An Exhumation Pulse From the Nascent Franciscan Subduction Zone (California, USA)

We investigated a suite of metabasite blocks from serpentinite matrix and shale matrix mélanges of the California Coast Ranges. Our new data set consists of 40Ar/39Ar dates of amphibole and phengite and U‐Pb dates of metamorphic zircon. Combined with published geochronology, including prograde Lu‐Hf garnet ages from the same blocks, we can reconstruct the timing and time scales of prograde and retrograde metamorphism of individual blocks. In particular we find that exhumation from amphibole‐eclogite facies conditions occurred as a single episode at 165–157 Ma, with an apparent southward younging trend. The rate and timing of exhumation were initially uniform (when comparing individual blocks) and fast (with cooling rates up to ~140°C/Ma). In the cooler and shallower blueschist facies, exhumation slowed and became less uniform among blocks. Considering the subduction zone system, the high‐grade exhumation temporally correlates with a magmatic arc pulse (Sierra Nevada) and the termination of forearc spreading (Coast Range Ophiolite). Our findings suggest that a geodynamic one‐time event led to exhumation of amphibole‐eclogite facies rocks. We propose that interaction of the Franciscan subduction zone with a spreading ridge led to extraction of the forearc mantle wedge from its position between forearc crust and subducting crust. The extraction led to fast and uniform exhumation of subducted rocks into the blueschist facies. We also show that the Franciscan subduction zone did not undergo significant cooling over time and that its initiation was not coeval with blueschist‐facies metamorphism of the Red Ant schist of the Sierra Nevada foothills.


Introduction
Metamorphic histories of subducted and exhumed rocks reflect a range of thermomechanical processes operative in subduction zones (Coleman & Lanphere, 1971;Ernst, 1973). Data from these rocks provide quantitative constraints on volatile recycling (e.g., Hacker, 2008;Hacker et al., 2003), plate-interface rheology (e.g., Agard et al., 2018, and references therein), and chemical-physical responses to changes in subduction-zone boundary conditions (e.g., Bebout, 2007). Quantifying these processes relies on a robust knowledge of subduction-zone thermal structure and its global variation (Syracuse et al., 2010), but subduction zone pressure-temperature (P-T) conditions determined on exhumed rocks exhibit a thermal structure that may not match results obtained from numerical modeling (Penniston-Dorland et al., 2015). Such discrepancies suggest that modeling approaches and/or rock observations are incomplete, for example, either by model (Kohn et al., 2018) or rock bias (van Keken et al., 2018). One way to assess the bias inherent to the subducted rock record is to examine variations in P-T-t histories within individual subduction complexes. These data suggest that most subduction zones record synchronous, punctuated exhumation, indicating that exhumation is likely driven by changes in boundary conditions (Agard et al., 2009) and may not represent a thermal steady state. On the other hand, some studies have determined lengthy residence times for individual blocks prior to exhumation (Tamblyn et al., 2019), such that comprehensive P-T-t data from multiple blocks may expose the time-resolved thermal structure, or metamorphic field gradient, of an individual subduction zone.
The Franciscan Complex is regarded as the classic example of a subduction-accretion complex, with a history of continuous subduction exceeding 150 Ma (Hamilton, 1969;Mulcahy et al., 2018;Wakabayashi, 2015). The oldest metamorphic ages presumably date subduction initiation, initially thought to be~150 Ma (Coleman & Lanphere, 1971), since revised to~169 Ma (Anczkiewicz et al., 2004), and recently determined to be~180 Ma (Mulcahy et al., 2018). These old ages stem solely from high-grade, amphibole-eclogite facies blocks in lower-grade, shale or serpentinite matrices that make up <1% of the Franciscan Complex but are the only material record of subduction zone metamorphism from~180 to 132 Ma; at~123 Ma there was a significant change in both the volume and type of accreted material (Figure 2a; Dumitru et al., 2010;Wakabayashi & Dumitru, 2007) (Dumitru et al., 2010). There is a wide range of hornblende and phengite K-Ar and 40 Ar/ 39 Ar dates (see Wakabayashi & Dumitru, 2007, for a review) that may indicate continuous exhumation of amphibolites and eclogites over a~20-30 Ma period. However, despite decades of research on the Franciscan Complex, coupled geochronology (U-Pb zircon, Lu-Hf garnet) and thermochronology ( 40 Ar/ 39 Ar hornblende, 40 Ar/ 39 Ar phengite) have rarely been performed on the same blocks-such that published radiometric ages are often sourced from different blocks with potentially distinct metamorphic histories.

Geologic Setting: Franciscan Subduction System
The Franciscan subduction system comprises the Franciscan accretionary complex, the Sierra Nevada magmatic arc, the interjacent Coast Range Ophiolite, and Great Valley group forearc basin (Figure 1). The Franciscan Complex consists of mixed mafic, metasedimentary, and mantle material in various mélanges, coherent blocks, and thrust sheets along the length of coastal California. Though a full discussion is beyond the scope of this study, we briefly review components of the subduction system relevant to this paper. We also discuss a unit from the Western Sierra metamorphic belt (Red Ant schist) that was previously suggested to relate to the Franciscan subduction system and is dated in this study.

Franciscan Mélanges
Mélanges with shale, serpentinite, or mixed matrices enclosing lithologically variable blocks are an important component of the Franciscan Complex. These mélanges occur between coherent nonmélange units of different metamorphic grade, along imbrications of such units, and along the boundary with the Coast Range ophiolite (see Wakabayashi, 2011, for a review). Recent observations suggest that the mélanges accreted together with the nonmélange units and should be considered part of the nappes (Wakabayashi, 2015(Wakabayashi, , 2017b. Two endmember mélange types occur in the Franciscan. The first includes block-in-matrix mélanges with all blocks having the same metamorphic grade as the matrix and lithologies of either or both flanking units. These are generally agreed on to be of tectonic origin, resulting from progressive deformation. The second class of block-in-matrix mélanges contain "exotic" blocks that differ in lithology and metamorphic grade from the flanking units and matrix, respectively. Their formation is disputed, and they are the focus of this study.

Franciscan High-Grade Metabasites
The largest coherent amphibole-eclogite grade units are kilometers in size (e.g., Willow Spring slab and Antelope Creek slab; Hermes, 1973;Wakabayashi & Dumitru, 2007), whereas the smallest investigated garnet-amphibolite block in this study (PG14) is approximately 40 cm in diameter. The large coherent high-grade units are differentiated from blocks by scale and by lacking a shale or serpentinite matrix surrounding them (Wakabayashi, 2015). No intermediate, felsic, or sedimentary bulk compositions are described within the larger coherent units, suggesting that their protoliths were never exposed to input of continental detritus. Absence of continental input is further supported by the lack of LREE-rich compositions in the high-grade units (Saha et al., 2005;Wakabayashi et al., 2010). Geochemical analyses of eclogites and garnet-amphibolite showed significant enrichments in immobile Th, Ba, and Pb relative to N-MORB and were interpreted to point to a nascent arc (suprasubduction) origin of the blocks (Saha et al., 2005;Wakabayashi et al., 2010). It should however be noted that based on differing assessment of chemical mobility during subduction zone metamorphism, the reliability of such interpretation is debated (Ghatak et al., 2012).
The P-T evolution of the high-grade blocks has been studied since the recognition of eclogite by Switzer (1945), and there has been general agreement on the observation of three parageneses of (i) taramitic/katophoritic/barroisitic amphibole + garnet, (ii) omphacite + garnet and (iii) glaucophane ± phengite ± lawsonite ± garnet (e.g., Mulcahy et al., 2018;Tsujimori et al., 2006; 10.1029/2020TC006305 Tectonics Wakabayashi, 1990). However, there is a prevailing disagreement on the temporal order of the hornblende and omphacite bearing parageneses. For example, Krogh et al. (1994) and Wakabayashi (1990) interpreted hornblende inclusions in omphacite and garnet to indicate a counter-clockwise P-T path with an initial amphibolite-facies peak T, followed by a later, eclogite-facies peak P overprint at similar or lower temperature. In contrast, Mulcahy et al. (2018) observed inclusion of omphacite in hornblende and interpreted a clockwise P-T path, with initial high-pressure eclogite-facies metamorphism followed by a higher-temperature amphibolite overprint. Moreover, Tsujimori et al. (2006) worked on the same blocks as Wakabayashi (1990) and concluded a hairpin PT-path toward the amphibole-eclogite facies in which the prograde and retrograde path approximately overlie each other. Despite these disagreements, all above cited studies agree on a variably strong, late blueschist-facies overprint of the eclogite/amphibolite blocks.
Independently of their now-enclosing matrix, many of the eclogite/amphibolite blocks have peripheral or internal reaction zones composed of hydrous minerals, primarily actinolite, epidote, phengite, talc, and chlorite (Wakabayashi, 2019). These rinds have been classically interpreted to have formed at the outer contact of the block with serpentinite by mechanical mixing and metasomatic alteration hybridizing the block and matrix (Coleman & Lanphere, 1971;Penniston-Dorland et al., 2014). A revised interpretation is proposed by Wakabayashi (2017bWakabayashi ( , 2019 who points out that the internal reaction zones are very common and likely formed along imbricates of ultramafic schists. Also, peripheral rinds are typically found only along parts of the periphery of a block. Based on these observations they suggest that many or all of the peripheral rinds formed as internal selvages and came into their peripheral position only upon breakup of the blocks. The ubiquity of these features indicates that many, if not all, of the high-grade blocks were at some point either encased in an ultramafic (serpentinite) matrix or imbricated with ultramafic layers, even if they are now exposed within a shale matrix.

Coast Range Ophiolite
Twenty-two remnants of the CRO occur scattered over~600 km along the California Coast Ranges (Bailey et al., 1970;Hopson et al., 2008). They encompass variably complete ophiolitic successions from upper mantle peridotite to volcanic and sedimentary cover, exposed in ≤3 km thick sections. Where contacts are exposed, these ophiolitic remnants have been tectonically juxtaposed with the structurally lower Franciscan Complex, whereas a sedimentary contact is preserved with the structurally higher Great Valley group (Bailey et al., 1970;Hopson et al., 2008). Based on crustal-scale seismic and gravity data, Constenius et al. (2000) estimate the overthrust length of CRO on the Franciscan Complex to be >60 km without any intervening forearc mantle.
A multiphase formation process for the CRO is well established (Figure 2a; Hopson et al., 2008;Shervais et al., 2004). The main phase of CRO crystallization is constrained with TIMS and LA-ICP-MS U-Pb dates of zircons ranging from 172.0 ± 4.0 to 161.13 ± 0.11 Ma; these zircons are separated from texturally late plagiogranites within gabbroic crust barren of zircon in eight of the remnants (Hopson et al., 2008). This magmatic phase is ascribed to a supra subduction origin based on geochemical data by Metcalf and Shervais (2008) and Shervais et al. (2005), while Hopson et al. (2008) ascribed it to a "typical" spreading center and related off-axis magmatism based on paleomagnetic and biostratigraphic arguments. A full consideration of these two interpretations is beyond the scope of this study, but the recent recognition of Franciscan subduction initiation by~176 Ma (Mulcahy et al., 2018)-that is, prior to the formation of the CRO-favors a supra-subduction origin for the CRO.
The Middle Jurassic ophiolite is disconformably overlain by an Oxfordian to upper Tithonian (~163.5-145.0 Ma) volcanopelagic succession of radiolarian ooze and waterlain tuff, as well as locally coarser andesitic to rhyolitic volcaniclastic marine sediments (Figure 2a). A later magmatic phase resulted in the emplacement of basaltic to silicic sheets crosscutting the CRO; these are as young as 150 ± 2 Ma ( 40 Ar/ 39 Ar hornblende; Evarts et al., 1992). Young basaltic magmatism has a distinct MORB-like signature and was interpreted to indicate subduction of a spreading ridge beneath the CRO (Shervais et al., 2004). Due to a lack of documentation of many geochronologic ages that are published only in conference abstracts, the correlation between ages and geochemical phase classifications remains mostly tentative (Shervais et al., 2004) and is partly driven by different interpretations regarding the tectonic setting of ophiolite formation (e.g., Hopson et al., 2008;Shervais et al., 2005). Additionally, several 40 Ar/ 39 Ar dates (e.g., Evarts et al., 1992) 10.1029/2020TC006305 Tectonics RUTTE ET AL. cannot be recalculated with a recent 40 Ar/ 39 Ar age standard recalibration because the data lack sufficient detail, although the new calibration will likely make these dates older by only a few Ma.

Great Valley Group
The Great Valley group was deposited atop the CRO. Where CRO remnants are absent, its contact with the structurally deeper Franciscan Complex is a fault contact (Constenius et al., 2000;Ernst, 1970). The Great Valley group consists of mudstone with turbiditic silt-and sandstone and submarine fanglomerate at the base (basin-plain deposits) that grade upward into a classical forearc basin succession (Ingersoll et al., 1978). The basal age of sedimentation is a matter of recent debate: Biostratigraphic constraints based on Buchia zones indicate sedimentation in the late Kimmeridgian (~155 Ma) in northern California to late Tithonian (~145 Ma) in central California (Figure 2a; Hopson et al., 2008, and references therein). However, detrital zircons with Cretaceous U-Pb ages in many of the near-base sandstones previously assigned to the Jurassic in Northern California contradict this assignment (Figure 2a; Orme & Surpless, 2019;Surpless et al., 2006). Calculated maximum deposition ages range 153-135 Ma and imply a temporal gap between CRO formation and the beginning of Great Valley group deposition of up to 15-23 Ma in four studied sites (Orme & Surpless, 2019).

Sierra Nevada Batholith
The Sierra Nevada batholiths ( Figure 1

Red Ant Schist
A major tectonic suture zone east of the Franciscan is traced by the Feather River ultramafic belt, a N-S trending string of blueschist-facies ultramafic, metabasaltic, and metasedimentary units (Figure 1; various classification scheme of Sierra Nevada foothills terranes are reviewed by Taylor et al., 2010). Among the best investigated higher-grade units along this suture is the Red Ant schist, a coherent unit of blueschist-facies metasedimentary and metabasaltic rocks cropping out along the Melones shear zone together with serpentinite ( Figure 1; Schweickert et al., 1980). Phengite from the Red Ant schist yielded K/Ar dates of 157-190 Ma and were interpreted as a minimum age of blueschist-facies metamorphism (Schweickert et al., 1980). These ages suggested that the suture zone may be temporally correlated with early Franciscan subduction (e.g., Ernst, 2011). The Red Ant schist is also lithologically similar to the Stuart Fork blueschists in the Klamath Mountains (Hacker & Goodge, 1990), from which phengites were dated at~214-223 Ma (K/Ar; Hotz et al., 1977). Hacker and Goodge (1990) and Hacker (1993) pointed out that the K/Ar ages from the Red Ant schist (in contrast to the Stuart Fork blueschists) may be influenced by a late, greenschist-facies overprint.

Other Regional Data
Several events and changes in the Mid-to Late Jurassic evolution of the north Pacific region are approximately coeval with the investigated processes in the Franciscan at 180-150 Ma. In the western Sierra Nevada, metamorphic belt forearc sedimentation occurred until 174 Ma, followed by deformation and metamorphism until 165 Ma (Hacker, 1993). North of the Franciscan Complex, in the Klamath Mountains, the crust of the Josephine ophiolite formed by~160 Ma (U-Pb zircon in plagiogranite) and experienced a second phase of magmatism at 154-149 Ma (U-Pb zircon in cross-cutting silicic intrusions; Saleeby et al., 1982). The Sixes River mélange of the Franciscan Complex of Oregon hosts block-in-matrix mélanges with blueschist, amphibolite, and eclogite blocks (Blake et al., 1985) that appear similar to the here studied units and may be their northward prolongation; few available K/Ar data are consistent with K/Ar data from the Franciscan mélanges of California (Coleman & Lanphere, 1971). The Easton metamorphic suite in the northwest Cascades documents metamorphic sole formation in an initiating subduction zone between >167 and 164 (Cordova et al., 2019). In the Canadian Cordillera, marine basins of the Slide mountain terrane were deformed during the accretion of the Quesnellia island arc to the continental margin at 187-174 Ma (Murphy et al., 1995). The Talkeetna arc in southern Alaska formed at 174-158 Ma, while in northern Alaska the beginning closure of the oceanic Angayucham basin is dated by obduction of the Brooks range 10.1029/2020TC006305 Tectonics ophiolite at 169-163 Ma (Wirth et al., 1993;40 Ar/ 39 Ar hornblende, not recalculated because age standard is unknown).

Sampling
By design of the study, most of the investigated blocks have been previously described and analyzed with various methods. Except for sample BR-5 that was made available by Sean Mulcahy, all blocks were resampled without knowledge of the exact previous sampling location(s) on the block. Thus, given sample names designate identical blocks, not the same individual samples.

40 Ar/ 39 Ar Analysis of Phengite and Amphibole
Samples were crushed, sieved, and ultrasonicated in deionized water. Phengite and amphibole (mostly hornblende sensu lato, cf. Table 1) were concentrated using magnetic separation and a dry shaking table. Pure mineral separates were handpicked from the largest possible sieve fraction.
The separates were loaded in wells in Al discs bracketed with Fish Canyon sanidine (FCs) as a geological age standard (28.294 ± 0.036 Ma; Renne et al., , 2011. Neutron irradiations were carried out in two batches in the CLICIT and CLOCIT facilities at the Oregon State University TRIGA reactor (Table 1). Argon was released from sample aliquots of 0.5-4 mg by step-wise heating using the defocused beam of a CO 2 laser. To reach most homogeneous heating of the grains possible with this technique, we ensured that grains were not stacked. The analyses were carried out using two MAP-215 mass spectrometers, each equipped with one analog electron multiplier at Berkeley Geochronology Center (Table 1). Several samples were analyzed in duplicate to optimize the heating schedule for better separation of isotopic reservoirs. We scanned for double-charged 82 Kr on m/z 41 to monitor possible interference from double-charged 80 Kr on m/z 40, produced from Br during irradiation (Rutte et al., 2017); no significant contents were measured. Correction factors for collateral Ar isotope production and 37 Ar Ca / 39 Ar K production ratio for CLICIT irradiation are taken from Renne et al. (2015); correction factors for CLOCIT are from Rutte et al. (2018), and we determined and used a 37 Ar Ca / 39 Ar K production ratio of 1.99 ± 0.08 for CLOCIT using Hb3Gr in irradiation 476.
We define plateau dates by at least three contiguous steps that overlap within 2σ uncertainty, yield >50% of the released 39 Ar K , and have homogeneous Ca/K ratios. Plateau dates are calculated as inverse variance weighted means.

40 Ar/ 39 Ar Age Interpretations
As in our results, previous 40 Ar/ 39 Ar studies employing stepwise-degassing analysis of Franciscan phengite and amphibole found partially disturbed age spectra; disturbances mostly stem from excess 40 Ar (radiogenic 40 Ar, separated from its parent 40 K) and the difficulty of preparing pure mineral separates from fine grained and sometimes intergrown phengite and amphiboles (Ross & Sharp, 1988;Wakabayashi & Dumitru, 2007). In most cases, reliable plateau or isochron ages can be calculated based on a subset of steps, but these deviate from the spectrum-integrated (also referred to as total fusion) ages. Early K/Ar and 40 Ar/ 39 Ar total fusion data from low-grade metasedimentary units of the Franciscan were comprehensively appraised by Dumitru et al. (2010). However, for lack of internal verifiability and testability of excess 40 Ar, we exclude all previous total fusion data from our discussion of the metamorphic blocks.

Laser-Ablation Split Stream (LASS) Zircon U-Pb Geochronology and Trace-Element Geochemistry
Samples were crushed, sieved, and washed with deionized water. Zircons were concentrated from about 2 kg of <150 μm sieve fraction using magnetic separation and heavy liquids. Though we attempted to isolate zircon from several more samples with ≤15 μm sized zircon visible in thin section, we were only able to retrieve zircons large enough for laser ablation from TIBB and BH-32. Zircon grains were handpicked in ethanol and mounted in epoxy, then polished to central sections and imaged using cathodoluminescence (CL) and backscattered electrons (BSE) detectors on an FEI Quanta 400f scanning electron microprobe at the University of California Santa Barbara ( Figure S1).  (Paton et al., 2011) for the Wavemetrics Igor Pro software was used to correct measured isotopic ratios and elemental intensities for baselines, time-dependent laser-induced inter-element fractionation, plasma-induced fractionation, and instrumental drift. Downhole fractionation was modeled using an exponential best fit.

Tectonics
The uncertainty of single measurements is dominated by counting statistics and signal stability. Each 207 Pb/ 206 Pb and 206 Pb/ 238 U measurement requires an additional 2% error attributable to variation in ablation or transport characteristics, mass-balance instabilities, or plasma loading effects; this was added in quadrature. Long-term analysis shows that this equipment and method are capable of measuring the baseline-and fractionation-corrected 206 Pb/ 238 U and 207 Pb/ 206 Pb of a primary reference material (e.g., 91,500 zircon) with a long-term precision of ≤1.5% (Kylander-Clark et al., 2013). When combined with uncertainties in the isotopic ratios of the primary RM and U decay constants, the date of a homogeneous unknown has an external uncertainty of~2%. All date uncertainties are reported at the 95% confidence interval, assuming a Gaussian distribution of measurement errors. Stated 2σ date uncertainties are internal; that is, they include propagated in-run and decay constant errors only. Low and imprecise Ti concentrations in zircon (coupled with uncertain silica activities) preclude robust temperature calculations for zircon (re)crystallization.

Comparability of Geochronologic Methods
We herein compare Lu-Hf, U-Pb, and 40 Ar/ 39 Ar data. The compiled Lu-Hf ages are calculated using a 176 Lu decay constant of 1.867 × 10 −11 a −1 calibrated directly with the U-Pb system (Scherer, 2001;Söderlund et al., 2004). We therefore consider only the given Lu-Hf isochron uncertainties with the original data. The external uncertainty of the U-Pb method using our methods is~2% (see above). New and compiled 40 Ar/ 39 Ar ages are calculated based on standard age and decay data intercalibrated with the U-Pb system using the optimization approach of Renne et al. ( , 2011. Systematic uncertainties based on inconsistencies between the three decay systems fall below the time scales interpreted in this study.

Results: Block Descriptions and Analytical Data
In the following we describe data from individual blocks with respect to their locale, petrography, and results of previous studies on the same blocks. We then report our new geochronologic results and their context, followed by a brief interpretation of the data for the block. Comparisons among individual blocks, and overarching interpretations, follow in section 5. New 40 Ar/ 39 Ar and U-Pb results are displayed in Figures 3 and 4 Ar compositions. Results from amphibole and phengite are in red and green tones, respectively. Uncertainties are plotted at the 1σ level, but calculated results are at the 2σ level for ease of comparability with other methods. Further analytical detail in Table 1 and supporting information. Abbreviations: IID, inverse isochron date; IntA, integrated age; Int., intercept; MSWD, mean square weighted deviation; cum., cumulative. Raw data used to construct these plots are in Table S1. 10.1029/2020TC006305 Tectonics respectively. Figure S2 provides a photographic documentation of the samples and Table S1 electron probe microanalysis data of amphibole and phengite.

Mount Hamilton: MH-3
MH-3 was a block at the west slope of Mount Hamilton that was destroyed by a road crew in 1990. The block was composed of an eclogite core with lenses of garnet-amphibolite, wrapped by a retrograde metamorphic Tectonics rind. Catlos and Sorensen (2003) dated individual phengite crystals from the rind using laser ablation 40 Ar/ 39 Ar analysis with a weighted mean dates of 142.9 ± 3.7 Ma (their block MH-90). Two aliquots of hornblende from the garnet-amphibolite in MH-3 yielded similar rising apparent age spectra and Ca/K ratios in the beginning of the step-heating experiment (Figure 3a). The second half of released 39 Ar K yielded plateau dates with a weighted mean date of 153.70 ± 0.81 Ma and stable Ca/K ratios of 10-13. Signals of 36 Ar are small, and the steps with homogeneous Ca/K have very little spread in inverse isochron space and thus define an imprecise date. The shape of age and Ca/K spectra and the comparably low Ca/K of the plateau date point to a successive homogenization of Ar from the hornblende and-likely-younger, low Ca/K phengite inclusions during the step-heating experiment, analogous to the observations of Ross and Sharp (1988). We thus interpret the weighted-mean date of 153.70 ± 0.81 Ma as a minimum age for crystallization or cooling of the hornblende. Phengite from the rind yielded a plateau date of 148.37 ± 0.67 Ma. The inverse isochron through the plateau steps yields a 40 Ar/ 36 Ar intercept of 290 ± 170 (i.e., atmospheric with no inherited Ar). We interpret the date as the age of crystallization or cooling of phengite.

Jenner: Sunset Boulders (Je1r and Je2)
Je1r and Je2 consist of multiple amphibolite fragments littered over a 100 × 60 m area. They were mapped atop the marine terrace south of the Russian River estuary by (Wakabayashi, 2015). Individual fragments are up to 15 × 10 × 10 m in size and are in some cases broken apart in place. Sample Je2 is an amphibolite sample from a central large fragment. It contains brown hornblende with a ferro-tschermakite to ferro-pargasite composition. Matrix hornblende (Je2) yields a plateau date of 159.85 ± 0.32 Ma with a Ca/K ratio of 30-34 (Figure 3b). An inverse isochron through all steps has a super-atmospheric 40 Ar/ 36 Ar intercept of 313.8 ± 5 and yields an inverse isochron date of 158.1 ± 1.6 Ma (Figure 3c). Because of the small component of excess Ar, we prefer the slightly younger inverse isochron date. Sample Je1r is a piece of chlorite-actinolite rind sampled from a fragment at the edge of the boulder field. Actinolitic amphibole from the rind yields a plateau date of 154.5 ± 2.3 with a Ca/K of 63-82 (Figure 2b). An inverse isochron through all steps has an atmospheric 40 Ar/ 36 Ar intercept. The dates are indistinguishable within uncertainty and permit three possible interpretations: (i) both dates are cooling ages, (ii) both dates are crystallization ages, or (iii) the matrix hornblendes date cooling during crystallization of the rind.

Jenner: Heart Rock (HR-2 and HR-3)
Heart Rock is a 3 × 5 × 4 m sized block of epidote + garnet-blueschist at Jenner Beach. Foliation parallel mineralogical layering is isoclinally folded and cut by veins of phengite and sodic amphibole. Remain of clinopyroxene were found in garnet, suggesting earlier eclogite-facies metamorphism (Carruthers & Rowe, 2017). Phengite from the matrix yields 3.4 Si pfu. It yields a subtly saddle shaped 40 Ar/ 39 Ar age spectrum with a plateau date of 148.9 ± 1.8 Ma (Figure 3d). An inverse isochron through steps 4-13 yields a 40 Ar/ 36 Ar intercept of 349 ± 57 (atmospheric within uncertainty) and an inverse isochron date of 147.8 ± 3.0 Ma (Figure 3e). Steps 1-3, however, bend toward a nonatmospheric 40 Ar/ 36 Ar value. Because of this hint to excess 40 Ar, we prefer the inverse isochron date over the weighted-mean age and interpret it to be the age of crystallization or cooling through phengite closure to Ar diffusion. Microprobe analysis of phengite from a crosscutting vein yields an indistinguishable composition from the host rock; two . Spot analyses, calculated isochrons, and chondrite-normalized rare-earth element (REE) plots are color-coded by texture (cores and rims). Two core spots from PG32/TIBB are clearly older and were excluded from the core isochron calculation in panel B. Chondrite values for REE normalization were sourced from McDonough and Sun (1995). Raw data used to construct these plots are in Table S2. Constructed using ISOPLOT (Ludwig, 2012).

10.1029/2020TC006305
Tectonics aliquots yielded plateau dates with atmospheric intercept of 283 ± 15 and 297 ± 16 and a resulting inverse variance weighted mean date of 141.49 ± 0.43 Ma, straightforwardly interpreted as the age of formation of the vein (Figure 3d). Together, the phengite host-rock and vein dates indicate a residence time of ≥6 Ma in the blueschist facies.

Jenner: Je3d
Je3d is from a 35 × 30 × 20 m sized block of garnet blueschist atop the marine terrace south of the Russian River estuary. The sample consists of-in thin section-mostly blue to purple amphibole with complex optical zonation patterns and few greenish amphibole grains, possibly indicating an earlier amphibole-eclogite facies metamorphism ( Figure S2). We were able to separate only few dark greenish (in separate) amphibole grains with low K-content and thus obtained comparably imprecise data. The sample yields a 40 Ar/ 39 Ar plateau date of 158 ± 14 Ma (Figure 3f). An inverse isochron through all steps yields an atmospheric 40 Ar/ 36 Ar intercept of 290.5 ± 8.9.

Tiburon Mélange: PG23/TIBB/TIBBvein
PG23 is a 3 × 3× > 2 m large garnet-and phengite-bearing amphibolite block from Ring Mountain that was first described by Wakabayashi, 1990; his sample TIBB). Phengite and hornblende define the main penetrative foliation and stretching lineation. Rare fold hinges preserve an older isoclinally folded foliation with the axial planes parallel to the main foliation. Omphacite-rich eclogite occurs in bands and patches throughout the block. Wakabayashi (1990) described eclogite "veins" crosscutting the foliation defined by the hornblende and concluded that the higher-pressure eclogite assemblage overprints the older hornblende paragenesis. A reevaluation of the data led Tsujimori et al. (2006; their sample A) to peak P-T conditions of the eclogite assemblage (Grt-Cpx-Phe) of 550-620°C at 1.9-2.5 GPa and interpretation of a hairpin P-T path in which retrograde and prograde metamorphism followed a similar trajectory. Cooper et al. (2011) calculated peak T of 510-560°C for the amphibolite paragenesis (Grt-Hbl). Varying modes of chlorite and glaucophane indicate variable retrograde overprint of both amphibolite and eclogite assemblages. During sampling we discovered a previously undescribed vein of coarse phengite (>10 mm) and bladed gray-blue amphibole aggregates (>20 mm long) cross cutting the foliation (sample TIBBvein; Figure S2). Saha et al. (2005) analyzed the bulk rock chemistry of PG23 (their sample TIBB), found little indication of trace-element mobility since crystallization, and interpreted the sample protolith as a nascent arc basalt. Anczkiewicz et al. (2004) determined a Lu-Hf garnet-WR date of 153.4 ± 0.8 Ma from a two-point isochron (their sample PG23).
Zircons from a garnet-amphibolite layer are up to 150 μm in length, but mostly <100 μm long and have a compact habit with aspect ratios around 1:2. Cathodoluminescence reveals patchy zoning, with bright rims overgrowing darker and partially metamict cores along lobate boundaries ( Figure S1). However, zircon cores and rims share a similar mineral chemistry; core and rim Th/U ratios range from 0.2-0.8 and 0.4-0.5, respectively, and cores and rims have similar REE abundances without a statistically significant Eu anomaly (Figure 4a, inset; Table S2). Core analyses range from concordant to a few percent discordant, and an isochron regression through 22 spot analyses yields a lower-intercept 206 Pb/ 238 U date of 165.9 ± 2.0 Ma (MSWD = 1.8; Figure 4a; cf. Wendt & Carl, 1991). Two core spots yielded much older dates of 180-185 Ma, albeit with significant uncertainties (±10 Ma) and with trace-element compositions indistinguishable from other core analyses. Only three inclusion-free rim spots were obtained; using the same upper intercept defined by the core isochron ( 207 Pb/ 206 Pb ≈ 0.5), we obtain a lower-intercept 206 Pb/ 238 U rim date of 157.1 ± 3.2 Ma (MSWD = 1.8; Figure 4a). A rim isochron calculated using the Stacey and Kramers (1975) 207 Pb/ 206 Pb for the calculated date (~0.85) yields the same date within uncertainty. We also attempted titanite and rutile U-Pb dating from this sample, but uranium concentrations for both minerals were below the LA-ICP-MS detection limit.
Two aliquots of matrix hornblende yield well defined 40 Ar/ 39 Ar plateau dates with a weighted mean date of 159.19 ± 0.81 Ma (Figure 3g). Inverse isochrons through all steps yield 40 Ar/ 36 Ar intercepts of 300 ± 23 and 286.6 ± 6.6. Phengite from the matrix yields a plateau date of 156.80 ± 0.54 Ma with a slightly supra-atmospheric 40 Ar/ 36 Ar intercept of 333 ± 24. We thus prefer the slightly younger inverse isochron date of 156.4 ± 1.1 Ma. Phengite from the coarse vein yields a similar apparent age spectrum but with a downward step in the middle. In inverse isochron space the first three steps describe a bend toward atmospheric 10.1029/2020TC006305 Tectonics 40 Ar/ 36 Ar, while the following 14 steps form an inverse isochron with a 40 Ar/ 36 Ar intercept of 690 ± 180 and a date of 150.4 ± 2.9 Ma (Figure 3h).
These dates and mineral chemistry data implicate metamorphic (re)crystallization of zircon cores without coexisting plagioclase from at least 166 Ma until 158 Ma; if the older zircon core dates (~180 Ma) are significant, they may record protolith basalt crystallization, or alternatively suggest a much earlier initiation of eclogite-facies conditions. Because the eclogite-facies assemblages define the main, penetrative foliation, the observed isoclinal folding must have occurred during this prograde metamorphic interval. With the calcic hornblende having grown along the prograde path (Wakabayashi, 1990), and a peak T of 550-620°C exceeding the hornblende Ar closure temperature (Tsujimori et al., 2006), the 40 Ar/ 39 Ar dates of hornblende dates are straightforwardly interpreted as cooling close to peak T conditions at~159 Ma, synchronous with zircon resorption and rim growth. The 40 Ar/ 39 Ar plateau date of matrix phengite is the age of cooling below phengite Ar closure at~157 Ma. The 40 Ar/ 39 Ar inverse isochron date of vein phengite is the age of vein crystallization at~150 Ma from an aqueous fluid carrying inherited 40 Ar. Considering these constraints, the TIBB block spent at least~7 Ma (and potentially up to~20 Ma) under high-grade prograde conditions and remained at blueschist facies or higher conditions for another~8 Ma in the subduction zone.

Tiburon Mélange: RM-8
RM-8 is a 5 × 12× > 2 m large epidote-garnet-amphibolite block on Ring mountain that was first detailed by Mulcahy et al. (2014). Phengite and hornblende define a strong foliation and stretching lineation. No variation in lithology or internal structure is visible given a comparably fresh and lichen free level of exposure. Late carbonate-filled fractures are oriented perpendicular to the long axis of the block. Mulcahy et al. (2014) determined a peak T of 636 ± 46°C by modeling Ca and Mg zonation in garnet and obtained a Lu-Hf garnet-WR date of 153.4 ± 0.8 Ma from a 6-point isochron.
Matrix hornblende is zoned with magnesio-hornblende, magnesio-taramite, barroisite, and actinolite compositions in grain cores, overgrown by ferroan glaucophane rims. Microprobe analysis yielded Ca/K ratios ranging 14 to 17 in hornblende and above 160 in sodic amphiboles. Matrix hornblende yielded a 40 Ar/ 39 Ar plateau date of 156.61 ± 0.86 Ma at a Ca/K of 15-17 ( Figure 3i) with an atmospheric 40 Ar/ 36 Ar of 240 ± 100 of the inverse isochron. We interpret this to be the age of retrograde hornblende crystallization or cooling through Ar closure in hornblende. In either case the date is most likely close to determined peak T conditions. Phengite yielded 3.4 Si pfu and a 40 Ar/ 39 Ar plateau date of 155.35 ± 0.50 Ma. In inverse isochron space, steps 3-15 form an isochron with an atmospheric 40 Ar/ 36 Ar intercept of 319 ± 44. We interpret this date to be the age of cooling or crystallization of phengite with minor incorporation of excess 40 Ar in the blueschist facies. We attempted titanite and rutile U-Pb dating, but U concentrations were too low for LA-ICP-MS analysis.

Tiburon Mélange: DRM-1/EA
DRM-1 is a 20 × 25× > 6 m large block on Ring mountain that was first described by Mulcahy et al. (2018; their sample EA). Its dominant lithology is a phengite and garnet-bearing amphibolite, with a penetrative foliation and stretching lineation defined by hornblende and phengite. Up to 40 mm thick and 5 m long layers of garnetite are found subparallel to the foliation; Mulcahy et al. (2018) found a symmetric chemical zonation inside the garnetite layers and suggested syntaxial growth in a successively opening fracture.
Zircons included in garnet and from the matrix yielded lower intercept inverse concordia dates of 176.5 ± 3.8 Ma and 160.3 ± 3.2 Ma, respectively (Mulcahy et al., 2018). These were interpreted as ages of early prograde-probably eclogite-facies-metamorphism at~176 Ma, followed by hornblende paragenesis overprinting at~160 Ma. Mulcahy et al. (2018) further analyzed garnet compositions and found variable zoning patterns suggesting a complex growth and resorption history. Lu-Hf geochronology of multiple garnet aliquots reflects this observation by scattering in isochron space; isochrons enveloping all isotopic data were interpreted to date episodes of garnet growth from 166 ± 1 Ma to 155 ± 1 Ma (Mulcahy et al., 2018).
Two aliquots of matrix hornblende yielded plateau dates with a weighted mean of 160.14 ± 0.78 Ma (Figure 3j). Inverse isochrons yield atmospheric intercepts of 283 ± 15 and 297 ± 16. We interpret this to be the age of hornblende crystallization or cooling through Ar closure in hornblende. Phengite yields a 10.1029/2020TC006305 Tectonics saddle shaped 40 Ar/ 39 Ar apparent age spectrum. In inverse isochron space, the first two steps bend toward an atmospheric 40 Ar/ 36 Ar intercept, while successive steps define an isochron with a date of 146.8 ± 2.3 Ma and a 40 Ar/ 36 Ar intercept of 357 ± 74 for steps 3-13 (Figure 3k). We interpret this date to be the age of cooling or crystallization of phengite under incorporation of minor amounts of excess 40 Ar.

Tiburon Mélange: BH-32
BH-32 is a 3 × 2 × 2 m sized block of garnet amphibolite in the Berkeley Hills. The block has weak foliation and stretching lineation and was described in detail by Mulcahy et al. (2018). It contains omphacite crystals included in garnet and as relicts within a matrix of sodic amphibole and hornblende. Garnets display complex zoning patterns indicating episodic growth and resorption. From these observations, Mulcahy et al. (2018) interpreted an initial eclogite-facies metamorphic event, followed by amphibolite paragenesis and blueschist-facies overprints. Lu-Hf isotopic compositions of multiple aliquots scatter in isochron space; Mulcahy et al. (2018) calculated enveloping isochrons and interpreted them to date episodic garnet growth between 172 ± 1 Ma and 156 ± 1 Ma.
Ten zircons from BH-32 have long axes of up to 120 μm with an elongated habit (aspect ratios~1:3). Cathodoluminescence reveals up to 15 μm wide rims overgrowing cores along rounded boundaries in three grains ( Figure S1); one core is metamict. The cores and rims have similar CL textures and have generally similar compositions with respect to Th/U, REE, and Eu/Eu*. However, the rim analyses typically have steeper HREE slopes than the core spots, which typically exhibit flat chondrite-normalized HREE patterns (Figure 4b, inset). Eight core analyses yield an imprecise unconstrained isochron (lower intercept 206 Pb/ 238 U date = 157.6 ± 6.6 Ma) that does not form a single population (MSWD = 3.6), with an upper intercept ( 207 Pb/ 206 Pb ≈ 0.38) that deviates significantly from the Stacey and Kramers (1975) value for the same date ( 207 Pb/ 206 Pb = 0.85). Using the same upper intercept as determined by the core isochron regression, four rim analyses yield a statistically significant isochron date of 157.0 ± 2.6 Ma (MSWD = 2.3) (Figure 4b). The synchronicity between core and rim zircon dates, coupled with changes in REE slopeand their similarity to the youngest garnet Lu-Hf dates from the same sample (Mulcahy et al., 2018)suggests that the zircon growth and recrystallization textures relate to the cessation of garnet growth and the initiation of garnet breakdown by~157 Ma.
In thin section, hornblende tends to have darker cores and lighter rims or patchy zoning and overgrowth of sodic amphibole. EPMA-determined amphibole compositions include barroisite, edenite, magnesiohornblende, and magnesio-katophorite with Ca/K ratios ranging 15-24. Our hornblende separate yields a 40 Ar/ 39 Ar plateau date of 160.09 ± 0.92 Ma at a relatively constant Ca/K of 18-20 (Figure 3l). An inverse isochron through all steps yields an atmospheric intercept. Phengite yields 3.4 Si pfu and a 40 Ar/ 39 Ar plateau date of 146.26 ± 0.52 Ma. An inverse isochron through all steps yields an atmospheric 40 Ar/ 36 Ar intercept.
The garnet and zircon data suggest that high-grade metamorphism started prior to~170 Ma with episodes of garnet and zircon (re)crystallization until~157 Ma. The hornblende date at~159 Ma could be interpreted as a cooling age or crystallization age, in either case during retrogression, but close to peak T. The phengite Ar/Ar date can be either interpreted as a cooling age or crystallization age during blueschist-facies retrogression. Prograde metamorphism thus lasted >11 Ma and later blueschist facies metamorphism continued for >15 Ma.

Tiburon Mélange: PG14
PG14 is a 40 × 40 × 30 cm sized block of garnet amphibolite in the Berkeley Hills. It is foliated and has a stretching lineation defined by hornblende; ≤1 mm wide calcite veins cut the block. Anczkiewicz et al. (2004) determined metamorphic peak-T of 580-610°C from garnet-hornblende thermometry, and a garnet crystallization age of 162.5 ± 0.5 Ma from a 4-point garnet-WR Lu-Hf isochron.
Amphiboles have a patchy zoning with magnesio-hornblende and barroisite compositions and overgrowths of actinolite. Microprobe analyses yield Ca/K ratios ranging 18-45. Two aliquots yielded similar but complex 40 Ar/ 39 Ar apparent age spectra with an initial rise from~45 Ma to~154 Ma, decreasing again to~130 Ma (Figure 3m). The Ca/K ratios determined from 37 Ar Ca / 39 Ar K rise in parallel to the rising apparent ages tõ 22, form a plateau and only the last steps show another increase. In inverse isochron space, individual steps exhibit excess scatter suggesting mixture of one nonradiogenic and two radiogenic sources (Figure 3n). The 10.1029/2020TC006305 Tectonics complex apparent age and Ca/K spectra likely relate to differential degassing and phase changes of the magnesio-hornblende and barroisite during the step-heating experiment (cf. Wartho et al., 1991). We consider the plateau date of aliquot 1 at 154.6 ± 1.4 Ma to be a minimum age for the cooling of hornblende through Ar-closure.

Tiburon Mélange: BR-5 (Big Rock)
Big Rock is a 10 × 12 × 12 m sized garnet-epidote blueschist block near Nicasio, CA. It is probably part of the Tiburon mélange. Mulcahy et al. (2014) determined peak T conditions of 542 ± 41°C from garnet cation diffusion zonation assuming a pressure of 10 kbar and further determined a 4-point Lu-Hf garnet-WR isochron date of 156.2 ± 1.0 Ma. Sample BR-5 contains remnants of hornblende in the matrix (Mulcahy et al., 2014), but we found these grains to be inseparable from phengite and did not attempt to date them. We found sodic amphibole, but no hornblende inclusions in garnet.
Phengite from BR-5 contains 3.4 Si pfu. It yields a subtly saddle shaped apparent age spectrum with a plateau date of 150.8 ± 1.1 Ma (Figure 3o). An inverse isochron through steps 3-15 yields a super-atmospheric 40 Ar/ 36 Ar intercept of 495 ± 88, indicating the presence of excess Ar, and a date of 148.5 ± 1.4 Ma (Figure 3p). Steps 1 and 2 bent toward an atmospheric intercept, likely related to adhered atmospheric Ar. We prefer the inverse isochron date over the plateau date and interpret it as the age of crystallization or cooling during retrograde blueschist facies metamorphism. The presence of sodic amphibole and the lack of hornblende inclusions in the previously dated garnet suggests that amphibolite-facies metamorphism occurred prior to~156 Ma ( Figure S2).

Red Ant Schist: YU4
To explore the significance of earlier K/Ar dates-which suggested a temporal correlation between early Franciscan subduction and suturing in the Sierra foothills-we performed 40 Ar/ 39 Ar analysis of phengite from one of the outcrops investigated by Schweickert et al. (1980) at the Yuba River near Downieville, outside our main study area. Phengite from sample YU4 yielded a complex apparent age spectrum rising from 140 Ma to mostly 205-220 Ma with Ca/K _x0003C; 0.007 (Figure 3q). The last steps yield apparent ages rising up to 256 Ma with a Ca/K rising to 0.03. The integrated age, approximately equivalent to a conventional K/Ar date for our sample, is 208.1 ± 1.1 Ma. Our new 40 Ar/ 39 Ar data highlight the difficulty of interpreting legacy K/Ar dates; it shows that phengite in the Red Ant schist records a complex thermal history and suggests that blueschist-facies metamorphism in the western Sierra foothils is older than~205 Ma and thus unrelated to early Franciscan subduction.

Timescales of Metamorphism of High-Grade Blocks
The application of multiple geo-and thermochronometers to individual blocks allows us to discriminate the duration of metamorphism that each block experienced, from (i) prograde-to-peak crystallization of garnet and zircon, to (ii) retrograde hornblende crystallization or cooling, to (iii) retrograde blueschist-facies phengite crystallization or cooling. These data allow comparison among blocks from different parts of the Franciscan Complex, including an understanding of whether the blocks were subducted and exhumed as part of a coherent package, and how the duration and absolute timing of retrograde metamorphism varied locally and regionally. As a first observation, red and blue boxes in Figure 2b highlight the absence of coeval prograde (grt and zircon dates) and retrograde (mostly hornblende and phengite dates) eclogite facies metamorphism along regions of the Franciscan subduction zone; this discrimination is broadly synchronous along the entire orogen. In the following we discuss the overall interpretation of the 40 Ar/ 39 Ar dates of hornblende and phengite, their regional variability, and then compare the results of multiple methods applied to individual blocks.
While phase relations strongly suggest that the PG23 40 Ar/ 39 Ar hornblende date is a cooling age, in all other samples the hornblende date may be the age of cooling or crystallization; nonetheless, a succinct interpretation is possible. Calculated peak T conditions of the amphibolites are in the range of 500-600°C. Growth of hornblende sensu lato in metabasite during subduction metamorphism requires temperatures above~470°C (Ernst, 1979;Hernández-Uribe & Palin, 2019). For retrograde hornblende growth, kinetics likely limit (re) crystallization to even higher T. A simple diffusion model of hornblende for representative width of 200-

10.1029/2020TC006305
Tectonics 500 μm (infinite cylinder) and cooling rates of 10-100°C/Ma suggests closure temperature of Ar in hornblende of 507-589°C (Brandon et al., 1998;Harrison, 1981). With peak T, minimum (re)crystallization temperature and closure temperature converging at 500-600°C, we propose that-be they cooling or crystallization ages-the 40 Ar/ 39 Ar hornblende dates are a minimum-but close-estimate of the age of peak T metamorphism.
The durations between 40 Ar/ 39 Ar hornblende and matrix phengite dates range 1.2-14 Ma. The shortest difference in RM-8 with 1.2 ± 1.0 is difficult to interpret because both dates may date cooling or crystallization. In contrast, the hornblende and matrix phengite dates of sample PG23 are straightforwardly interpreted as cooling ages (see above). Matrix phengite is 2.8 ± 1.4 Ma younger than the 40 Ar/ 39 Ar hornblende date. Assuming closure temperatures of 550°C for hornblende at fast cooling rates and~360°C for phengite (Wijbrans et al., 1993;Wijbrans & McDougall, 1986) points to cooling rates of 45-136°C/Ma for a few Ma time interval. In the other samples, phengite dates may be the age of crystallization, in which case the dates can be linked to temperatures ≤360°C (less than the closure temperature of Ar in phengite). Durations of 13.3 ± 2.4 and 13.8 ± 1.1 Ma between 40 Ar/ 39 Ar hornblende and 40 Ar/ 39 Ar matrix phengite of DRM-1 and BH-32 indicate slower average cooling rates of~15°C/Ma. Crystallization of a second generation of phengite in veins and rinds in PG23, MH-3, and HR2&3 indicates that blueschist-facies metamorphism lasted at least another 6.0 ± 1.1, 5.3 ± 4.1, and 6.6 ± 3.0 Ma, respectively. In general, the scatter of 40 Ar/ 39 Ar phengite dates is larger than that of 40 Ar/ 39 Ar hornblende dates (Figures 2b and 5a).
Our new and compiled 40 Ar/ 39 Ar hornblende dates point to a distinct phase of block cooling from 168 to 157 Ma (n = 12; Figures 2b, 5a, and 5b, two disturbed dates excluded). When considering the dates in respect to their paleolatitudinal position a north-to-south younging trend is apparent. The oldest 168-166 Ma (n = 3) dates stem from the Tehama-Colusa mélange in northern California. The Antelope Creek slab from central California east of the San Andreas fault system-with an intermediate geographic position prior to Cenozoic strike-slip motion (reconstruction based on Wakabayashi, 1999)has a date of 163.3 ± 2.1 Ma (Ross & Sharp, 1988). The samples from Ward Creek and Jenner were at a similar latitude, whereas those from Ring Mountain and the Berkeley Hills were further south during Franciscan subduction, and range from 160.14 ± 0.78-156.61 ± 0.86 Ma (n = 8; Figure 5). This trend may be tested in the future by geochronologic analysis of high-grade blocks from the Sixes River mélange in Oregon that likely represents the northward continuation of the mélanges studied in California (Coleman & Lanphere, 1971) and would extend the study area by 350 km (Figure 1). In contrast to the tight clustering of the 40 Ar/ 39 Ar hornblende dates, 40 Ar/ 39 Ar phengite dates are more variable, ranging between 156.4 ± 1.1 and 134.6 ± 0.4 (n = 24; Figure 5) with most samples at 155-140 Ma. Considering only the blocks and coherent units that experienced amphibole-eclogite facies metamorphism (i.e., excluding those samples with peak P-T conditions in the blueschist facies), the range remains at 156.4 ± 1.1-143.4 ± 3.7 Ma (n = 12). All phengite data stem from Central California and do not allow comparison with the N-S trend observed in the hornblende data. The varying range of 40 Ar/ 39 Ar hornblende and phengite dates indicate relatively uniform initial exhumation of the high-grade Ar/ 39 Ar hornblende dates which approximate the age of peak T metamorphism.
To further compare durations of metamorphism based on multiple geochronometer we use the 40 Ar/ 39 Ar matrix hornblende dates to approximate the age of peak T conditions. Figure 5c shows the difference between the matrix hornblende 40 Ar/ 39 Ar date with the oldest available prograde metamorphic age (Lu-Hf garnet or U-Pb zircon), matrix phengite 40 Ar/ 39 Ar date and (if present) any second phengite generation in a vein or rind. This estimate returns durations of prograde metamorphism of 5-18 Ma. However, only a rough estimate of the metamorphic conditions dated by the garnet and zircon is possible. While twõ 180 Ma zircon dates in samples PG23 (Figure 4a) may reflect protolith (igneous) processes, most of the zircon U-Pb dates in this study overlap with Lu-Hf or Ar/Ar dates from the same samples (Figure 2), suggesting that zircon ages are intimately related to metamorphic (re)crystallization. There is no evidence that these zircons grew from in situ melts, or that these blocks ever experienced suprasolidus conditions. Our interpretation is supported by trace-element data tied to each U-Pb age. For example, the absence of a negative europium anomaly (Eu/Eu* < 1) in zircons is commonly interpreted to indicate the crystallization in a plagioclase-free, eclogitic assemblage (e.g., Rubatto & Hermann, 2003). However, it should be noted that plagioclase often is only an accessory phase in the investigated blocks and in Franciscan amphibolites in general (e.g., Wakabayashi, 1990). The positively sloping chondrite-normalized HREE patterns for all investigated zircons ( Figure 4) suggest that zircon crystallized prior to garnet crystallization or after garnet breakdown, that the amount of garnet in equilibrium with zircon was insufficient to deplete the bulk rock in HREE, or that zircon REE uptake was governed by local rather than bulk-rock factors.
Garnet occurs in a range of Franciscan blueschists with peak-T conditions as low as 400°C, such that Lu-Hf dates may record a range of prograde-to-peak metamorphic conditions. Additionally, available Lu-Hf garnet dates have been accompanied by a characterization of major-element zonation of the garnets (Anczkiewicz et al., 2004;Cooper et al., 2011;Mulcahy et al., 2014Mulcahy et al., , 2018 but did not include documentation of Lu distribution within the garnets. Further, Moore et al. (2013) showed that REE distributions in undated Franciscan garnets are exceedingly complex, far more so than represented by the major element data. Therefore, the limited information prevents robustly determining if the Lu-Hf dates represent mostly early, late, or mixed growth intervals. Though most Franciscan Lu-Hf data define single isochrons, the isotopic data of samples BH-32 and DRM-1 span an age range and their complex elemental zoning point to a complex growth and resorption history (Mulcahy et al., 2018).
All of the above-discussed complexities discourage a detailed direct comparison of the apparently variable duration of prograde metamorphism between blocks. However, the apparent long prograde-to-peak metamorphic durations of >15 Ma for blocks that reached~600°C (PG23, DRM-1), that is,~15 Ma for~300°C of heating at 20°C/Ma, point to a relatively slow sinking rate of~18 km/Ma, assuming an intermediate geothermal gradient of 11°C/km (discussed below in more detail). Even considering that 40 Ar/ 39 Ar hornblende ages may partially postdate peak-metamorphism by few Ma, or considering that zircon and garnet growth may have started at lower T-that is, both resulting in slight underestimation of the actual rateit is at the lower limit of known sinking rates of 12-100 km/Ma in established modern subduction zones (England et al., 2004;Jarrard, 1986). Either the subduction rate was slow, or the sampled rocks were decoupled during prograde slab sinking and dragged down at a lower rate than the remainder of the slab. Further, detailed quantitative petrologic investigation is required to reveal the P-T conditions dated by the Lu-Hf method and to test whether the blocks share a similar subduction rate.
In summary, our findings point to a lengthy progression of these blocks at high P and T, followed by relatively uniform cooling at ≤136°C/Ma at amphibole-eclogite facies conditions. Blueschist facies lower-T cooling became nonuniform and slowed to~15°C/Ma with a potential residence time.

Deformation and Rind Formation
Early isoclinal folding and late fracturing has been observed in multiple Franciscan high-grade blocks including the herein studied PG23 (eclogite/amphibolite; Tsujimori et al., 2006) and Heart Rock (epidoteblueschist; Carruthers & Rowe, 2017). The isoclinal folding of PG23 occurred under prograde amphibole-eclogite facies conditions (Tsujimori et al., 2006;this study). Ages of phengite bearing crosscutting veins in Heart Rock and TIBB indicate blueschist facies fracturing of the blocks during exhumation 10.1029/2020TC006305 Tectonics and lack of significant ductile deformation after~141 and~150 Ma, respectively. For the Sunset boulders block, our data indicate that the actinolite-chlorite rind formed approximately at the time of retrograde hornblende paragenesis overprinting, prior to 154.5 ± 2.3 Ma. This is in line with the results of Catlos and Sorensen (2003). Depending on the interpretation of the rind as having formed at the block periphery or as internal selvage (section 2.2), it indicates encasement in a serpentinite matrix or imbrication with serpentinite during retrogression. Current understanding of the rheology of metabasites and serpentinites does not allow for isoclinal folding of the former in a matrix of the latter, because metabasites are significantly stronger than serpentinized peridotite (Okazaki & Hirth, 2020;Proctor & Hirth, 2016). The chronology of deformation and rind formation thus indicates that PG23 and (possibly all blocks) were part of larger coherent units at the time of prograde metamorphism, underwent plastic deformation to varying degrees, and were encased in or intercalated with serpentinite only during exhumation and retrogression. Anczkiewicz et al. (2004), Cooper et al. (2011), andMulcahy et al. (2014) compared peak temperatures and metamorphic ages of blocks and coherent terranes. They found that the oldest blocks (dated by Lu-Hf garnet) preserved the highest temperatures and interpreted this trend to result from a slowly subducting and slowly cooling subduction zone (Figure 6a). Numerical models of an initiating subduction zone best fit the trend for a slow subduction rate of 10 km/Ma (Cooper et al., 2011). Evaluating the expanded P-T-t dataset sheds a different light: In P-T space (Figure 6b), the blocks and coherent slabs with quantitative P and T estimates span a field with average geothermal gradients ranging 8-14°C/km, covering most of the observed global range of subduction zone metamorphism (Figure 6b; cf. Penniston-Dorland et al., 2015). Both coherent slabs and blocks span the full range of average geothermal gradients. The ages of prograde to peak metamorphism determined here and elsewhere do not support an overall cooling geothermal gradient in the subduction zone from 170-120 Ma. On the contrary, the units metamorphosed after~158 Ma (blueschist portion of Willow Spring slab and South Fork Mountain schist) preserve average geothermal gradients in the hotter portion of the observed range. We recognize that quantitative pressure constraints for these metabasites are limited and often rely on quite general constraints and that the instantaneous Franciscan geotherm represents a balance between the thermal state of the incoming and overlying plates (e.g., Wakabayashi, 2015). However, given the available data, our collated observations support a scenario in which the average geothermal gradient of the Franciscan subduction zone did not significantly change, or even subtly increased, after~158 Ma.

Tectonic Implications
The slow and potentially variable tempo of prograde metamorphism of the high-grade blocks and coherent units with sinking rates ≤18 km/Ma can be explained by either subduction rates at the slow end of modern subduction zones or early decoupling of the units from the downgoing plate as proposed by the circular blockin-mélange flow model after Gerya et al. (2002). However, additional observations leave the hypothesis of metabasite block subduction to eclogite facies conditions and exhumation from there inside a circular flow

10.1029/2020TC006305
Tectonics mélange unsubstantiated: (1) The prograde isoclinal folding observed in some blocks later encased in a weaker serpentinite matrix indicates that they were still parts of larger (coherent) units during subduction, which is further supported by (2) the lack of documented prograde metamorphic rind formation. This may still be consistent with a version of the model in which plucking of metabasite from the slab occurs only at peak conditions. However, prediction of a "chaotic" timing of progression and retrogression (predicted by circular mélange flow models) is not substantiated, as (3) there is little evidence for prolonged, coeval prograde and retrograde amphibole-eclogite-facies metamorphism among different blocks and (4) punctuated, event-like, high-grade exhumation contradicts the predicted "chaotic" timing of progression and retrogression (Figure 2b). Instead, we suggest that the eclogite/amphibolite blocks were subducted at very slow rates and as coherent units-maybe part of the downgoing slab-that experienced significant differential stresses and deformation during the initial~15 Ma of the Franciscan subduction zone.
The regionally uniform cooling of blocks and coherent sheets from peak-temperatures to <500°C has an event-like character (Figure 2b). This high-grade part of the retrogression varies along strike from 165 Ma in northern California to 160-157 Ma in central California.
Our data suggest forcing of the onset of exhumation by a feature in the subducting plate that asynchronously entered the subduction zone. Besides the above discussed exhumation of blocks-in-mélange relative to the boundaries of the mélange, four exhumation mechanisms have been previously proposed for exhumation of the Franciscan Complex: (1) extrusion of material driven by relatively low-density serpentinite from beneath the forearc mantle (Ernst, 1970;Wakabayashi, 2017a), which can be similar to (2) extrusion of a wedge bounded by a thrust below and a normal fault above (Maruyama et al., 1996), (3) normal faulting (Platt, 1986), and (4) erosion and thrust faulting (Ring & Brandon, 1994; see also Wakabayashi, 1999, for critique).
Mechanisms 3 and 4 explain exhumation within an accretionary wedge, i.e., in the blueschist facies. They cannot explain the initial exhumation from depths up to 80 km, particularly under the absence of a significantly large accretionary wedge at the time. Assuming buoyant ascent or extrusion (mechanisms 1 and 2) of serpentinite is responsible for exhuming eclogite-facies units, what may have caused this process to have occurred for only a brief phase at~160 Ma, but not before or after? A prominent change in the subduction zone proposed for this time is the subduction of a ridge interpreted to explain near-trench, late-stage magmatism observed in the CRO that occurred at 157-150 Ma (Hopson et al., 2008;Shervais et al., 2004; Figure 7). Subduction of the successively younger crust toward an active ridge would change the dip and bending radius of the subduction zone and thus of the buoyancy of weak material in the subduction channel (Mancktelow, 1995). The absence of a thermal overprint of the blocks suggests that they did not pass through the thermal anomaly created by the subducting ridge near the trench, that is, they did not exhume to near surface conditions in one step. Instead they were only exhumed to an intermediate depth as long as the thermal anomaly persisted (Figure 7). This agrees with our observation of a decreasing cooling rate in the blueschist facies.
Today the CRO overlies the Franciscan Complex along a >60 km (perpendicular to strike) long contact without intermittent forearc (CRO) mantle (Constenius et al., 2000). To explain the petrologic and Figure 7. Proposed tectonic evolution of the early Franciscan subduction zone system. (a) The map shows a plate configuration in which a spreading ridge subducts beyond the Franciscan trench. The ridge-trench-trench triple junction moves south over time. (b) After initiation at~170 Ma the subduction zone rolled back and the CRO formed along the retreating trench. Coherent units of metabasites subducted at a slow rate and were partly isoclinally folded. At~160 Ma ridge subduction (Shervais et al., 2004) or near-trench approach of a ridge (Wakabayashi, 2017b) with young buoyant crust forced trench advance and sheared the CRO crust of its forearc mantle. The forearc mantle-assumed to have a high density-then sank at a higher rate compared to the subducting slab resulting in forearc mantle extraction and exhumation of the slab. Fragmentation of coherent units began forming the high-grade blocks. Buoyant serpentine ascent likely aided exhumation. Forearc mantle extraction allows exhumation of the blocks without them passing through the thermal anomaly above the subducting ridge. By 150 Ma the blocks were exhumed into the blueschist facies and welded beneath the CRO crust along an extraction fault. Following exhumation was slower and may have occurred along discrete faults. CRO, Coast Range ophiolite.

10.1029/2020TC006305
Tectonics geochronologic data and modern crustal structure we propose forearc-mantle extraction as an additional (or alternative) exhumation mechanism to buoyant ascent, for the initial high-grade exhumation of eclogite/amphibolite units into the blueschist facies. In this context, the forearc-mantle wedge beneath the CRO and above the subducting Franciscan slab detached off the CRO crust when a relatively buoyant subducting ridge entered the subduction zone. The forearc mantle then sank, due to its relatively higher density (Figure 7). We define forearc mantle extraction as the forearc mantle being removed at a higher motion rate compared to the subducting slab resulting in differential exhumation of the slab. In this respect it differs from forearc mantle erosion where the subducting slab has a higher relative downward velocity.
The apparent increase of the average geothermal gradient documented in blueschist-facies units exhumed after~157 Ma (Figure 7) and the initially low sinking rate are also plausible with a subducted ridge, particularly due to younging of the subducting oceanic crust. The Franciscan Complex at~160 Ma may thus be analogous to the proposed extraction of the forearc mantle between the Yuli Belt and the Coastal Range of Taiwan after its detachment from the Philipine sea plate (Sandmann et al., 2015;Shyu et al., 2011). There, the anticipated trigger for the detachment and extraction was the buoyant Chinese continental margin entering the subduction zone.
After exhumation into the blueschist facies and welding to the base of the CRO crust, cooling/exhumation rates decreased and became more variable because discrete faults or shallowly operating subduction channel return flow became responsible for further exhumation (Platt, 1986;Ring & Brandon, 1994;Ukar & Cloos, 2019). Formation of the present-day outcropping mélanges involved further tectonic and sedimentary mixing after exhumation from the amphibole-eclogite facies into the blueschist facies by~150 Ma (Wakabayashi, 2011(Wakabayashi, , 2012(Wakabayashi, , 2015. The lack of exhumation of any rocks higher-grade than blueschist facies metamorphosed after~157 Ma supports our interpretation that the exhumation of amphibole-eclogite facies units was a punctuated effect related to a geodynamic anomaly, and not a steady-state process in the nascent Franciscan subduction zone. Similar interpretations have been drawn in other subduction zones, for example, in the Zagros in Iran (Agard et al., 2006)-suggesting that exhumed subduction-zone rocks may provide unrepresentative snapshots of the thermal and geodynamic structure compared to modern subduction zones.
Our expansion of the chronologic understanding of Franciscan subduction initiation and timing of high-grade exhumation correlates with other significant tectonic events within the Cordilleran chain. Initiation of Franciscan subduction prior 175 Ma correlates with accretion of the Quesnellia island arc in the Canadian cordillera and initiation of magmatism in the Talkeetna arc in southern Alaska. High-grade exhumation in the Franciscan subduction zone at~160 Ma follows shortening and ophiolite obduction in the north Alaskan Brooks range and was followed by overthrusting of the Josephine ophiolite in the Klamath Mountains. Though these connections are preliminary, further high-precision geochronologic work on the precise timing of processes in circum-Pacific regions-along with geodynamic modelingmay allow genetic linkages between these apparently correlated geodynamic changes.

Conclusion
We can draw the following conclusions regarding the P-T-t history of individual blocks and coherent units: 1. Units subducted into the amphibole-eclogite facies were only exhumed during the first~20 Ma (180-160 Ma) after Franciscan subduction initiation. 2. Prograde metamorphism of some units lasted ≥15 Ma suggesting slow sinking rates ≤18 km/Ma. 3. Chlorite-actinolite rinds formed in the amphibole-eclogite facies by ≥155 Ma, indicating high-grade exhumation of the blocks occurred in contact with serpentinite. 4. Exhumation of metabasites from the amphibole-eclogite facies was fast and occurred in a narrow time window at 168-157 Ma with a southward younging trend. It was a punctuated process that coincided with tectonomagmatic discontinuities in the Franciscan subduction system and the circum-Pacific subduction systems. 5. Exhumation of high-grade units slowed in the blueschist facies and became more variable (comparing different units) with the duration between 40 Ar/ 39 Ar hornblende and phengite dates ranging 1.2-14 Ma. 6. The herein expanded P-T-t data set for Franciscan high-grade units does not support a previously interpreted cooling of the subduction zone. To the contrary, units exhumed after~157 Ma yield subtly higher average geothermal gradients.

10.1029/2020TC006305
Tectonics Contrary to previous speculations, metamorphism in the Sierra foothills Red Ant schist along the Melones shear zone is significantly older and temporally unrelated to Franciscan subduction initiation. Our findings indicate that the exhumation of eclogites and amphibolites in the Franciscan Complex was triggered by a geodynamic one-time event rather than a continuous steady-state process. We propose that forearc-mantle extraction may best explain the available observations made in the Franciscan subduction system at~160 Ma.

Data Availability Statement
All analytical data are accessible via OSF.IO/Q3GJN.