Detrital signals of coastal erosion and fluvial sediment supply during glacio-eustatic sea-level rise, Southern California, USA

Coastal erosion, including sea-cliff retreat, represents both an important component of some sediment budgets and a significant threat to coastal communities in the face of rising sea level. Despite the importance of predicting future rates of coastal erosion, few prehistoric constraints exist on the relative importance of sediment supplied by coastal erosion versus rivers with respect to past sea-level change. We used detrital zircon U-Pb geochronology as a provenance tracer of river and deep-sea fan deposits from the Southern California Borderland (United States) to estimate relative sediment contributions from rivers and coastal erosion from late Pleistocene to present. Mixture modeling of submarine canyon and fan samples indicates that detrital zircon was dominantly (55%–86%) supplied from coastal erosion during latest Pleistocene (ca. 13 ka) sea-level rise, with lesser contributions from rivers, on the basis of unique U-Pb age modes relative to local Peninsular Ranges bedrock sources. However, sediment that was deposited when sea level was stable at its highest and lowest points since the Last Glacial Maximum was dominantly supplied by rivers, suggesting decreased coastal erosion during periods of sea-level stability. We find that relative sediment supply from coastal erosion is strongly dependent on climate state, corroborating predictions of enhanced coastal erosion during future sea-level rise.


INTRODUCTION
A majority of the world's coastlines are net erosive with retreating sea cliffs (Young and Carilli, 2019). Coastal erosion can form an important component of sediment budgets in some sediment-routing systems, contributing to beach nourishment, tourism, and sand resources (Young and Ashford, 2006;Patsch and Griggs, 2007). Rates of coastal erosion are predicted to increase with global sea-level rise in the coming century (Hackney et al., 2013;Limber et al., 2018;Mentaschi et al., 2018), likely exacerbating the consequences of sea-cliff retreat and coastal gullying along heavily populated coastlines, such as in Southern California, where residential and industrial cliff-top developments are widespread (Young and Ashford, 2006). For example, Limber et al. (2018) estimated rates of cliff retreat increasing twofold or greater over the coming century in Southern California.
As coastal erosion is episodic, and historical observations are limited to the past few centuries, estimates of long-term erosion rates are relatively sparse and often have high uncertainties (Patsch and Griggs, 2007), making future predictions challenging (Limber et al., 2018). Although estimates of sediment supply from coastal erosion are available over historical (i.e., ≤10 2 yr) time scales (e.g., Young and Ashford, 2006), relatively few studies (e.g., Rogers et al., 2012;Hurst et al., 2016) have estimated rates of coastal erosion over longer "intermediate" time scales (i.e., 10 2 -10 6 yr; sensu Romans et al., 2016). Therefore, we used detrital zircon (DZ) as a sediment tracer to track the relative contributions from rivers and coastal erosion to deep-sea fans in the Southern California Borderland over the past 40 k.y.

GEOLOGIC SETTING
The Oceanside margin consists of a narrow shelf (∼5 km wide) that connects mountainous topography of the northern Peninsular Ranges with deep-marine basins of the Southern California Borderland (Fig. 1A). The northern Peninsular Ranges are underlain by plutonic rocks of a Lower to mid-Cretaceous magmatic arc and associated volcanic rocks (128-91 Ma;Herzig and Kimbrough, 2014;Kimbrough et al., 2014;Premo et al., 2014). Sea cliffs are present along 80% of the coastline, average 25-35 m in height (locally up to 100 m), and are composed of ∼80% siliciclastic sand-sized material (Young et al., 2010). The Oceanside shelf, sea cliffs, and coastal lowlands are composed of Cretaceous-Cenozoic sediments that onlap older, crystalline basement (Fig. 1B;Darigo and Osborne, 1986;Young et al., 2010).
Five rivers with headwaters within the northern Peninsular Ranges account for ∼80% of the fluvial drainage area to the Oceanside shelf ( Fig. 1). Although dams have reduced sediment supply by rivers to the Oceanside shelf by ∼50% (Patsch and Griggs, 2007), 10 Be-derived denudation rates that equate to ∼2 Mt yr −1 mass flux to the Oceanside margin are in broad agreement with estimates of predam mass flux from 20 th -century stream gauge data (∼2.2 Mt yr −1 ; Inman, 2008;Covault et al., 2011). Rates of historical (1930s to 2010) sea-cliff retreat along the Oceanside margin have varied between ∼1 and ∼100 cm yr −1 (Limber et al., 2018), with an average rate of ∼8 cm yr −1 calculated between 1998 and 2004 (Young and Ashford, 2006). Patsch and Griggs (2007) estimated that bluff erosion accounted for ∼34% of the total Oceanside littoral cell budget prior to anthropogenic modification (i.e., river damming and sea-cliff armoring). Young and Ashford (2006) reported higher contributions (∼84%) from sea-cliff erosion and coastal gullying based on data collected from 1998 to 2004.

Published online 8 September 2021
The Oceanside littoral cell is sand-starved at its northern end, and longshore currents transport sediment in a predominantly southeastward direction along the coastline, thus forming a closed system (Patsch and Griggs, 2007). Three major submarine canyons and numerous gullies incise the Oceanside shelf edge (Fig. 1). During sea-level lowstand, rivers cross the exposed Oceanside shelf to deliver their sediment loads directly into submarine canyons at the shelf edge, including the Oceanside and Carlsbad Canyons, which were actively receiving terrigenous sediment prior to latest Pleistocene transgression ( Fig. 1; Darigo and Osborne, 1986;Covault et al., 2007). During highstand, sediment is transported by longshore currents southward to the Scripps and La Jolla Canyons, which extend within close proximity of the shoreline. The La Jolla fan is only active during highstand because its feeder canyons lack direct connection with a river (Covault et al., 2007).
Sediment sources in the Oceanside margin can be differentiated on the basis of DZ U-Pb ages: (1) The crystalline bedrock of the northern Peninsular Ranges and Upper Cretaceous-Paleocene strata are dominated by Cretaceous U-Pb ages, and (2) Upper Paleocene-Eocene strata are characterized by an abundance of latest Cretaceous, Permian-Triassic, and Proterozoic zircon that lacks a local source in the northern Peninsular Ranges ( Fig. 2; Jacobson et al., 2011;Premo et al., 2014;Sharman et al., 2015). The distinct signatures produced by Cretaceous versus Upper Paleocene-Eocene strata reflect an early Cenozoic shift from local to nonlocal sediment sources that extended as far as Arizona (USA) and Sonora (Mexico) ( Fig. 2; Kies and Abbott, 1983;Sharman et al., 2015).

METHODLOGY
To characterize the DZ signature of river sediment supply and submarine canyon-fan deposits, we sampled (1) each of the five major rivers of the Oceanside margin, and (2) five offshore samples from core and the shallow subsurface (Figs. 1 and 2; Table S1 in the Supplemental Material 1 ). The three core samples were collected from sections with calibrated 14 C ages from the Oceanside fan (sample 503-P1; ca. 15.8-7.2 ka), Carlsbad fan (H5-P1; ca. 20.3 ka), and La Jolla fan (EM3-4, ca. 40 ka) (Covault et al., 2007;Normark et al., 2009). Although the age of the Oceanside fan sample is constrained to be between ca. 15.8 and 7.2 ka based on calibrated 14 C ages (Normark et al., 2009), we interpret that the sample is likely older than ca. 13 ka based on the timing when the Oceanside Canyon was drowned and the Oceanside fan became inactive (Covault and Romans, 2009). Two additional samples were collected from the shallow subsurface (0.3-0.8 m depth) within the La Jolla Can-yon (sample L2-79-SC g68) and La Jolla fan (sample L2-79-SC 98) (Fig. 1). Samples were processed using standard mineral separation and U-Pb analytical procedures (see the Supplemental Material text for details and Tables S2 and S3 for analytical results).
We used the forward mixture modeling approach of Malkowski et al. (2019) to estimate the relative contributions of DZ from rivers and coastal erosion. The five sampled rivers were combined into two parents (P1 and P2) based on similarity in DZ U-Pb age distributions between San Juan and San Mateo Creeks (P1) and between Santa Margarita, San Luis Rey, and San Dieguito Rivers (P2) (Fig. 2). A compilation of eight Upper Paleocene-Eocene samples (523 grain analyses) from sea cliffs and coastal outcrops was used as a proxy for sediment input from coastal erosion (P3; Fig. 2; Table S4). Bestfit mixing proportions and 95% confidence intervals were determined using 5000 iterations of a bootstrapping sampling-with-replacement routine, with V max used as a goodness-of-fit metric (Fig. S1, Table S6; Malkowski et al., 2019).

RESULTS
DZ U-Pb ages from rivers were all dominated by Cretaceous zircon, with a progression from multimodal to unimodal age peaks from northwest to southeast (Fig. 2). San Juan and San Mateo Creeks displayed more latest Cretaceous (90-66 Ma) zircon (16%-20% of the total number) relative to rivers to the southeast (Fig. 1). Four of the submarine canyon  and fan samples (Carlsbad and La Jolla) displayed broadly similar U-Pb age distributions to the rivers. Early to mid-Cretaceous age peaks (114-97 Ma) were dominant in these samples, with modest abundances of latest Cretaceous (6%-13%) or Proterozoic (2%-9%) zircon. The Oceanside fan sample, however, displayed abundant Late Cretaceous zircon (33%) with an age peak of 82 Ma and had an elevated proportion (26%) of Proterozoic zircon. The Oceanside fan sample also displayed a minor Late Jurassic peak (ca. 149 Ma) that was not present in the other marine or river samples (Fig. 2).
Mixture modeling indicated elevated contributions of zircon derived from recycling of Cenozoic coastal outcrops (P3) in the Oceanside fan sample (55%-86% inner 95 th percentile range; Table S6). The four other deep-sea fan samples yielded markedly lower estimates for zircon supply from coastal outcrops (inner 95 th percentile ranges between 0% and 37%; Table S6).

DISCUSSION
DZ ages from the four submarine fan samples deposited during stable lowstand and highstand sea levels (Carlsbad and La Jolla canyon-fans) are consistent with primary derivation from rivers (Fig. 3). For instance, samples show general overlap with the integrated river U-Pb age distribution that was computed by normalizing to the expected sediment load of each river based on catchment area and 10 Be-derived millennial erosion rates (Fig. 2;Covault et al., 2011). Mixing calculations confirmed that these four samples were dominated by river input (average of 76%-89%; Fig. 3; Table S6). Although the DZ ages from the La Jolla canyon-fan system indicate river input, the river sediment is delivered to the littoral zone, where longshore currents deliver it to canyon heads (Fig. 3).
In contrast, the ca. 13 ka Oceanside fan sample, deposited during sea-level rise, contained abundant latest Cretaceous (90-66 Ma) and Proterozoic DZ grains, which lack a local source in the northern Peninsular Ranges (Fig. 2; Premo et al., 2014). Instead, these grains were likely recycled from Upper Paleocene-Eocene, and possibly younger, strata that compose the majority of the Oceanside shelf and coastal inland exposures (Fig. 1B). Correspondingly, mixture modeling results indicated that the ca. 13 ka Oceanside fan sample was dominantly supplied via coastal erosion (i.e., 55%-86%; Fig. 3).
We considered two explanations for the anomalous result from the single Oceanside fan sample (Fig. 2): (1) local derivation from submarine erosion (e.g., mass wasting) of the proximal Oceanside shelf and/or canyon, or (2) sediment supplied from coastal erosion (Fig. 3). Although the first explanation is challenging to rule out, several considerations suggest that the Oceanside fan sample resulted from enhanced coastal erosion during latest Pleistocene and early Holocene sea-level rise. On the basis of sand mineralogy on the Oceanside shelf, Darigo and Osborne (1986) interpreted a Peninsular Ranges batholith source for lowstand Pleistocene deposits and a local, Eocene sea-cliff source for transgressive Holocene deposits that blanketed the shelf north of La Jolla during sea-level rise. Similarly, Covault et al. (2011) noted that deep-sea sediment budgets indicated an imbalance between fluvial input (∼2 Mt yr −1 ) and deep-sea sediment deposition (∼3 Mt yr −1 ) during Holocene sea-level rise; sediment supply from coastal erosion could make up the deficit and close the sediment budget. Terraces on the Oceanside shelf, correlated to latest Pleistocene to early Holocene (ca. 15-10 ka) sea-level oscillations and stillstands, provide additional evidence for shelfal and coastal erosion during sea-level rise (Darigo and Osborne, 1986). Although relative DZ supply cannot be unequivocally related to relative sediment supply from rivers and coastal erosion, given the lack of constraint on zircon concentration (Amidon et al., 2005;Malkowski et al., 2019), our estimate of relative DZ yield from coastal erosion during highstand and lowstand (average of 15%, range of 0%-37%) is within the uncertainty of the Patsch and Griggs (2007) estimate of 33% sediment supply from sea-cliff erosion prior to river damming and cliff armoring.
The DZ results presented herein and sedimentological observations from previous work are consistent with historical observations and predictions of accelerated sea-cliff retreat during periods of sea-level rise (Hackney et al., 2013;Limber et al., 2018;Mentaschi et al., 2018). For example, 4-87 m of sea-cliff retreat has been estimated for 2 m of sea-level rise over the coming century in the vicinity of the city of Oceanside, California (Young et al., 2014;Limber et al., 2018). Changes in wave power, the frequency and intensity of storms, and anthropogenic activity may also promote coastal erosion (Mentaschi et al., 2018;Reguero et al., 2019). Terrestrial paleoclimate proxies from Lake Elsinore (north of Oceanside) suggest decreased precipitation and storm intensity from early to late Holocene time (Kirby et al., 2007). Thus, although river sediment supply may have been high during the relatively wet early Holocene (Wells and Berger, 1967;Kirby et al., 2005Kirby et al., , 2007, deep-sea sediments suggest that any increases in fluvial sediment supply were overwhelmed by even larger increases in coastal erosion, perhaps driven by increased frequency and magnitude of storms in conjunction with sea-level rise.

CONCLUSIONS
We used DZ as a sediment tracer to reveal that rivers supplied the bulk of sediment to deepsea fans in the Southern California Borderland during periods of stable sea level (lowstand and highstand). However, we interpret one sample deposited during latest Pleistocene sea-level rise (ca. 13 ka) to be dominantly supplied by sediment from coastal erosion. These findings suggest that the role of sediment supplied from coastal erosion versus rivers is dependent on sea level and climate state, supporting predictions of enhanced coastal erosion as a consequence of future sea-level rise. Furthermore, we demonstrated the utility of DZ as a sediment tracer of different components of coastal sediment budgets. The sand-sized fraction of deep-sea depositional systems is thus a valuable archive of the effects of environmental changes (such as sea-level rise in response to climate change) on coastal erosion and sediment supply.