Sedimentary signals of recent faulting along an old strand of the San Andreas Fault, USA.

Continental transform fault systems are fundamental features in plate tectonics. These complex systems often constitute multiple fault strands with variable spatio-temporal histories. Here, we re-evaluate the complex history of the San Andreas Fault along a restraining bend in southern California (USA). The Mission Creek strand of the San Andreas Fault is a major geologic structure with ~90 km of strike-slip displacement but is currently mapped as inactive. Quaternary deposits record sediment dispersal across the fault from upland catchments and yield key markers of the fault's displacement history. Our sediment provenance analysis from the Deformed Gravels of Whitewater and the Cabezon Fanglomerate provide detrital geochronologic and lithologic signatures of potential sources within the San Bernardino Mountains and Little San Bernardino Mountains. Statistical analysis shows that the Cabezon Fanglomerate is most compatible with the Mission Creek and Morongo Valley Canyon sources, rather than the Whitewater Canyon as previously suggested. We propose that displacement since deposition ~500-100 ka across the Mission Creek strand has separated these deposits from their original sources. These findings challenge the current paradigm that the Mission Creek strand is inactive and suggest that the fault continues to be a primary structure in accommodating deformation along the Pacific-North American plate boundary.


Sedimentology and stratigraphy 8
We conducted field mapping and sedimentary facies analysis of a ~220 m thick composite measured 9 stratigraphic section, here called the Sagebrush Section, through the tilted Deformed Gravel of 10 Whitewater (Qd) and overlying sub-horizontal Cabezon Fanglomerate (Qo), located in the Mission 11 Creek Preserve ( Figure S1). Lithological descriptions, bed thicknesses, facies, and sedimentary 12 structures were described at decimeter scale ( Figure S1). Where preserved, paleoflow directions 13 were determined from the orientations of imbricated conglomerate clasts. Sedimentary provenance 14 data collected from this section include clast counts, sandstone petrography, and detrital zircon U/Pb 15 geochronology. Lithofacies shown in Figure S1 are modified from Miall (1978) and DeCelles et al. 16 (2015). 17

Modal analysis of clast compositions 18
Clast compositional information was collected from twenty sampling stations, including the active 19 drainages and Quaternary deposits (Table S1). To minimize bias toward more durable clast types, we 20 used the area counting technique (e.g., Howard, 1993) for all cobble-sized clasts (between 64 mm 21 and 256 mm in diameter) until 100 counts were reached for each station. We report normalized 22 compositions from twelve diagnostic clast lithologic compositions, Biotite Gneiss, Deformed 23 Granite, Amphibolite and Mafic Schist, Monzonite, Pink K-feldspar monzonite, Granite and 24 Granodiorite, Diorite, Coarse diorite, Volcanic, Quartzite, Phyllite, and Marble. These clast types are 25 modified from Sadler et al. (1993) for comparison with published clast compositional data from 26 modern drainages. Recalculated data are shown in Table S2 and Table S3 for modern drainages and  27 Quaternary deposits, respectively. 28 29 3. Modal analysis of sandstone petrography 30 31 Fifteen samples from medium-grained sandstones were collected from the Deformed gravels of 32 Whitewater (Qd), Cabezon Conglomerate (Qo), Qt3 terrace deposits, and active drainages for 33 sandstone petrographic analysis. Each thin section was prepared by Quality Thin Section, LLC., and 34 stained with Alizarin red S to aid in identification of calcic plagioclase and other Ca-rich phases. 35 Samples were point-counted (for 400 grains) following the Gazzi-Dickinson method 5,6 using a Pelcon 36 automated point counting system and a Leica DMZ2700 petrographic light microscope at the 37 University of Connecticut. Grain parameters identified in these point counts are listed in the data 38 repository, and recalculated data are provided in Table S4. All samples are classified as arkosic to 39 lithic-arkosic sandstones (after Folk et al., 1980). Figure S2  are shown on Pb*/U concordia diagrams ( Figure S3) and relative age-probability diagrams using the 87 routines in Isoplot 10 ( Figure S4). The age-probability diagrams show each age and its uncertainty 88 (for measurement error only) as a normal distribution, and sum all ages from a sample into a single 89 curve. 90

Pearson Chi-squared Statistical Analysis of Modern Catchments 91
We evaluated the probability of the detrital clast types and detrital zircon U/Pb ages sourced from 92 each catchment by performing a nonparametric statistical analysis on the clast populations from 93 Quaternary deposits and modern catchments. 94 First, we grouped the twelve observed clast types into three genetically related categories: (1) 95 crystalline metamorphic, (2) intrusive, and (3) metasedimentary, sedimentary, and volcanic. We 96 applied the Pearson chi-square (χ 2 ) statistic 11 as a measure of the goodness of fit between the 97 observed clast lithology categorical data, Oi, (i.e., the Qo sample) and the expected clast lithology 98 distribution based on exposed bedrock, Ei, (i.e. the catchment source) for n = 100 total counts. 99 χ 2 c-1 = Σ (Oi -Ei) 2 /Ei 100 Calculated χ 2 statistics were compared to a P-value = 9.488 for 0.05 level of significance and two 101 degrees of freedom (Table S6). For samples that yield χ 2 values less than the P-value, the null 102 hypothesis is satisfied and we interpret the observed clast or zircon U/Pb age distribution as 103 equivalent to the predicted rock types exposed bedrock lithology. 104 The χ 2 results show that observed detrital datasets satisfy the Chi-squared goodness of fit test and 105 match the expected distributions, with χ 2 values between 0.011-13.530. Only the observed clast 106 distribution of Little Morongo Canyon does not match the predicted distribution, possibly due to 107 higher metasedimentary clasts derived from Morongo Valley. We note that clast type categories that 108 contribute to largest Pearson residuals are the metamorphic lithology (biotite gneiss, amphibolite), 109 and the metasedimentary lithologies, as observed qualitative differences. Figure S5 compares the 110 proportions of rock types based on exposed bedrock lithology, observed detrital zircon U/Pb ages, 111 and clast types. See manuscript text for discussion. 112 We also performed the χ 2 statistical goodness-of-fit test between the modern catchments and the 113 Quaternary deposits. Canyon. 119

Kolmogorov-Smirnoff Statistical Analysis 120
We compare the detrital zircon U/Pb age distributions from the Quaternary deposits with potential 121 source areas using the Kolmogorov-Smirnoff (K-S) statistic on the calculated cumulative zircon He 122 date distributions as a goodness-of-fit criterion 12,13 . Bernardino Mountains (Fig. 1). These sand samples were sieved to a grain size range of 250 to 500 139 µm, and then leached through a series of 2-3% HF acid to isolate the grains of quartz, the beryllium-140 bearing mineral 16 . Following quartz separation and purification, 9 Be was added to the sample as a 141 spike to determine the amount of 10 Be naturally present in the sample. Beryllium was extracted from 142 the sample using ion chromatography and subsequently converted to beryllium oxide 16,17 , which was 143 then mixed with powdered niobium and targeted for accelerator mass spectrometry.  Table S1. Sample locations for detrital zircon U/Pb LA-ICPMS geochronology and sandstone 203 petrography. 204 Table S2. Modal clast compositional data from active drainages. 205 Table S3. Modal clast compositional data from Quaternary deposits. 206 Table S4. Recalculated modal sandstone petrographic point-count data. 207 Table S5. Detrital zircon U/Pb ICP-MS geochronological data. Detrital zircon U-Pb 208 geochronologic analyses by LA-ICP-MS analysis. The * indicates radiogenic Pb (corrected for 209 common Pb). All errors are reported at the 1σ level. 210 Table S6. Chi-squared statistics of observed proportion of rock types and associated detrital 211 zircon age categories (crystalline metamorphic, intrusives, metasedimentary/sedimentary) with 212 expected proportions from exposed bedrock lithology. For samples that yield χ2 values less than 213 the P-value (gray highlight), the null hypothesis is satisfied and the observed and expected 214 distributions are statistically equivalent. 215 Table S7. Chi-squared statistics of observed proportions of clast lithology from the Quaternary 216 deposits with expected proportions from potential source areas. For samples that yield χ2 values 217 less than the P-value (gray highlight), the null hypothesis is satisfied and the observed and 218 expected distributions are statistically equivalent. 219 Table S8. 10 Be data from active drainages. 220 Figure S1. Stratigraphy and sedimentology of the Mission Creek fanglomerates exposed in the 222 Sagebrush section, showing lithofacies, paleoflow measurements, and locations of clast counts 223 and sample for detrital geochronology and sandstone petrography. The tilted Deformed Gravels 224 of Whitewater (Qd) are unconformably overlain by sub-horizontal deposits of the Cabazon 225 Fanglomerate (Qo). Figure S2. Sandstone petrographic data from Quaternary deposits and 226 modern rivers showing relative proportions of framework minerals, tectonic provenance fields, 227 and sandstone composition. Refer to Table S4 for

Quaternary Deposits
Qt2 terrace ll Cabezon Fanglomerate (Qo) Deformed Gravels of Whitewater (Qd) Figure S2. Sandstone petrographic data from Quaternary deposits and modern rivers showing relative proportions of framework minerals, tectonic provenance fields, and sandstone composition. Refer to Table S4 for point-counting data. a) Monocrystalline Quartz, Feldspar-Total Lithics (Qm-F-Lt) data show a dissected and transitional magmatic arc, and to a lesser extent, basement uplift provenance. b) Total Quartz-Feldspar-Lithics (Qt-F-L) data show a stronger dissected magmatic arc provenance, likely biased by polycrystalline quarts (derived from basement gneisses) contributing to higher total quartz component. c) Quartz-Feldspar-Rock Fragments (Q-F-R), showing predominantly arkosic-to-lithic-arkosic sandstone compositions (after Folk, 1968). d) Quartz-Plagioclase-K-feldspar (Q-P-K) data, showing variable quartz and K-feldspar and uniformly low plagioclase content, except for Whitewater River, which has the highest relative proportion of plagioclase. Modal sandstone compositions from Qd are characterized by an upsection transition from basement uplift to dissected magmatic arc provenance fields and increase in plutonic character (Fig. S2), consistent with unroofing of the Mesozoic Sierra Nevada batholith and Proterozoic metamorphic basement of the Mojave Province.   Figure S4. Relative probability distributions of zircon U/Pb ages from the a) modern catchments draining the San Bernardino and Little San Bernardino Mountains, b) Late Pleistocene terrace ll, and c) Mid-Pleistocene deposits measured in the Sagebrush Section. Note break in scale between 400 and 1300 Ma (no zircons of this age range).  Figure S5. Predicted proportions of exposed bedrock lithology, based on contributing areas of drainage basins, and the observed detrital zircon U/Pb age categories and clast types. Observed zircon U/Pb results yield acceptable representations of bedrock lithology, whereas clast types generally over-represent more durable rock types (i.e., Mesozoic intrusives). Creek (see Fig. 1 for location). Red lines show the location of the north and south splay of the Mission Creek Fault. Note the uplifted planar surface between the two splays. We map and interpret the uplifted planar surface is the result of a left step along the Mission Creek Fault from the north splay to the right splay. Base hillshade was generated with ESRI ArcMap v.10.4.1 software (under fair terms of use https://www.esri.com/en-us/legal/copyright-trademarks).