context of modern fault kinematics (Galehouse, 2002; Freymueller et al., 1999), and the role of creep in long-term evolution of the fault zone is largely unknown. For this reason, differences in long-term displacements and slip rates determined from the surface geology compared to near term rates from paleoseismic or geophysical data may reflect real differences in the kinematics of the Maacama fault zone over time both at the surface and at depth, and not merely uncertainties inherent in the comparison of results derived from geologic versus geophysical approaches. The data sets presented here suggest there is reasonably close agreement between geologic and potential field-derived displacement data for the Maacama fault zone.

The total slip contributed to the Hayward fault zone by the Rodgers Creek fault zone amounts to at least 28 ± 0.5 km (Table 3). Based on averaged maximum and minimum displacements of all displacement criteria (Table 3), the Maacama fault zone separately contributes ∼20.4 ± 3.6 (16.8–24) km (Table 3) of displacement to the Hayward fault zone southeast of the Sears Point–Donnell Ranch area via the Bennett Valley fault zone (Figs. 11 and 13; Table 3). At least 44.4–52.5 (48.4 ± 1.4) km of slip is therefore contributed to the Hayward fault zone by the Rodgers Creek–Maacama stepover fault system south of the Sears Point–Donnell Ranch area.

Although antiformal axes appear to align after restoring displacement of a fault scarp breccia across the northern and southern Rodgers Creek fault zones, the correlation of the antiform axes is nonunique and their alignment could be fortuitous. The breccia exposures east of the Rodgers Creek fault in the Sears Point area also have an unknown distribution beneath the alluvium of southern Sonoma Valley. It is not recognized at the surface south of San Pablo Bay or reported in the subsurface of San Pablo Bay (Wright and Smith, 1992). A conservative interpretation of this data set infers the 28 ± 0.5 km offset of the fault scarp breccias to be a minimum displacement since 7.0 ± 0.3 Ma (Table 3).

As determined from different geologic criteria, long-term displacement has been 17.5–24 km and slip rates have been between 5.5 and 7.9 mm/yr on the Maacama fault since ca. 3.17 ± 0.04 Ma (Table 3). Displacement based on the matching of aeromagnetic anomalies across the Maacama fault zone yield a similar range of displacement (12–26 km) and a slip rate of 6.0 ± 2.3 mm/yr, if it is assumed that slip was initiated at 3.17 ± 0.04 Ma.

The total contribution of the Rodgers Creek–Maacama fault system to slip of the entire East Bay fault system south of San Pablo Bay since 7.0 ± 0.3 Ma appears to be at least 44.5–52.5 km, for a median long-term slip rate of 6.95 ± 0.85 mm/yr (6.1–7.8 mm/yr). Larger amounts of slip attributed to the East Bay fault system to the south are contributed from faults east of the Hayward fault and probably from poorly constrained pre–7 Ma slip on a proto-Hayward fault zone north of Burdell Mountain.

Although much of the northern Coast Ranges is now in compression (Fig. 1; Berry, 1973; Wentworth et al., 1984; Wentworth and Zoback, 1990; Argus and Gordon 2001) and the San Andreas fault is curved, with a restraining bend located at its northernmost end, the restraining bend is trailed to the south by a prominent releasing bend configuration (Fig. 1). Some studies (Stanley, 1987; Wilson et al., 2005) also suggest that this releasing and restraining bend configuration of the northern San Andreas fault has formed the Pacific–North American plate boundary since some time in the Miocene, and as such, its northward migration with the Mendocino Triple Junction should have influenced successive distributions of transtensional and transpressional structures for some distance east of the main plate boundary (the San Andreas fault). To a first order, this concept may be valid (that is, strike-slip–related basins become younger northward east of the San Andreas fault; Blake et al., 1978; McLaughlin and Nilsen, 1982; Nilsen and McLaughlin, 1985).

Numerous studies also point to the northward migration of a slab window beneath the Coast Ranges as having influenced the distribution of volcanism and related extension in the crust (Dickinson and Snyder, 1979; Lachenbruch and Sass, 1980; Fox et al., 1985; McLaughlin et al., 1996; Graymer et al., 2002). Thermal response of the crust to slab window migration may, in turn, have combined with the migrating releasing bend segment of the northern San Andreas fault (Fig. 1) to form the northward-younging transtensional (strike slip) basins of the Rodgers Creek–Maacama fault system. Releasing bend-related extensional fault geometry is viewed here as a structural element, separate from the migrating slab window, that provided needed open conduits and pathways for upward migration of magma from asthenospheric depths and for volcanic venting coeval with, or younger than, the surface faulting.

The Rodgers Creek–Maacama fault system as characterized here has evolved in conjunction with lengthening of the San Andreas transform margin. The fault system evolved as a series of extensional right steps and northeastward clockwise splays beginning ca. 7.0 Ma, with the opening of basins beneath the Santa Rosa Plain. Several fault zone reorganizations between ca. 7 Ma and the present are inferred from the orientations, slip characteristics, and timing of different fault sets of the northern and southern Rodgers Creek fault zones. From these relations we infer a sequence of superposed fault zone reorganizations that began with extensional strike slip followed by transpression and uplift, in turn followed by pure strike slip, and most recently by younger than 1 Ma reoriented extensional strike-slip faulting (Figs. 2, 3, 9, and 13).

The right-stepped Maacama fault zone exhibits a younger overlapping history of at least two reorganizations beginning ca. 3.2 Ma, with eruption of the upper part of the Mount St. Helena eruptive sequence that was accompanied or closely followed northeast of Healdsburg by initiation of extensional strike-slip faults of the Maacama fault zone. These faults bound the basins of Little Sulfur Creek (Fig. 15) and their associated syntectonic sedimentary fills (McLaughlin and Nilsen 1982; Nilsen and McLaughlin, 1985). Deposition in these strike-slip basins along the Maacama fault zone was followed by transpression that uplifted, dissected, and compressed the basins. The recent north-northwest–trending, younger than 1 Ma extensional strike-slip faults that are associated with opening of the Santa Rosa pull-apart basin splay from the Matanzas Creek, Bennett Valley, and Rodgers Creek fault zones. Southeast of Santa Rosa, these north-northwest–trending faults overprint earlier, more northwest-oriented basin-bounding faults of the Maacama fault zone (Figs. 2, 3, 9, and 13).

The long releasing bend in the northern San Andreas fault zone is currently adjacent to and west-northwest of the Rodgers Creek–Maacama fault system, and thus may influence the extensional strike-slip setting of the Rodgers Creek–Maacama fault system relative to motion of the Pacific plate. However, the timing and sequence of reorganizations of the Rodgers Creek–Maacama fault system that we have observed do not have a straightforward correspondence with the regional-scale patterns of compression and extension associated with bends in the northern San Andreas fault zone. The succession of extensional and compressional components of the Rodgers Creek–Maacama fault system with time instead appears to be a more complicated response to Mendocino Triple Junction migration. The reorganized fault geometries seen with the Rodgers Creek–Maacama fault zone actually may be initiated sequentially at the southern end of the fault system, as a separate but necessary response to continual lengthening and changing of fault geometry at the northern end of the fault system with triple junction migration. Local compressional structures also are shown in laboratory models to be integral parts of active pull-apart basin settings (e.g., pop-up structures described by Dooley and McClay, 1997) and thus may not always represent temporally separate transpression.

Other studies (Wells and Simpson, 2001; Williams et al., 2006) suggest that faulting kinematics in the northern Coast Ranges east of the San Andreas fault are significantly influenced by basement fault block interactions north and south of the Mendocino Triple Junction (Fig. 1). Thrust faults of east-directed structural wedges formed during early Tertiary plate convergence that uplifted and unroofed the Mesozoic basement of the Coast Ranges are examples of preexisting block boundary structures that can be reactivated in later transpressional settings and influence locations and geometries of Quaternary blind thrusts (Unruh et al., 2004, 2007; Wentworth et al., 1984; Wentworth and Zoback, 1990). In contrast, recent seismic experiment results interpret the Maacama and other active strike-slip faults in the northern Coast Ranges to extend through the entire crust of the North American plate (Beaudoin et al., 1998; Hole et al., 1998, 2000; Henstock and Levander, 2003), raising questions of how the strike-slip faults of the Rodgers Creek–Maacama fault system might interact with reactivated wedge thrusts. The nature of Mesozoic terrane boundary faults in the upper to mid-crust and their unknown contribution to the kinematics of the Rodgers Creek–Maacama fault system are beyond the scope of this paper, but we note their demonstrated significance to the east along the boundary between the Sacramento Valley and northern Coast Ranges (Unruh et al., 2004, 2007; Wentworth et al., 1984; Wentworth and Zoback, 1990).

Scaled-sandbox models of stepping strike-slip faults and derived pull-apart basins (Dooley and McClay, 1997) provide insight into several features of the Rodgers Creek–Maacama stepover fault system. The modeling shows that northwest-trending principal bounding faults of an evolving dextral right-stepped fault system initially do not overlap along strike. Rhombic-shaped pull-apart basins that form with this geometry of non-overlapping strike-slip faults have bounding extensional faults with north-northwest orientations. These basins are referred to as 30° non-overlapping releasing sidestep pull-apart basins (Figs. 9 and 10). As the fault system evolves, the principal strike-slip faults bounding the right step area lengthen, to where their ends are at 90° to each other, resulting in a box-shaped basin geometry referred to as a 90° releasing sidestep pull-apart basin (Fig. 10). Additional lengthening of the principal bounding strike-slip faults results in a right-stepover region, the principal northeastern and southwestern bounding strike-slip faults of which overlap considerably along strike (e.g., the 150° releasing sidestep pull-apart basin of Fig. 10). This more highly evolved stage of stepover fault development possibly is analogous to some overlapping elements of the Maacama and northern Rodgers Creek fault zones north of Santa Rosa. These faults of the Rodgers Creek and Maacama fault zones exhibit much longer lengths of overlap and more complex histories, however, than those in the sandbox models (Fig. 10).

Comparison to the laboratory models suggests that progressive development of along-strike overlap in the principal bounding faults of the Rodgers Creek–Maacama stepover system has resulted in the local development of pull-apart basins of different geometries at different stages in the lengthening of these faults. The models also suggest that at least some compressional structures adjacent to the pull-aparts may be coeval pop-ups or flower structures (Dooley and McClay, 1997). In contrast to this progression from non-overlapping (immaturely evolved) to substantial overlapping (maturely evolved) geometry seen in sandbox models, an immature, 30° non-overlapping pull-apart basin geometry is associated with the recently developed Santa Rosa pull-apart basin. This geometry is apparently related to reorganization of fault orientations in the stepover, reverting to a less mature stage of stepover development that is superposed on the more evolved stepover geometry seen in the overlapping relation between the northern Rodgers Creek (Healdsburg fault segment) and Maacama fault zones (Figs. 2, 3, 10, and 13).

This pattern of fault zone reorientation may, to first order, account for abandonment of some older segments of the southern Maacama fault zone and further provide the rationale for an apparent southwestward migration of the southern Maacama fault zone toward the Rodgers Creek and Bennett Valley fault zones during the recent stepover fault system reorganization. Overlapping faults of the northern Rodgers Creek and Maacama fault zones that evolved between ca. 3 and 1 Ma were overprinted by the immature non-overlapping stepover geometry of the Santa Rosa pull-apart basin (Figs. 3, 9, 10, and 13) after ca. 1 Ma as the result of this reorganization, which as discussed herein may have been in response to fault zone lengthening at the northern end of the fault system closer to the Mendocino Triple Junction, rather than a direct response to the releasing bend geometry of the San Andreas fault zone to the northwest.

1. The Rodgers Creek fault zone was initiated between ca. 7.3 and 6.7 Ma, when faulting splayed northeastward from a west-northwest–oriented proto-Hayward fault zone, forming a new zone of faults having northwest orientations. We interpret a distinctive breccia that is derived from extensional fault scarps along the east side of Santa Rosa Plain to mark the normal fault-bounded (transtensional?) margin of basins beneath the Santa Rosa Plain and the time of initiation of the Rodgers Creek fault zone. Extensional displacement on the early Rodgers Creek fault zone was replaced after ca. 5.4 Ma by compression and associated east-directed thrusting that uplifted the east side of the Santa Rosa Plain. The thrusting and compression partitioned the initial strike-slip basin beneath the Santa Rosa Plain into the separate Windsor and Cotati basins.

2. Composite strike-slip fault displacement for the northern and southern Rodgers Creek fault zones since ca. 7.0 ± 0.3 Ma is ≥28 ± 0.5 km, based on right-lateral separation of the fault scarp breccia between the Sears Point and Santa Rosa areas. This displacement is viewed as a minimum, because the southeastward extent of fault scarp breccia beneath Sonoma basin east of the southern Rodgers Creek fault zone is unknown. The Rodgers Creek fault zone slipped right-laterally at a median rate of ∼2.8 ± 1.1 mm/yr from the Late Miocene to early Pleistocene, but the rate has increased to ∼5.3 ± 2.2 mm/yr since the earliest Pleistocene. Low early slip rates reflect significant dip-slip components of displacement prior to 2.78 ± 0.02 Ma. A part of the southern Rodgers Creek fault zone may be partitioning slip toward the Maacama fault zone via the Spring Valley fault and the Bennett Valley fault zone at depth. However, Holocene surface faulting and earthquake distribution north of Santa Rosa indicate that an unconstrained component of slip is still taken up by the Healdsburg fault segment of the northern Rodgers Creek fault zone along its 40-km-long map overlap with the Maacama fault zone.

3. Similar surface displacement of the Sonoma Volcanics (20.8 ± 3.3 km) and basement rocks of the Mesozoic Coast Range Ophiolite (21.5 ± 0.5 km) indicate that the Maacama fault zone north of Santa Rosa was initiated at or soon after 3.17 ± 0.04 Ma and it has maintained a long-term slip rate of ∼5.5–7.9 mm/yr (median rate of 6.7 ± 1.2 mm/yr). Offset magnetic anomalies along the Maacama fault zone suggest a similar maximum displacement of 19 ± 7 km and median long-term rate of 6.0 ± 2.3 mm/yr. The slip rate since ca. 1.0 ± 0.2 Ma has been ∼6.3 ± 1.8 mm/yr.

4. The total contribution of the Rodgers Creek–Maacama fault system to slip of the East Bay fault system south of San Pablo Bay since 7.0 ± 0.3 Ma appears to be >48.4 ± 1.4 km, for a median long-term slip rate of at least 6.95 ± 0.85 mm/yr. Larger slip attributed to the East Bay fault system to the south results from slip contributed from faults east of the Hayward fault and to poorly constrained pre–7 Ma slip on the proto-Hayward fault zone north of Burdell Mountain and southwest of the Rodgers Creek fault zone.

5. We infer, from comparison to analogous laboratory generated sandbox models (Dooley and McClay, 1997), that the most recently organized geometry of the Rodgers Creek–Maacama fault system is an immature stage of stepover fault zone development characterized by north-northwest–oriented pull-apart basins with principal bounding faults that do not overlap. The immature stepover geometry is superimposed on an older geometry with principal bounding strike-slip faults having a more west-northwest orientation that overlap for ∼40 km along strike. Westward migration of the south end of the Maacama fault zone since the Pleistocene (ca. 1.2 Ma) may be the result of the superposition of these successive fault zone geometries. However, though the transtensional strike-slip basins of the Rodgers Creek–Maacama fault system have evolved within the realm of migrating major restraining and releasing bends of the northern San Andreas fault zone, the succession of the fault system reorganizations is not simply relatable to the migration of these bend geometries. Superimposed fault reorganizations with time at the southern end of the Rodgers Creek–Maacama fault system are probably a more direct kinematic response to the lengthening and reorganization of faulting at the northern end of the fault system, with northward migration of the Mendocino Triple Junction.

An early version of this manuscript was reviewed by R.G. Stanley and J.J. Rytuba of the U.S. Geological Survey, who suggested numerous changes that were incorporated into the paper. Two anonymous reviewers for Geosphere provided extensive additional comments and suggestions for reorganizing and improving the manuscript, prompting rethinking, refinement, and clarification of some of our interpretations. We thank many colleagues at the U.S. Geological Survey, including Elmira Wan, David Wahl, Carl Wentworth, Russ Evarts, Russ Graymer, Carol Prentice, and David Schwartz, for providing data, insights, worthwhile discussions, and encouragement during this research over the past 12 years. McLaughlin is indebted to Tor Nilsen (deceased) for raising an awareness of strike-slip–related sedimentary basins and providing many insights into their recognition and unique sedimentologic and tectonic characteristics.