Evidence for coseismic subsidence events in a southern California coastal saltmarsh

Paleoenvironmental records from a southern California coastal saltmarsh reveal evidence for repeated late Holocene coseismic subsidence events. Field analysis of sediment gouge cores established discrete lithostratigraphic units extend across the wetland. Detailed sediment analyses reveal abrupt changes in lithology, percent total organic matter, grain size, and magnetic susceptibility. Microfossil analyses indicate that predominantly freshwater deposits bury relic intertidal deposits at three distinct depths. Radiocarbon dating indicates that the three burial events occurred in the last 2000 calendar years. Two of the three events are contemporaneous with large-magnitude paleoearthquakes along the Newport-Inglewood/Rose Canyon fault system. From these data, we infer that during large magnitude earthquakes a step-over along the fault zone results in the vertical displacement of an approximately 5-km2 area that is consistent with the footprint of an estuary identified in pre-development maps. These findings provide insight on the evolution of the saltmarsh, coseismic deformation and earthquake recurrence in a wide area of southern California, and sensitive habitat already threatened by eustatic sea level rise.


Results
We collected 55 sediment cores in the saltmarsh to explore the geologic history of the wetland (Fig. 2, see Supplementary Table S2). Most sediment cores collected were reconnaissance gouge cores, examined visually to look for sharp lithological changes and to identify locations in the saltmarsh to collect large diameter cores for sedimentological analyses (see Supplementary Fig. S3). Lithological descriptions of the reconnaissance gouge cores combined with sedimentological analyses of four vibracores and one piston core indicate that four distinct lithostratigraphic units extend across the saltmarsh (Figs 3 and 4). We describe the lithostratigraphic units based on the sedimentological analyses conducted on the large diameter cores. The basal unit consists of gray organic-poor silty sand containing sparse mica and scattered and clustered shell fragments, which is overlain by interlayered peaty to organic-rich mud or coarse-grained micaceous silty sand. Marine diatoms occur in the basal unit, and most are broken and show abrasion at the edges (Fig. 5, see Supplementary Data S4). We also found fragmented and abraded disarticulated pieces of a bivalve, Chione californiensis, which lives in mud flats, estuarine environments and shallow marine settings 22 . The fragmented and abraded diatoms and shells suggest that the marine organisms were living in an environment exposed to wave energy, indicating the basal unit may represent a period when the site was a shallow marine embayment.
To understand the ecological changes associated with the layers above the basal unit, we paired microfossil analyses with sedimentological data in two representative vibracores, 02VC and 18VC (Figs 5 and 6, see Supplementary Data S4 and S5), which are separated in the saltmarsh by approximately 650 m (Fig. 2). As illustrated in Fig. 5, the transition in 02VC across the sharp depositional contact at 222.5 cm depth is characterized by a change from the expected saltmarsh deposits to organic-poor micaceous silty sand, accompanied by key changes in the microfossils. For example, at 229 cm we document intertidal saltmarsh deposits characterized by preservation of abundant intertidal diatoms and foraminifera in organic-rich silt with low magnetic susceptibility, similar to the deposits on the saltmarsh surface today (Fig. 5, see Supplementary Data S4). At 227 cm, the deposits show a slight drop in the abundance of intertidal diatoms and a decrease in the intertidal foraminifera abundance by ~45%, commensurate with the appearance of fresh-to-fresh brackish (f-fb) diatoms. The f-fb diatoms are an important indicator, as they are abraded and fractured and likely allochthonous, suggesting introduction of a freshwater source to the saltmarsh. The f-fb diatom taxa preserved are consistent with subaerial and shallow freshwater and alkali environments; possibly similar to those historically mapped fringing the Anaheim Bay estuary (Fig. 2). The sediments in this interval show an increase in silt and magnetic susceptibility and decrease in organic content, changes also consistent with a shift in paleoenvironmental conditions that occurs between 229 cm and 227 cm.
Indications of freshwater and tidal water mixing persist up the core (Fig. 5). At 225 cm, abundant allochthonous f-fb diatoms occur with sparse intertidal diatoms, and no foraminifera are present. At 222.5 cm there is a sharp depositional contact where the underlying organic-rich silt is buried by a thick (~100 cm) layer of organic-poor coarse-grained micaceous silty sand. By 200 cm the limited intertidal diatoms are absent; the only remaining microfossils are sparse abraded and fractured f-fb diatoms (Fig. 5). At ~100 cm depth, the sediment starts to return to saltmarsh deposits.
Microfossil analyses in 02VC across the lithologic transition at 40 cm indicate that muddy peat containing intertidal diatoms is overlain by organic-rich silt containing f-fb diatoms and intertidal diatoms (see Supplementary Data S5). At 324 cm, abundant intertidal diatoms occur in organic-rich sandy silt (Fig. 5). An abrupt lithological transition occurs at 320 cm. A spike in MS and decreased organic content and percent sand characterizes the transition. At 305 cm, the lithostratigraphic unit contains abundant intertidal diatoms and sparse allochthonous f-fb diatoms hinting at the introduction of freshwater at the site. By 295 cm, f-fb diatoms are more abundant than intertidal diatoms. At approximately 280 cm, coarse-grained micaceous sandy silt buries the organic-rich mud. The character of this transition is similar to what occurs shallower in the core across the sharp depositional contact at 222.5 cm (Fig. 5). Microfossil analyses in 18VC indicate that coarse-grained micaceous silty sand deposits completely devoid of intertidal diatoms and foraminifera abruptly bury relic intertidal deposits at three distinct depths: 45 cm, 198 cm, and 253.5 cm (Fig. 6). Based on the combined analyses in 02VC and 18VC, we identify three separate events (E1, E2, and E3) in the vibracores where similar lithologic transitions occur and paleoecological data indicate a change in depositional environment. Furthermore, using stratigraphic position and sedimentological observations documented in the reconnaissance cores, we correlate the events across the saltmarsh (Figs 3 and 4). Radiocarbon dating of charcoal, organic remains and shells indicates the events occurred 2085-1910 cal yrs BP (E3), 1603-1302 cal yrs BP (E2), and 632-363 cal yrs BP (E1) (Fig. 7, see Supplementary Table S6).

Discussion
We interpret the lithostratigraphic sequence in the Seal Beach saltmarsh as recording three abrupt paleoenvironmental changes from saltmarsh deposits, to predominantly freshwater influenced silty sand, and then a return to a saltmarsh environment. What historic and geologic factors could contribute to the deposition and preservation of these events? During the winter of 1938, a historically large flood blanketed the coastal plain of Los Angeles and Orange County with > 1-m thick deposits of silty sand 12 , suggesting the possibility that the thick micaceous silty sand deposits in the saltmarsh could be the remains of terrestrial flooding events. However, radiocarbon ages establish that the uppermost silty sand deposits at Seal Beach predate historic flooding events and suggest the estuary may not be a good recorder of major floods 12,23 . Thus, preservation of the silty sand units likely requires production of accommodation space or other mechanisms to isolate the silty sand from coastal erosion and sea level rise. Vibracore 02VC shows that intertidal marsh deposits originally deposited at sea level are now at a depth of ~350 cm below the modern marsh surface (Fig. 5). These marsh deposits have a maximum age of about 2700 cal BP (Fig. 7), leading to an average accumulation rate in 02VC of about 1.3 mm/yr. This rate exceeds the late Holocene relative sea level rise rate of 0.8 + /− 0.3 mm/yr recently established for southern California 24 indicating subsidence could be controlling the preservation of the silty sand units. Additionally, the sedimentary sequence at Seal Beach is different than the wetland sites used for the sea level curve, which show relatively constant marsh aggradation as sea level rises 24 . This suggests that a feature local to the Seal Beach saltmarsh is needed to produce and preserve the abrupt transitions indicated by events E1, E2, and E3.
The geometry of the Newport-Inglewood fault within Sunset Gap (Fig. 1) provides a mechanism that could produce coseismic subsidence, episodically increasing accommodation space and enabling preservation of the micaceous coarse-grained sediment. Extensive oil and water well data in the region show that within Sunset Gap At an even more local scale, the geometry of the fault zone between Bolsa Mesa and Landing Hill would also tend to accentuate subsidence in the center of the saltmarsh due to a small (5°) right step in the fault (Figs 1 and 2). The step in the fault zone could be resolved as a small pull-apart basin at least 1 km wide, similar in scale to other steps and bends in the fault such as a graben northwest of Huntington Beach Mesa where paleoearthquakes are documented 26 and a small push-up ridge at Signal Hill located to the NW (Fig. 1).
The work of 4 was modified by 5,6 in order to establish criteria indicative of coseismic subsidence in wetland environments straddling extensional step-overs along strike slip faults. The criteria established are: (i) an abrupt upper sedimentary contact; (ii) diatom evidence for relative sea level change; (iii) evidence for sustained submergence and/or rapid aggradation; (iv) graben or releasing fault geometry; (v) wide lateral extent of submergence; and (vi) synchrony with other paleoseismic events. Comparing the Seal Beach saltmarsh data set to these criteria suggests that the abrupt burial events are consistent with the hypothesis that earthquakes along the Newport-Inglewood fault system resulted in coseismic subsidence of the relic wetlands surface three times in the past 2000 years, occurring approximately every 700 years (Fig. 7).
The earthquakes episodically lower the saltmarsh where a small 1 km wide right step occurs in the fault trace, instantly creating accommodation space and increasing the tidal prism and fetch (Fig. 8). Immediately following the subsidence, mixing of tidal water and freshwater begins reducing the concentrations of intertidal diatoms and foraminifera that accumulate in the submerged basin with sparse allochthonous f-fb diatoms. As the surrounding freshwater wetlands erode, intertidal microfossils disappear and only f-fb diatoms are preserved in low concentration. The basin is subsequently filled during freshwater floods by sediment that is nearly devoid of microfossils or does not contain microfossils (Figs 5 and 6). Once the accommodation space created by the coseismic subsidence fills, the wetland returns to a similar pre-earthquake saltmarsh (Fig. 8).
Comparison of the proposed paleoearthquake record at Seal Beach to regionally documented earthquakes reveals interesting patterns in the size and frequency of potential earthquakes on the Newport-Inglewood fault. The largest historic earthquake on the fault is the 1933 (M L = 6.3) Long Beach earthquake, which ruptured a 13-16 km-long section of the fault between the cities of Newport Beach and Long Beach 27,28 . The earthquake resulted in extensive structural damage in the Los Angeles region and 120 deaths 27,28 . There was no surface rupture mapped during field investigations following the 1933 earthquake, although liquefaction and lateral spreading did occur 13,20,27 . Cracking was reported "along and near the structural [fault] zone" in the saltmarsh 15 but we observe no evidence of the 1933 earthquake in the cores, consistent with minimal land-level changes and no lateral offsets measured during this moderate sized earthquake. Consequently, the M L 6.3 historic earthquake Distance Across Saltmarsh (m) 21   provides a minimum estimate for the size of the paleoearthquakes, since the historic earthquake was not associated with any subsidence. The most recent earthquake (E1) identified in the saltmarsh is contemporaneous with a surface-rupturing earthquake documented along the Rose Canyon fault, during which approximately 3 m of slip occurred 29,30 . If the 3 m offset is the average for this Rose Canyon event, empirical scaling 31 permits that the rupture could be ~225-km long, extending northwest of the saltmarsh, providing both temporal and geomorphic evidence for major (Mw7.6) earthquakes on the system. Alternatively, considering dating limitations at each site and Coulomb modeling 17 , E1 could represent the northernmost rupture of a sequence of intermediate sized earthquakes that sequentially ruptured up the coast 21 . The only long-term estimate of paleoearthquake recurrence for the Newport-Inglewood fault is from near the Huntington Beach oil field, where 26 used continuous cone penetrometer borings to locate offsets in stratigraphic units in a small graben on the fault. They interpret five Holocene earthquakes. Of this record, only the youngest earthquake near the Huntington Beach oil field could be correlative with one of the events in the Seal Beach saltmarsh.
We propose that the saltmarsh coseismically subsided three times in the past 2000 years. It is possible that the entire historic Anaheim Bay estuary is related to coseismic deformation along the Newport-Inglewood fault (Figs 1 and 2). Future earthquakes that result in subsidence of the saltmarsh may present serious hazards to the U.S. Naval Weapons Station Seal Beach, Seal Beach National Wildlife Refuge, Huntington Harbor, and other southern California coastal communities.

Methods
Field Techniques. Reconnaissance cores were described visually, specifying sediment type, color, presence/ no presence of mica flakes, and the relative amount of organic material. For deeper probing at a site, 1-m long extension rods were attached to the end of the core barrel and the procedure was repeated until penetration was no longer possible. In order to collect sediment for laboratory analysis, we collected three Livingstone piston cores (5.08 cm diameter) and four vibracores (7.62 cm diameter) (Fig. 2).
Quantitative Sediment Analyses. The large diameter cores were cut in half lengthwise, scraped clean with a razorblade (perpendicular to the length of the core), measured, photographed, visually described, and subsampled at 1-cm contiguous intervals into 1-cm 3 plastic cubes. Changes in lithology were determined through measurement of magnetic susceptibility 32 , loss-on-ignition 33 (% total organic matter), and grain size. Mass magnetic susceptibility was determined using a Barrington MS2 Magnetic Susceptibility meter at 0.465 kHz in SI   units of 10 −7 m 3 kg −1 . Grain size analysis was performed at 1-cm sample resolution using a Malvern Mastersizer 2000 laser diffraction grain size analyzer. Prior to running the grain size analysis, the samples were treated with 25-50 mL of 30% H 2 O 2 to obliterate organic matter, 10 mL 1 N HCl to dissolve carbonates, and 10 mL 1 N NaOH to eliminate biogenic silica. The grain size analyzer's performance was measured regularly every 10th sample using a 1-16 micron standard tuff.

Diatoms.
To remove organic material from samples, approximately 1-3 cm 3 of sediment was placed in a beaker with enough 30% H 2 O 2 added to submerge the sample. After the reaction completed, the sample was rinsed clean by filling the beaker with distilled H 2 O and left to settle for at least 1 hr before decanting. This step was repeated several times to remove all remaining H 2 O 2 as well as clays in suspension. The material was then transferred to a graduated tube and diluted 10:1 with distilled water. An aliquot of the cleaned sample was withdrawn using a mechanical pipette, and transferred to a glass coverslip and allowed to dry. The cover slip is permanently mounted to a glass slide with a mounting medium applicable for diatom studies; Naphrax mounting medium was used in this project.
Evaluation of modern samples from the vegetated saltmarsh surface and tidal mud flat indicated that prominent taxa are similarly distributed in both environments (see Supplementary Data S4). A few diatom taxa show a stronger affinity for either the mud flat (e.g., Achnanthes intermedia) or saltmarsh (e.g., Nitschia fasciculata), but many clearly occur in both environments. Thus, for the purposes of this study, no effort was made to differentiate between deposits that may have accumulated on a tidal mud flat from those originating in a vegetated saltmarsh environment. Diatoms were abundant in the modern intertidal samples, but abundance varied widely in down core samples. For all samples, an attempt was made to count at least several hundred valves to provide reliable comparisons between different stratigraphic units. This was not possible for some down core deposits that contained only sparse, presumably reworked valves, or that were devoid of diatoms. Results are expressed as concentration in estimated valves per cc of sediment, as well as relative percentage of the total number counted per sample. Diatoms were identified to species level whenever possible using 34-48 . Foraminifera. To remove plant macrofossils and fine-grained material, samples consisting of 1.2-4.1 cm 3 of sediment were wet-sieved through 500 μ m and 63 μ m sieves. The > 500 μ m fraction was examined for larger foraminifera before being discarded. To split the foraminiferal, the fraction between 63 and 500 μ m was subdivided into eight equal aliquots using wet splitter 49,50 . Where possible, at least 250 specimens were wet-counted under a stereoscopic binocular microscope using reflected light. Taxonomic identifications follow [50][51][52][53][54][55] . Juvenile specimens of Trochammina inflata, Jadammina macrescens that were difficult to distinguish from each other because of their small sizes were lumped into a single group and assigned as juvenile Trochamminids. Specimens of the genus Ammobaculites were combined into a single group, because these species were often broken making it difficult to identify them to the species level 54,55 . All counts were expressed as numerical abundance of living and dead foraminiferal numbers per 10-cm 3 of bulk sediment and as a relative abundance (%) of species out of the total group. Radiocarbon analysis. Accelerator Mass Spectrometry radiocarbon dating was performed on 35 samples selected from cores 02VC, 18VC, 19VC, and 14PC (see Supplementary Table S6). Shell samples were picked directly from the sediment cores with metal tweezers, rinsed with deionized water and placed in labeled glass vials to dry before being sent for analysis. To sample for charcoal and organic material, approximately 1-cm 3 of sediment was removed from the cores using a metal spatula while attempting to avoid large vertical roots that could be visually traced extending downward in the core from the surface of the saltmarsh. The sediment was placed in a 125 m sieve, rinsed with deionized water, and charcoal and organic material for dating was picked using a binocular microscope. The samples were placed in labeled glass vials and dried before being sent for analysis.
The radiocarbon samples were analyzed at either the Keck Carbon Cycle AMS Facility at the University of California, Irvine, or at the Center for Acceleration Mass Spectrometry at the Lawrence Livermore National Laboratory (see Supplementary Table S6). The radiocarbon ages obtained on samples of vegetated plant material reflect the initiation of photosynthesis in the saltmarsh or at a time out of the saltmarsh prior to deposition as detritus. Modern roots injected below the saltmarsh surface may have contaminated detrital material or material growing in-situ. Thus, we infer that stratigraphically "young" ages result from samples contaminated by roots introducing younger carbon to the sample. We infer that "old" stratigraphic units are detrital, resulting from residence time within the environment prior to deposition in the saltmarsh 56 . Shells were corrected for the marine reservoir effect using the local reservoir correction for Anaheim Slough 57 . In order to develop a chronological sequence of the abrupt changes in the depositional environments of the saltmarsh, and constrain the timing of the earthquakes, we combined twenty-seven laboratory reported radiocarbon ages obtained from stratigraphically correlated deposits using the radiocarbon calibration program, OxCal v4.2 58 IntCal13 and Marine13 59 (see Supplementary Data S7). OxCal v4.2 58 applies Bayesian statistics that integrate chronological restrictions to better constrain the calibrated radiocarbon probability distributions. Reported radiocarbon ages of the charcoal, plant material, and shells are reported in calibrated years BP 59 .