The Eastern Belt of the Franciscan Complex in the northern California Coast Ranges consists of coherent thrust sheets predominately made up of ocean floor sediments subducted in the Early Cretaceous and then accreted to the overriding plate at depths of 25-40 km. Progressive packet accretion resulted in the juxtaposition of a series of thrust sheets of differing metamorphic grades. This study utilizes laser Raman analysis of carbonaceous material to determine peak metamorphic temperatures across the Eastern Belt and phengite barometry to determine peak metamorphic pressures. Locating faults that separate packets in the field is difficult, but they can be accurately located based on differences in peak metamorphic temperature revealed by Raman analysis. The Taliaferro Metamorphic Complex in the west reached 323-336°C at a minimum pressure of ~11 kbar; the surrounding Yolla Bolly Unit 215–290°C; the Valentine Springs Unit 282-288°C at 7.8±0.7kbar; the South Fork Mountain Schist 314–349°C at 8.6–9.5 kbar, a thin slice in the eastern portion of the SFMS, identified here for the first time, was metamorphosed at ~365°C and 9.7±0.7kbar; and a slice attributed to the Galice Formation of the Western Klamath Mountains at 281±13°C. Temperatures in the Yolla Bolly Unit and Galice slice were too low for the application of phengite barometry. Microfossil fragments in the South Fork Mountain Schist are smaller and less abundant than in the underlying Valentine Springs Unit, providing an additional method of identifying the boundary between the two units. Faults that record a temperature difference across them were active after peak metamorphism while faults that do not were active prior to peak metamorphism, allowing for the location of packet bounding faults at the time of accretion. The South Fork Mountain Schist consists of two accreted packets with thicknesses of 300 m and 3.5 km. The existence of imbricate thrust faults both with and without differences in peak metamorphic temperature across them provides evidence for synconvergent exhumation.

Accretionary wedges such as the Franciscan Complex of California represent the near-surface expression of the subduction zone interface, and as such have considerable geodynamic significance. The rheology of the accreted material may affect rates of subduction and the transfer of stress from the subducting to the upper plate [1]; and on shorter time-scales controls the frequency and magnitude of subduction zone seismicity [2]. Both solid and fluid materials are recycled on various scales in subduction zones by mechanisms that are still very much under investigation [39]. The thermal and mechanical structure of accretionary complexes, and the kinematics of deformation within them, are thus of both societal and geotectonic importance. Despite many decades of investigation, however, we still have a very poor understanding of some fundamental questions about these complexes. The thermal structure is in a state of extreme disequilibrium, with very low gradients parallel to the subduction zone, and at certain times and places, very high inverted gradients normal to the subduction zone interface (e.g., [10, 11]). Mechanisms of accretion at depth (underplating) are debated, in part because of uncertainties in the thermal structure, fluid pressure, metamorphic reactions, and the distribution and rheologies of the various rock types involved [12, 13]. The mechanisms by which rocks buried to depths of 50 km or more in subduction zones are exhumed are even more strongly debated, particularly as they are likely to disturb or fundamentally modify the internal structure of the accretionary complex [14]. For all these reasons, further structural, petrological, and geochemical investigation of accretionary complexes is required.

The Franciscan Complex of California was one of the first to be recognized as a fossil accretionary wedge [15], yet its internal structure and the distribution of metamorphic temperature and pressure are still very poorly defined. It consists primarily of ocean-floor sediments and lesser amounts of mafic material that were accreted both by large-scale frontal accretion and by underplating along the convergent margin of North America in Cretaceous to Paleogene time [16]. The underplated material experienced significant exhumation between 100 and 70 Ma [17, 18]. High pressure/low temperature metamorphism is widespread and, while extensive coherent terranes are present, especially in the eastern portion of the Franciscan, there are also significant bodies of mélange [19]. Structural analysis of the Franciscan is difficult due to the combination of complex structure and the monotonous lithological sequence of predominately metapelites and metagreywackes.

The recent development of multiple geothermometers based on laser Raman analysis of carbonaceous material (LRCM) has extended the lower bound of the temperature range that can be investigated with this method, making it a viable method for determining peak metamorphic temperatures in the low temperature Franciscan Complex. This study uses the geothermometer of Kouketsu et al. [20] together with phengite barometry to better constrain peak metamorphic pressure and temperature attained across multiple transects in the coherent terranes of the Eastern Franciscan Belt. This allows a much more detailed definition of the metamorphic zonation in this part of the Complex. In a region that consists of imbricate thrust packets of differing metamorphic grade, peak metamorphic temperatures can also be used to reveal the location of tectonic boundaries. The combination of improved structural constraints and improved temperature and pressure determinations also allows for more focused discussion of the processes of underplating and exhumation that have led to the present large-scale distribution of these units.

2.1. The Mesozoic Accretionary Margin in California

California is dominated by three NW-SE trending tectonic regions, the Franciscan Complex, the Great Valley Group, and the Sierra Nevada batholith. Created by the subduction of thousands of kilometers of sea floor beneath North America [21], these three zones are, respectively, the accretionary prism, forearc basin, and magmatic arc of the subduction zone [15, 22]. Subduction began in the Middle Jurassic (Wakabayashi, 1992; [23]) and ended with a transition to a transform boundary during the Neogene [24]. The E-dipping Coast Range Fault, which approximates the position of the paleosubduction zone, separates the Franciscan from essentially unmetamorphosed ophiolitic rocks and the Great Valley Group to the east [15]. The ophiolitic rocks to the east of the Coast Range Fault belong to the Coast Range Ophiolite [25] and to the Tehama-Colusa Melange, both deformed ocean floor units that are in fault contact with each other [26]. These ocean floor rocks have an uncertain origin and are overlain by the basinal sediments of the Great Valley Group to the east [27]. In some locations, there are small slices of low-grade metasediments between the Coast Range Fault and the ophiolitic rocks (Figure 1); these have been tentatively attributed to the Jurassic Galice Formation of the Western Klamath Mountains [28]. The Coast Range Fault was originally interpreted to be the subduction zone megathrust [15], but most workers have since recognized it to have been reactivated with a normal sense of motion, contributing to the exhumation of the underlying Franciscan [8, 2932]. Contrary to this view is the argument that exhumation was driven by erosion, rather than by extensional faulting [3335].

In the Coast Ranges of northern California and southern Oregon, the Franciscan has traditionally been divided into three belts, the Coastal Belt, the Central Belt, and the Eastern Belt (Figure 1), based on the timing of accretion, the grade of metamorphism, and the overall structural style [36, 37]. With the exception of the younger King Range terrane, the Coastal Belt was accreted between 65 and 45 Ma [38] and is only lightly metamorphosed [19]. It experienced a maximum temperature close to the closure temperature for fission tracks in apatite and cooled between 20 and 10 Ma [18]. Much of the Central Belt is composed of deformed pelitic material carrying distributed exotic blocks [36], fitting the definition of a mélange as proposed by Hsü [39] and, after being largely accreted between 95 and 88 Ma [38], attained metamorphic temperatures of 100-250°C and pressures of 3-10 kbar [40]. The Eastern Belt was accreted and metamorphosed in the Early Cretaceous; it is the oldest and highest grade of the three belts [17] and is described in more detail below. The peak temperatures of the three belts reveal a pattern of a metamorphic grade that increases to the east across the Franciscan as a whole. The Eastern and Central Belts both experienced significant synsubduction exhumation between 100 and 70 Ma, as determined by zircon fission track analysis [18, 41].

2.2. Eastern Belt

The Eastern Belt consists mainly of coherent thrust sheets with some intercalated mélange units. Metagreywackes and metapelites are the dominant lithologies and are accompanied by lesser amounts of metabasalt and chert. Metamorphic grade is thought to increase to the east internally across the Eastern Belt, as evidenced by changes in mineral assemblage and degree of quartz recrystallization. The change in mineral assemblage is predominately reflected by the presence of pumpellyite only in the west, and the presence of progressively coarser-grained lawsonite and sodic amphibole towards the east [4244]. The presence of epidote in only the easternmost margin of the Eastern Belt has also been noted as an indication of varied metamorphic grade [45]. Suppe [45] identified one unit, the Taliaferro Metamorphic Complex (TMC), that does not conform to the pattern of eastwardly increasing metamorphic grade. This is a fault-bounded body of blueschist-facies metasedimentary and metabasaltic rocks that is intercalated with lower grade lawsonite-albite facies metasediments. There are no reliable age data available for the TMC [17]. In contrast to the view that metamorphic grade increases regularly to the east, Bröcker and Day [46] have argued that, save for the epidote in the eastern SFMS and the unusual TMC, all variations in mineral assemblage within the Eastern Belt can be explained as a result of differences in composition. This view holds that while metamorphic grade increases across the Franciscan as a whole, it does not vary significantly within the Eastern Belt.

The Eastern Belt has been subdivided into the Yolla Bolly terrane to the west and the Pickett Peak terrane to the east [43]. Both terranes were metamorphosed under high-pressure low-temperature conditions as evidenced by the presence of lawsonite, blue amphibole, and sporadic jadeite [45, 4751], but the Yolla Bolly terrane is thought to be of somewhat lower grade (lawsonite albite facies) than the overlying Pickett Peak terrane (blueschist facies). The lowest estimated temperatures are 100-187°C ([43, 46], and references therein). The TMC lies within the Yolla Bolly terrane and was subjected to higher temperatures than the main body of lawsonite-albite facies rocks; its peak metamorphic temperature was estimated to be 300°C by Suppe [45], while Bröcker and Day [46] calculated 240°C–280°C based on phase equilibria. Peak pressures are thought to have been ~7 kbar [43] and 8.5-10 kbar ([46], and references therein) for the main body of the Yolla Bolly and the TMC, respectively. Ar-Ar dating of metamorphic white mica has placed the age of metamorphism of the main body of the Yolla Bolly at ~110 Ma, while the age of its deposition is interpreted to be ~111 in northern California based on detrital zircons [17, 52]; depositional ages are seen to decrease to ~100 Ma in the San Francisco Bay area and to ~89 Ma in the Nacimiento block further to the southeast [51, 53, 54].

The Pickett Peak terrane has been further subdivided into the Valentine Springs Unit in the west and the South Fork Mountain Schist (SFMS) in the east [43]. Both units consist predominantly of metagreywacke and pelitic schist, but the SFMS includes a body of metabasalt ~1 km thick, which was named the Chinquapin Member by Blake et al. [42]. Unit divisions within the Pickett Peak terrane, and within the Eastern Belt as a whole, were defined based on the textural characteristics of medium-grained greywackes. Blake et al. [42, 47] defined three textural zones, ranging from an unmetamorphosed appearance in hand sample to totally recrystallized. The South Fork Mountain Schist (SFMS), the easternmost unit, was defined as belonging to textural zone three, totally recrystallized. In those papers, no faults were described within the Eastern Belt, an interpretation which favored a gradational contact between the SFMS and underlying units. Subsequent work identified a fault along the basal contact of the SFMS [44, 45, 55]. Suppe [45] named this fault the Log Spring Thrust (LST) and identified an abrupt change in textural grade across the contact. Worrall [44] then identified additional faults and described a series of eastward dipping imbricate thrust faults as the defining structures of the Eastern Belt (Figure 2), also finding an abrupt change in textural grade across the LST. Suppe [45], Worrall [44], and Jayko et al. [43] all used subtly different definitions of the textural zones, and Jayko et al. [43] did not find an abrupt transition in textural grade across the LST. Disagreements over textural grade classifications can arise because different scales are used by different workers, because classifications are qualitative, and because the degree of quartz recrystallization depends not only on metamorphic grade but also on protolith grain size and intensity of deformation. In addition to disagreements over the nature of the basal contact of the SFMS, there are also disagreements over its location; within Thomes Creek (Figure 1) different workers have identified the LST in locations separated by as much as two kilometers [45, 56, 57].

The Valentine Springs Formation in the area of this study is dominated by metagraywacke sandstones that were deposited ~123 Ma, as determined by detrital zircon analysis [17]. The metamorphic age was determined by whole rock 40Ar/39Ar analysis, yielding an average age of 117 Ma [58]. Peak temperatures in the Valentine Springs Formation are thought to be 240–280°C [43, 46], while peak pressures are thought to be ~5.5–7 kbar [35, 43].

The SFMS is the oldest and highest-grade coherent unit in the Pickett Peak terrane and in the Franciscan. It is made up of pelites metamorphosed into quartz-mica schist as well as metagraywacke, metabasalt, and chert. Estimates for peak temperatures range from <200°C to 350°C, while peak pressure is estimated at ~7 kbar with upper pressure limits in the range of 8–10 kbar, based on the absence of jadeitic clinopyroxene ([43, 46, 48], and references therein). Detrital zircon data places deposition between 131 and 137 Ma while metamorphic ages for the SFMS were determined by 40Ar/39Ar step heating of white micas, yielding a weighted average age of 120.8±1.2Ma (2σ error) [17]. The white mica ages are interpreted as dating the growth of metamorphic white mica, with the differences in ages between the Valentine Springs and the SFMS being due to the progressive subduction and accretion of each unit, creating the current imbricate thrust geometry.

3.1. Temperature

Amorphous carbon becomes progressively more graphitized with increasing temperature, and this increase in crystallinity can be measured. Because the process is not reversed when the carbonaceous material (CM) is cooled, peak metamorphic temperature is recorded [59]. To measure the degree of crystallinity, a laser is focused on the sample. A portion of the incident photons induce vibrations in the molecular bonds of the sample, and the wavelength of the scattered photons is changed as a result. The change in wavelength is measured and different wavelength changes correspond to different vibrational modes. Measurement of amorphous carbon results in a larger number of vibrational modes than does measurement of graphite. When graphically plotted, these additional wavelength changes are termed defect bands and are numbered, e.g., D1, D2, etc. The wavelength change corresponding to graphite is termed the G band (Figure 3) [60], and references therein). Laser Raman analysis of CM was used to assess peak metamorphic temperatures across the Eastern Belt.

Samples of metapelite and metagreywacke were collected from streambeds across the Eastern Belt following the structural transects shown in Figure 1 and described by Schmidt and Platt [57]. The main focus of this study is the Thomes Creek transect and the Middle Fork Eel River transect; the Cottonwood Creek, Grindstone Creek, and Salt Creek transects sample the same units at different latitudes and were used to establish the regional applicability of our results. These supplemental transects were studied at a reconnaissance level, rather than in detail. The extent of the Thomes Creek transect was designed to capture the differences in metamorphic temperature between the Galice, the SFMS, and the Valentine Springs Units. The crucial boundaries that were investigated were the CRF, which separates the Galice from the SFMS, and the LST, as mapped by Schmidt and Platt [57], which separates the SFMS from the Valentine Springs. The Middle Eel transect was designed to reveal temperature differences between the TMC and the surrounding lawsonite-albite facies rocks. Samples were selected based on CM content and on location; multiple samples were collected from every identifiable fault-bounded tectonic block, with enough sample density to capture differences in peak metamorphic temperature across the unit.

As part of their work calibrating a geothermometer for contact metamorphism, Aoya et al. [61] explored the differences between two calibrated Raman geothermometers, one meant for use on deformed, regionally metamorphosed rocks and one for use on weakly or undeformed rocks that had experienced contact metamorphism. They found that, while there are meaningful differences between the calibrations for highly ordered CM, corresponding to higher peak metamorphic temperatures, this is not true for low temperature, disordered CM with R2 ratios (R2=D1/G+D1+D2 greater than 0.6. The two calibrations produce the same results at ~345°C, and differences between them are nearly negligible for disordered CM corresponding to investigated temperatures below ~375°C. A full study of the effect of deformation on Raman spectra is beyond the scope of this paper. As this previous work suggests that the effect of deformation is negligible for rocks that have experienced peak metamorphic temperatures consistent with those found in the Eastern Belt of the Franciscan, no consideration was given to the degree of deformation when collecting samples.

Samples were cut perpendicular to foliation and parallel to lineation before being investigated using LRCM within the samples. All analysis points were selected to be slightly below the surface of the thin section, to avoid analyzing CM damaged by polishing [62]. Analyses were carried out on a Renishaw M1000 Micro Raman Spectrometer System at the California Institute of Technology using a 2400 lines/mm grating and a 514 nm green laser, with collection times of 30 seconds. Laser power on the sample surface was 2-3 mW. Peaks were then deconvolved using the computer program PeakFit 4.12 (SeaSolve Software Inc.). All samples in this study had a minimum of 12 analysis spots. Absolute error for Raman analysis is typically taken as ~50°C (e.g., [60]). Errors reported in this paper are measurement errors, reported at 1σ based on the ~12 measurements per sample, and errors reported in figures are as described in captions.

This study processed the Raman spectra following the method developed by Kouketsu et al. [20]. It was important for this study to use a single geothermometer to assess the entire range of recorded temperatures, as a major goal of our work is locating abrupt changes in recorded temperature across faults. If separate geothermometers are used on either side of a fault, it would be possible to ascribe the difference in recorded temperature to differences between competing calibrations. When using a single calibration, the relative difference in temperature between two areas is independent of the accuracy of the absolute temperature and, as such, is a more reliable indicator of a real difference in peak metamorphic temperature. There are four regional metamorphism calibrations to consider. The original Raman thermometer developed by Beyssac in 2002 is applicable in the range of 330°C–650°C and is inappropriate for the lower temperature rocks of the Eastern Belt. The calibration of Lahfid et al. [63] is useful in the range of 200°C–320°C, a range that is too small to capture the higher temperature portions of the Eastern Belt. Rahl [64] developed a calibration for the range 100°C–700°C. Because there is not a unique solution when deconvolving the G and D2 bands with this method, it is inappropriate for some samples and was not used. In contrast, Kouketsu et al.’s [20] thermometer is applicable in the range of 150°C–400°C, sufficient for capturing expected Eastern Belt temperature variation. It provides two methods for determining temperature, one based on the full width at half maximum height (FWHM) of the D1 band and one based on the FWHM of the D2 band. FWHM-D1 is the preferred method for temperatures between 200°C and 400°C, and it was used for all calculations in this study.

3.2. Pressure

The Si content of white mica generally increases with increasing pressure from 3 Si atoms per formula unit (apfu) in muscovite to 4 Si pfu in celadonite via the Tschermak substitution Fe,MgSiAl1VIAl1IV, allowing pressure to be determined based on the Si content of phengite [65]. The Si content of white mica was analyzed at the University of California Los Angeles, using a JEOL JXA-8200 electron microprobe equipped with five wavelength-dispersive spectrometers. Analyses were performed using the following conditions: an accelerating voltage of 15 kV, a sample current of 15 nA, and a spot size of 5 μm. Correction of measured intensities used the ZAF method. Natural and synthetic standards were used to calibrate the microprobe; for major oxides (>0.2 wt %), standard deviations were ≤8.65% of the total measured value. Quantitative compositional X-ray maps were made of phengite grains to check for the presence of compositional zoning, and were made on the same machine using the following analytical conditions: 15 kV accelerating voltage, 160 ms dwell time, 2 μm step size, and a 95-150 nA sample current. Data analysis and calibration were done using the program XMapTools 3.3.1 [66, 67]. Standardization was done using high-quality measurements performed within the X-ray map area [68]. All Si apfu reported or discussed in this study are based on 11 oxygen atoms.

To calculate a pressure from a measured Si apfu, Si apfu isopleths were plotted on pseudosections along with temperatures determined by laser Raman; the intersection of a calculated temperature with a calculated Si apfu isopleth in P-T space then gives a pressure that is based on an assumption that peak temperature and peak pressure occurred simultaneously. Pseudosections were constructed with the software package Perple_X 6.8.3 ([69, 70], downloaded from the internet site for the system MnO-Na2O-CaO-K2O-FeO-MgO-Al2O3-SiO2-H2O-TiO2. The thermodynamic data set of Holland and Powell ([71], with 2002 updates) was used, as was the fluid equation of state (CORK) of Holland and Powell [72]. The following thermodynamic solution models were used: for clinoamphibole, GlTrTsPg [73, 74]; for biotite, Bio (HP) [75]; for chlorite, Chl (HP) [76]; for chloritoid, Ctd (HP) [77]; for clinopyroxene, Omph (HP); for orthopyroxene, Opx (HP) [78]; for feldspar, feldspar [79]; for garnet, Gt (HP); for phengite, Pheng (HP); and for staurolite, St (HP) [71]. An ideal solution model was used for ilmenite (IlGkPy). Excluded components were as follows: diopside (di) and stilpnomelane (stlp, msnp, and mstl). H2O was treated as existing in excess. The obtained pseudosections were contoured with Si apfu isopleths using the Perple_X subprograms werami and pstable.

Construction of pseudosections required whole rock bulk compositions for each of the six analyzed samples. Whole rock bulk compositions were measured using a Bruker M4 Tornado Micro-XRF Spectrometer at California Institute of Technology. Bulk composition data was slightly modified prior to pseudosection calculations. P2O5 was ignored and, because all P was assumed to be bound to apatite, a corresponding amount of CaO was also removed. All Fe was treated as divalent. Bulk compositions are reported in Table 1.

4.1. Temperature

4.1.1. Thomes Creek

Peak metamorphic temperatures vary spatially across the transect. The highest temperatures obtained are in the eastern portion of the transect, while the lowest temperatures are in the western portion; the full temperature range is 282-377°C (Figure 4). The differences in temperature are also seen qualitatively in changes in Raman spectra across the transect (Figure 5). Differences in peak metamorphic temperature occur across some mapped faults, but not across all faults. The CRF separates the Galice slice, which attained a temperature of 281±13°C, from a small slice of SFMS which reached temperatures as high as 377°C. An unnamed thrust zone that duplicates a thin slice of mafic schist separates this small, high-temperature slice from the main body of the SFMS, which experienced temperatures of 314-349°C. The previously estimated range for SFMS temperatures was 200-350°C (Bröcker and Day, and references therein). Our temperature results that are higher than these previous estimates are confined to the small, high-temperature slice of the SFMS in the eastern portion of the transect. Our results for the main body of the SFMS, 314-349°C, are entirely within the upper limits of the previously estimated range.

The lower boundary of the SFMS in Thomes Creek is a fault oriented at 56/025 (dip/dip direction) that separates the SFMS from underlying Valentine Springs rocks with very consistent peak temperature estimates of 282°-288°C, in good agreement with the upper limit of the previously estimated range of 240-280°C [43, 46]. Schmidt and Platt [57] interpreted the fault as a normal fault that cuts the original LST, based on its steep orientation and the omission of a SFMS mafic blueschist body at this point in Thomes Creek. Foliation orientations in the vicinity of the fault are highly variable, and broken formation is present in the disrupted zone on both sides, but the rocks structurally above the fault are distinguishable from the rocks below it by the more strongly differentiated foliation, produced by pressure solution [57]. Further below the LST, to the west, the peak metamorphic temperature does not vary significantly within the Thomes Creek transect, although Schmidt and Platt [57] noted significant changes in structural style, and a decrease in overall intensity of the deformational fabric. The western four samples in the Thomes Creek transect produced Raman spectra that could not be accurately fit with the D4 band fixed at 1245 cm-1 as prescribed by Kouketsu et al. [20], raising the possibility that these four temperatures may be erroneously high despite the fact that they closely match nearby temperatures to the east.

Two of the major faults mapped by Schmidt and Platt [57] do not separate rocks of different peak metamorphic temperatures. There is no recorded change in temperature across the Tomhead Fault, which bisects the Chinquapin Member, or across the fault at the base of the Chinquapin, which places Chinquapin metabasalts structurally above the ocean floor sediments of the remainder of the SFMS. The temperature structurally above the Chinquapin is 321±12°C, while the temperature just below the Chinquapin is ~320°C.

4.1.2. Middle Fork of the Eel River

The Middle Eel transect crosses the TMC and captures the temperature difference between it and the surrounding lawsonite-albite facies rocks of the Yolla Bolly Unit (Figure 6). The TMC reached a peak metamorphic temperature of 323-336°C, while rocks to the south reached temperatures in the range of 282-293°C. Rocks to the north reached temperatures ranging from >280°C at the boundary with the TMC to ~217°C a short distance north of the boundary. One additional sample was collected from metasediments structurally overlying a slice of TMC in Beaver Creek (39°56.23N, 122°59.23W), a tributary of the Middle Eel, ~11 km north of the Middle Eel transect. It records a temperature of 219±21°C, similar to the lowest temperatures from the Yolla Bolly unit in the main transect. Previously estimated temperatures for the lawsonite-albite facies rocks of the Yolla Bolly and the TMC were both slightly lower than our temperatures, at 187°C and 300°C, respectively [43, 45, 46].

4.1.3. Cottonwood Creek

The Cottonwood Creek transect (Figure 7) examined the extreme eastern and western portions of the section and yielded results similar to those seen in Thomes Creek. Peak temperatures in the easternmost Cottonwood Creek SFMS are ~370°C, similar to the thin, high-temperature slice on the eastern margin of the SFMS in Thomes Creek. Metamafic and metasedimentary rocks are intercalated by minor thrust faults a short distance west of the high-temperature zone in Cottonwood Creek, but we have not examined these for temperature differences. The western limit of the Cottonwood Creek transect records temperatures of ~280°C, similar to those determined for the Valentine Springs Unit in Thomes Creek.

4.1.4. Grindstone Creek and Salt Creek

Grindstone Creek samples range in temperature from 266 to 304°C, but the temperatures do not regularly increase or decrease to the east or west (Figure 8). This transect did not cross any known or observed faults. One sample was also analyzed from Salt Creek, to the south of Grindstone Creek. The sample was collected from just west of the Coast Range Fault (39°38.07N, 122°36.33W) and it yields a temperature of 306±9°C.

4.2. Pressure

Microprobe analyses of white mica oxides add up to ~95 weight percent, while chlorite oxides add up to ~88 weight percent. Measurements of white mica with totals below 93 weight percent were interpreted to have been contaminated with intergrown chlorite and were discarded. Some measurements also revealed a Cr component in the white mica. Cr can exist in a variety of valence states including Cr2+ and Cr3+. Cr2+ has the same charge and a similar ionic radius as Fe2+ and Mg2+, making it likely to also participate in a substitution similar to the Tschermak substitution, CrSiAl1VIAl1IV. Cr3+ shares the same valence state as Al3+ and so may be able to substitute directly for Al3+. Because the role of Cr in white mica substitution is somewhat unclear, measurements including Cr in excess of 0.2 weight percent were discarded, consisting of three measurements of sample SS10 and four measurements of sample SS53. The remainder of the probe data results are reported in Table 2 with standard deviations.

Quantitative compositional X-ray maps were also affected by the intergrowths of other phases, especially chlorite. Where intergrowths of other phases in phengite are too small, the XMapTools software interprets the measurement as lower or higher Si apfu phengite, without recognizing the coexisting second phase (Figure 9). As a result, the compositional maps must be compared to a backscatter image to avoid erroneously interpreting intergrowths of other phases as compositional zoning. As an example, the map of SS118 captures a phengite grain which is parallel to the main foliation of the Valentine Springs Unit and which has been slightly folded. Small intergrowths visible in the backscatter image show as pixels of phengite with a lower or higher Si apfu, depending on the composition of the additional phase. There are no variations in Si apfu away from the inclusions, and there is no overall compositional zoning (Figure 9). Similar results are seen for samples SS69 (Figure 10), SS53 (Figure 11), and SS1 (Figure 12).

The map of SS10 shows multiple phengites, aligned with both a folded S2 and with an incipient S3. The phengite grains have varying amounts of additional included phases. There is no compositional variation across phengites of either population, and Si apfu values are the same for S2 and S3 aligned phengites (Figure 13). These two foliations were described by Schmidt and Platt [57] and interpreted to have formed during or shortly after accretion, based on lawsonite grains which formed before or during the creation of the main S2 foliation and on lawsonite porphyroblasts with rotational inclusions which indicated continued growth during D2. This, in conjunction with the lack of compositional zoning in all SFMS and Valentine Springs samples, leads us to interpret their measured phengite compositions as reflecting peak or near peak pressure conditions.

Calculated pseudosections closely match the dominant mineral assemblages for each sample. Because S and P were not included in the modeled system, pyrite and apatite are not predicted by the modeling, despite being common in the thin section. A pressure was not obtained for the Galice formation, as the majority of Galice white micas visible in the thin section were equal to or smaller than the microprobe spot size and we were unable to obtain reliable compositional analyses as a result. We were also unable to calculate a reliable pseudosection for the TMC. Calculated and observed assemblages do not match, suggesting disequilibrium, and exploratory phengite measurements revealed two populations with Si apfu of 3.56 and 3.61. An estimate of minimum pressure is possible based on the presence of jadeite in the TMC, which at the measured temperature of 330°C indicates a minimum pressure of ~11 kbar.

4.2.1. Sample SS118, Upper Valentine Springs

The pseudosection produced for SS118 (Figure 14) accurately predicts the dominant mineral assemblage. SS118 contains quartz, chlorite, phengite, albite, sphene, pyrite, and apatite. Phases with constituents that were not modeled, pyrite, and apatite, are necessarily absent from the pseudosection. Paragonite and rutile were predicted, but not observed. The prediction of paragonite is seen in all other modeled samples in this study as well. In all cases, it is predicted as an accessory phase and, while it may have been erroneously predicted, it is also possible that it was not observed because it exists as a minor amount of small crystals that are optically quite similar to the ubiquitous phengite. The result is a calculated pressure of 7.8±0.7kbar, from just beneath the Log Spring Thrust, a result that is somewhat higher than the previously estimated pressures of 56.7kbar for the Pickett Peak terrane [43].

4.2.2. Samples SS69, SS53, and SS10, Main Body South Fork Mountain Schist

Sample SS69 contains quartz, chlorite, phengite, albite, lawsonite, sphene, and pyrite. The corresponding pseudosection (Figure 15) predicts a matching mineral assemblage, minus pyrite and with the addition of paragonite, as discussed above. The resulting pressure is 9.2±0.6kbar. Sample SS53 contains the same mineral assemblage as SS69 and also has a pseudosection (Figure 16) that predicts the observed assemblage minus pyrite and plus paragonite. SS53 reached a pressure of 8.7±0.8kbar. SS10 has Si apfu isopleths that can produce multiple intersection points with the Raman temperatures; however, all but the lowest pressure of these plot within the jadeite stability field (Figure 17). As SS10 was not observed to contain jadeite, and jadeite has never been reported in the SFMS, we consider the lower pressure intercept to produce the most accurate pressure and to indicate the calculated mineral assemblage to which the actual mineral assemblage should be compared. Sample SS10 contains quartz, chlorite, phengite, albite, lawsonite, sphene, and pumpellyite. The pumpellyite exists as a fine-grained replacement of an earlier phase and is present on the pseudosection at lower pressures; as such, it is interpreted as resulting from retrograde growth and is not part of the peak assemblage. Actinolite was predicted as an accessory phase, at 1.46 weight percent, yet was not observed. The resulting pressure for SS10 is 8.4±1.5kbar. Because no major faults separate these samples, we interpret them to have experienced the same peak pressure and take the pressure of the main body of the SFMS to lie within the overlapping error of each of the three samples, 8.6–9.5 kbar. Our results exceed the pressure of 6.5-7.5 determined by Bröcker and Day [46].

4.2.3. Sample SS1, High Temperature South Fork Mountain Schist

The observed mineral assemblage of SS1 is quartz, chlorite, phengite, albite, sphene, and apatite. The pseudosection (Figure 18) predicts this assemblage, minus apatite as discussed above. It also predicts paragonite and a trivial amount of lawsonite, 0.35 weight percent, neither of which was observed. Sample SS1 reached pressures of 9.8±0.7kbar. There are no previous estimates of pressure for this small, fault-bounded slice of metasediments in the eastern portion of the SFMS. This study is the first to identify it and to determine that it is metamorphically distinct from the main body of the SFMS.

4.3. Fossils

Microfossils of unknown identity, composed exclusively of carbon, and only observable in thin sections under reflected light, are common in the metasediments of the Pickett Peak terrane. These fossils exist as fragments ranging in size from ~50 to 1200 μm, and their primary features are regularly spaced holes, giving them an appearance similar to that of some diatoms. However, the microfossils are made of C, rather than of silica, as determined optically and by laser Raman analysis. Some fossil fragments have quartz growing in pressure shadows along their margins. The size and abundance of these fossils are different in adjacent tectonic units. Fossils found in SFMS samples tend to be smaller than the fossils in the underlying Valentine Springs Unit (Figures 19(a)–19(c)). Additionally, while fossils were observed in ~58% of Thomes Creek Valentine Springs samples, they were found in only ~14% of SFMS samples in the same transect. This change in size and abundance coincides exactly with the discontinuity in peak metamorphic temperature revealed by laser Raman analysis (Figure 4). Fossil abundance was also examined within the Grindstone Creek transect and is similar to that seen in the Valentine Springs, as fossils are large and are present in ~42% of the examined samples (Figures 4 and 8).

5.1. Temperature

Our results from Grindstone Creek and Salt Creek are somewhat surprising, as these locations recorded temperatures lower than those seen in the main body of the SFMS in Thomes Creek, despite being mapped as a continuation of that unit. It is possible that this is the result of a north-south variation in peak metamorphic temperature within the SFMS, but it is also possible that a lower grade unit, possibly the Valentine Springs, has been incorrectly mapped as SFMS. Our sampling transect was designed to cross the boundary between the SFMS and the Valentine Springs, but we neither observed a fault in Grindstone Creek nor discovered a jump in peak metamorphic temperature similar to the difference observed across the LST in Thomes Creek. In addition to the similarity between the Grindstone Creek temperatures and the temperatures recorded by the Valentine Springs in Thomes Creek, the two units have microfossils of similar size and abundance. These two similarities lead us to suggest that the area of Grindstone Creek mapped as SFMS may be more appropriately ascribed to the Valentine Springs, but a full analysis of this unit designation is beyond the scope of this paper.

Our Cottonwood Creek transect assessed temperatures near the eastern and western limits of the SFMS. The temperatures of ~280°C in the west match those determined for the Valentine Springs in Thomes Creek; it is likely that the LST lies east of these samples. The high temperature of 375±5°C in the east of Cottonwood Creek is nearly identical to the measured temperature of 377±4°C at the eastern limit of the SFMS in Thomes Creek. A zone of faulting, which juxtaposes metasediments and metamafics, was discovered to the west of this location and, as it defines a small, high-temperature slice in the easternmost portion of the transect, it is likely that this is the continuation of the fault in Thomes Creek that separates a thin slice of high-temperature SFMS from the main body of the unit (Figure 19(d)). These results suggest that temperature patterns revealed in the main Thomes Creek transect extend for tens of kilometers to the north.

At the northern boundary of the TMC, the temperature decreases abruptly from >320°C to 287±7°C and then to 215±15°C in the lawsonite-albite facies rocks over a horizontal distance of less than half a kilometer. The systematic decrease in peak temperature away from the TMC boundary could be explained as a result of conductive heat transfer from the TMC into colder Yolla Bolly rocks emplaced beneath it. In contrast, at the southern boundary, the peak temperature of the lawsonite-albite facies rocks was determined to be 293±11°C a full 0.75 km from the boundary and temperatures in the 215-219°C range were not observed.

Temperatures elsewhere in the Franciscan are generally in good agreement with our results. Lahfid et al. [80] used LRCM to determine peak metamorphic temperatures of the Lucia subterrane near San Simeon, finding that temperatures ranged from 220 to 315°C. In northern California, Underwood [81] used vitrinite reflectance to assess temperatures over multiple Franciscan terranes. Because the calibration used returned higher temperature values than competing calibrations, the results were considered maximum estimates. Underwood [81] reported an average temperature of 190°C for the Central Belt and an average temperature of 140°C for the Coastal Belt, not including the anomalously higher temperature of the King Range terrane, which was interpreted to have experienced a later hydrothermal overprint. Underwood [81] also found that, within the Central Belt, temperatures were higher in the east and lower in the west. The Central Belt was seen to have experienced temperatures of ~250°C near the border with the Eastern Belt and ~150°C near the border between it and the Coastal Belt. The Coastal Belt results were somewhat more complicated, but much of the Coastal Belt was seen to have experienced temperatures of 100-150°C. The temperatures of 215-219°C determined for the western portion of the Eastern Belt in this study are quite similar to the temperature of ~250°C determined by Underwood [81] for the eastern limit of the Central Belt. The slight discrepancy between these two results may be due to the different methods used, or to the fact that Underwood [81] employed a calibration that produced a maximum temperature estimate. Our results clearly show that temperatures increase to the east across the Eastern Belt and a comparison of them with previously determine temperatures provides additional evidence for an eastward increase in temperature across the northern Franciscan as a whole.

5.2. Pressure

An important assumption made when calculating pressures from Si apfu isopleths for this paper was that peak pressure and peak temperature coincided with each other. Previous studies have revealed that high-grade blocks in the Franciscan have experienced counter-clockwise P-T-t paths, some of which have experienced widely separated peak temperatures and pressures [8284]. However, these blocks are likely to have experienced a different thermal history than the coherent blueschist facies metasediments, as they were emplaced within a subduction zone that was still cooling slowly due to a low convergence rate [85]. In contrast to this, the metasediments were emplaced long enough after subduction initiation for the zone to be in a thermally mature steady state. Ernst [86] argues that the preservation of a blueschist facies mineral assemblage without a significant retrograde overprint requires that the retrograde path approximately retraced the prograde path. A P-T-t path such as this produces a peak temperature and peak pressure that occur simultaneously or near simultaneously. Evidence for such a P-T-t path has been found by Radvanec et al. [87], giving support to our assumption that peak temperature and pressure were nearly synchronous.

5.3. Exhumation

The existence of differences in peak metamorphic temperature across mapped faults provides information on the relative timing of the faulting, as these faults must have been active after peak metamorphism. Units of different temperatures that were juxtaposed with each other prior to peak metamorphism would have thermally equilibrated during peak metamorphism. Using this as a guide, faults that record different temperatures on either side are interpreted as having been active after peak metamorphism on the higher grade side of the fault, while faults that do not record such differences in temperature are interpreted as either having been active prior to peak metamorphism or as having a small amount of offset. The LST, the easternmost mapped thrust within the SFMS, and the Coast Range Fault were all active after peak metamorphism in the SFMS. The Tomhead Fault has duplicated the originally ~450 m thick Chinquapin metabasalt member [57] and appears to have accommodated significant offset, and yet records no differences in peak metamorphic temperature on either side. It is therefore interpreted as having been active prior to peak metamorphism, as is the fault that places the Chinquapin structurally above the main bulk of the SFMS sediments. The coexistence of these subparallel, imbricate thrust faults that were active both prior to and after peak metamorphism provides strong evidence for synconvergent exhumation that proceeded without altering the local kinematics of thrusting and underplating, suggesting a widespread regional exhumation mechanism. Continual underplating during regional exhumation is consistent with the lack of a greenschist facies overprint in the Eastern Franciscan, as subduction refrigeration maintained depressed geothermal gradients during exhumation.

Exhumation of deeply subducted material could have been driven by a number of different mechanisms, including extension of the accretionary wedge caused by underplating at its base [8]; steady state return flow in a parallel-sided subduction channel, driven by buoyancy [6, 88] or topographic gradients in the forearc [89]; forced return flow in a downward-closing subduction channel [5, 6, 90]; and buoyancy contrasts between subducted material and overlying rock [9, 91, 92]. The evidence for episodic accretion of discrete thrust slices in the Eastern Belt, and the absence of evidence for a reversal of the sense of shear across the SFMS, for example, as observed by Xia and Platt [89] in the Pelona Schist, suggest that viscous return flow was not the primary mechanism of exhumation. Previous work has identified extensional faulting within the Franciscan Eastern Belt [57], and so we favor interpretations that include synconvergent extension. Platt [8] suggested that underplating directly beneath the wedge increases its thickness and taper, resulting in extension in the upper rear of the wedge. Warren et al. [9] advocated buoyancy-driven flow leading to high rates of return from deep within the subduction zone (~100 km); the exhumed material then reaches the base of the accretionary wedge and triggers extension at shallower levels. Both of these models can explain the observed synconvergent extension.

5.4. The Subduction Interface

The nature and thickness of the subduction interface vary with a number of factors, including depth, types of lithologies present, fluid migration, and subduction geometries and rates ([12], and references therein). At shallow levels, the interface has been observed both directly via drilling [93] and indirectly in exhumed complexes [16, 94], revealing it to generally be ≤~300 m thick. An extensive compilation of underplated slab thicknesses shows that this thickness is common even at greater depths [12], though some workers have reported interfaces on the order of 2–10 km thick [89, 9597] and these greater thicknesses are consistent with the interface as revealed by some seismic imaging (e.g., [98101]).

For the purpose of this discussion, we assume that the subduction interface width in the SFMS is roughly equal to the thickness of the most recently accreted packet. Previous workers have found that subduction interfaces are bordered by basal and roof shears ([102], and references therein), and the packet bounding faults identified in this study are good candidates for these shears as they accommodated subduction-related deformation at the time of accretion. We have no constraints on whether or not motion ceased on structurally higher bounding faults after the activation of structurally lower faults, and it is likely that deformation extended for some distance beyond the bounding faults of a given accreted packet, but this assumption allows for a first order estimation of the interface thickness. Packets that were both emplaced within an actively exhuming complex and that experienced near synchronous peak pressures and temperatures must have experienced these peak conditions at or immediately after the time of accretion. As such, faults that lack a temperature difference across them are seen to have ceased activity prior to accretion and are eliminated as candidates for packet bounding faults at the time of accretion; this includes the Tomhead Fault and the fault which places the Chinquapin structurally above the bulk of the SFMS metasediments. The remaining faults, the CRF, the eastern most fault within the SFMS, and the LST, are interpreted as packet bounding faults at the time of accretion. This divides the SFMS at this latitude into two packets, an ~300 m thick packet to the east and an ~3.5 km thick packet to the west. The high-temperature slice of the SFMS, at ~300 m structural thickness, closely matches those most commonly reported by Agard et al. [12], while the western packet is significantly thicker. It is possible that the thin eastern packet was originally thicker before being subjected to subduction erosion, but in the absence of evidence for this, we interpret it to have simply been a narrower interface. Our results indicate that subduction interface thickness can vary from a few hundred meters to a few kilometers, even within a single subduction zone. This is somewhat striking, given that the two packets were accreted under quite similar conditions. Lithologic control over different accretion styles is suggested by the proximity of the ~1 km thick Chinquapin metabasalt member to the boundary between the two slices as well as by the lack of any significant metabasalt or coarse-grained metagreywacke component in the high-temperature slice. It is possible that the difference in accreted packet thickness could be due to some combination of the rheological differences caused by the presence or absence of the metabasalts and greywackes or by differing fluid content at the time of accretion.

5.5. Underplating

Underplating is thought to be facilitated by mantle scale changes in plate dynamics and transient changes to mechanical coupling ([12] and references therein), especially by dewatering of subducted sediments leading to a lower rheological contrast between them and the overriding plate [90, 102, 103]. While extensive veining is present in the SFMS, indicating large scale fluid flow consistent with a dewatering model, this study does not provide the observations needed to differentiate between different underplating mechanisms. After underplating, Franciscan Eastern Belt sediments occupied the base of an actively exhuming wedge and so peak metamorphic conditions were experienced close to the time of accretion. Due to progressing exhumation, each successive packet, despite being accreted at a structurally lower level, was accreted under lower temperature and pressure conditions than the preceding packet. This model adequately explains the stepwise variations in metamorphic grade, metamorphic age, and depositional age across the eastern Franciscan, and this pattern of accretion is supported by analyses of the Central and Coastal Belts as well [104].

How the TMC was emplaced within the lawsonite-albite facies rocks is an outstanding question, and it appears that it must have been exhumed, part way at least, before or during the subduction and accretion of the Yolla Bolly unit. At present, we lack reliable data on both the protolith age and the timing of metamorphism of the TMC, which hinders further discussion.

Peak metamorphic temperatures revealed by laser Raman analysis of carbonaceous material can be used to identify major faults and tectonic boundaries within the Franciscan. Metamorphic temperatures in the Eastern Belt decrease from east to west, and abrupt changes in peak metamorphic temperature are observed across mapped faults. The timing of faulting can be constrained by the existence or lack of a difference in peak metamorphic temperature on either side of the fault. The Log Spring Thrust, the unnamed thrust fault zone in the eastern SFMS, and the Coast Range Fault were all active after peak metamorphism and are the bounding faults of accreted sediment packets, while faults without a temperature difference on either side likely ceased activity prior to accretion. The SFMS consists of two accreted packets of significantly different thicknesses, ~300 m and~3.5 km, and these are taken to be the thickness of the subduction interface at the time of accretion. Sediments were accreted to the base of an actively exhuming wedge, as evidenced by subduction related faults that were active both before and after peak metamorphism.

Data not included in this manuscript or its supplementary file can be made available upon request.

The authors declare no conflicts of interest.

This research was funded in part by NSF grant EAR-1250128 to J. Platt. We are grateful to Daniel Platt, Daniel Schmidt, and Francisco MeldeFontenay for their help in the field, and to Tom MacKinnon for allowing us to work on his samples from Grindstone Creek. We would also like to thank Estibalitz Ukar, Gary Ernst, and Pierre Lanari for their detailed and constructive reviews of the manuscript.

Exclusive Licensee GeoScienceWorld. Distributed under a Creative Commons Attribution License (CC BY 4.0).