Spotila, J.A.; Marshall, J.; DePew, K.; Prince, P.S., and Kennedy, L., 2016. Potential for geologic records of coseismic uplift and megathrust rupture along the Nicoya Peninsula, Costa Rica.
This study presents the first paleoseismic investigation of the Middle America subduction zone at the Nicoya Peninsula, Costa Rica. This megathrust has been intensely studied over the past decade using a range of geologic and geophysical techniques, and it experienced interplate rupture in 2012 (moment magnitude 7.6). Despite many factors that could hinder preservation of a paleoseismic record in this coastal environment, including complex deposition in a tropical mangrove setting, this reconnaissance identifies two sites where stratigraphic evidence may record relative sea-level changes associated with Holocene earthquakes. Although more work is required to better constrain the timing and nature of these events, this study suggests that the Tamarindo and Playa Carrillo estuaries contain possible paleoseismic records. At the main study site of Tamarindo, alternations between mud and peat below 1 m in depth may record relative sea-level change associated with multiple earthquakes between ∼5 and 8 ka. However, this site offers no paleoseismic evidence of late-Holocene earthquakes. Much younger stratigraphy occurs at a Playa Carrillo, where mud-peat alternations from 200 to 500 years ago could represent recent coseismic ground motions. The work presented here is limited to litho- and chronostratigraphy, however, and full interpretation of the relative sea-level histories of both sites will require quantitative, high-resolution biostratigraphic analysis. Although the results suggest paleoseismic records exist along the Nicoya Peninsula, they appear fragmentary and complex and ultimately may not provide a continuous, high-resolution paleoseismic record like those obtained at other subduction zones worldwide.
Megathrust seismogenesis is dynamically controlled as a complex system involving numerous interconnected components, including plate roughness, deformation, mantle convection, hydrology, heat flow, sediment cycling, metamorphism, and other processes (DeShon et al., 2006; Hyndman and Wang, 1993; Kinoshita et al., 2006). Despite obvious societal relevance, it remains incompletely understood how physical and chemical processes affect mechanical behavior, which subduction zones are capable of great earthquakes, and the scale of inherent variability in mode of megathrust rupture (Bilek, Schwartz, and DeShon, 2003; Cisternas et al., 2005; McCaffrey, 2008; Satake and Atwater, 2007; Schwartz, 1999; Sieh et al., 2008; Stein and Okal, 2007).
Extensive effort has been invested in understanding the seismogenesis of the Middle America subduction zone at the Nicoya Peninsula, Costa Rica (see MARGINS Staff, 2004). The behavior of this megathrust at timescales exceeding the instrumental record, however, is poorly known. Paleoseismology is generally the only methodology able to connect a real-time, dynamic, physical-chemical understanding of a fault with its seismogenic behavior over numerous seismic cycles. Perturbations to sedimentary deposits and other features can record the occurrence of large earthquakes as shaking events, tsunamis, and changes in elevation relative to sea level (Carver and McCalpin, 1996). Geologic evidence has contributed significantly to understanding the behavior and hazards of subduction zones with moderate (>0.3 ka) recurrence intervals and significant (>1 m) relative sea-level change in temperate coastal environments, particularly from environments that exhibit strong contrast between freshwater and saline intertidal deposits (e.g. Atwater, 1987; Cisternas et al., 2005; Nelson, Shennan, and Long, 1996; Sawai et al., 2004; Shennan et al., 2014). This approach is more challenging in the complex deposits of tropical mangrove-dominated wetlands, yet it has locally been shown to be effective at documenting paleoseismic events and relative sea-level change (Belperio, Harvey, and Bourman, 2002; Berkeley et al., 2009; Dura et al., 2011; Grand Pre et al., 2012; Jankaew et al., 2008; Monecke et al., 2008; Rashid et al., 2013). The use of recent coastal stratigraphy to document paleoearthquakes may be even more challenging where recurrence intervals are short (<0.1 ka) or coseismic motions are small (<1 m). These and other factors make the feasibility of obtaining a paleoseismic record from the Middle America Trench uncertain. We have explored this feasibility by reconnaissance examination of estuarine deposits along the Nicoya Peninsula.
The Nicoya Peninsula in northern Costa Rica is underlain by an emergent forearc of the Middle America Trench, where the ∼20–25 Ma Cocos plate subducts beneath the Caribbean plate and Panama block at ∼8.3 cm/y (DeMets, Gordon, and Argus, 2010) (Figure 1a). A clear Benioff zone defines the limits of a 6°–13° dipping seismogenic megathrust at a depth of ∼10–28 km (DeShon et al., 2006; Newman et al., 2002; Protti, Guendel, and McNally, 1995; Schwartz and DeShon, 2007). Because of the roughness of the subducting slab, the Middle America Trench is segmented and has tended to fail only in moderate interplate events (i.e. <M8, length <200 km). Segment boundaries and other factors, such as the presence of fluids, result in significant variation in megathrust behavior, including the distribution of locking and upper-plate deformation (Audet and Schwartz, 2013; DeShon et al., 2006; Feng et al., 2012; Fisher et al., 1998; Iinuma et al., 2004; Newman et al., 2002; Norabuena et al., 2004).
Historic earthquakes demonstrate that the trench is seismogenic beneath the Nicoya Peninsula (Figure 1a). Eight large earthquakes (M > 7) have occurred in this area in the past 200 years (Figure 1a) (Protti et al., 2014). Four of these occurred to the NW or SE and define the Nicoya segment. The other four (1853, 1900, 1950, 2012) are considered to have been caused by interplate rupture of the Nicoya segment itself (Iinuma et al., 2004; Protti, Guendel, and Malavassi, 2001; Protti et al., 2014). The recent earthquake of 5 September 2012 (moment magnitude [Mw] = 7.6) ruptured a partially locked zone in the middle of the Nicoya segment, releasing the accumulated strain in the region of the Nicoya seismic gap (Nishenko, 1989; Protti, Guendel, and Malavassi, 2001; Protti et al., 2014). This event had a maximum interplate slip of 4.4 m and maximum geodetically determined coastal uplift of 0.53 m and geologically observed uplift of 0.67 m, thus recovering nearly all of the accumulated strain since the penultimate event in 1950 (Mw = 7.7) (Protti et al., 2014; Yue et al., 2013). The rupture did not extend through the entire up-dip portion of the geodetically defined locked zone (Figure 1a) (Feng et al., 2012), however, suggesting the potential for a future M∼7 event offshore (Protti et al., 2014; Yue et al., 2013).
The similarity in size and rupture area of the 2012 and 1950 events suggests that these are representative for interplate earthquakes on the Nicoya segment (Marshall and Anderson, 1995; Protti et al., 2014). Because of proximity to the trench (∼60 km), the central Nicoya coastline (including our Carrillo site) experiences interseismic subsidence at ∼1 cm/y, which is recovered by 0.5–1.0 m coseismic uplift every ∼50 years (Iinuma et al., 2004; Marshall and Anderson, 1995; Protti et al., 2014). The unusually short recurrence interval could give the Nicoya Peninsula unique value for understanding the dynamic processes that control subduction zone seismogenesis. A complicating factor is that the pattern of elastic subsidence and uplift should decrease away from the center of the locked zone to zero at segment boundaries, but it is unclear where our Tamarindo site falls along this gradient of decreasing strain (Figure 1a). Another complication is that superimposed on the seismic cycle is a component of long-term net uplift, which increases steadily from ∼0.1 mm/y along the northern Nicoya Peninsula to 1–2 mm/y at Cabo Blanco in the SE (Marshall et al., 2012).
There are several potential problems for finding reliable paleoseismic sites in the Nicoya Peninsula. The historical high frequency and small magnitude of relative sea-level change could hinder preservation of a useful paleoseismic record, particularly in the complex, heavily bioturbated deposits of a tropical mangrove environment. Given the short recurrence interval, a paleoseismic record would require rapid sedimentation, such that the sedimentary or paleontological environmental indicators would be preserved and stratigraphically separated into recognizable packages. The expected smallness of the coseismic sea-level changes relative to the noise associated with the large tidal range of the Nicoya coast (∼2–3 m) may be another potential hindrance to stratigraphic recognition of paleoseismic events. Additionally, the Nicoya coast experiences coseismic uplift, whereas subduction zone paleoseismology is more commonly based on coseismic subsidence (Dura et al., 2011; Nelson, Shennan, and Long, 1996; Shennan et al., 2014). Coseismic subsidence is generally easier to capture, given that datable soil horizons are submerged and subsequently preserved by subtidal burial. Records of relative sea-level change are possible in areas that experience coseismic uplift (e.g., Atwater et al., 2004; Briggs et al., 2014; Sawai, 2001; Sawai and Nasu, 2005; Sawai et al., 2004), but only if long-term depositional subsidence or lateral progradation create the accommodation space necessary to preserve subsidence stratigraphy (Nelson, Kashima, and Bradley, 2009). Such a condition may be difficult to find along the Nicoya coastline, given the presence of long-term net tectonic emergence of the forearc (Marshall et al., 2012).
To explore the feasibility of obtaining a paleoseismic record from Nicoya, this study searched for suitable wetlands with intertidal sedimentary environments where recent relative sea-level changes might be recorded. Initially, possible wetlands were identified along the northern two-thirds of the outer Nicoya coastline using maps and satellite images (Figure 1b). Suitable wetlands are rare, given the small size of drainage basins on the narrow outer coast of the uplifting Nicoya forearc (Figure 1b). The depositional environments of four sites appeared promising. These sites were cored for basic stratigraphic characterization and age control, but the focus here was only on the two sites that yielded organic material suitable for radiocarbon dating.
The Tamarindo Estuary lies at the mouth of the 73 km2 Matapalo basin (Battistini and Bergoeing, 1980) (Figure 1b). Although this estuary is located near the northern edge of the Nicoya segment and outside of the locked zone (Figure 1a), it also falls outside of the zone of significant permanent tectonic uplift (Marshall et al., 2012); therefore it is more likely to preserve subsidence stratigraphy. This site can be viewed as a test case for extracting a relative sea-level history from this environment, although the anticipated ground-level changes due to the seismic cycle (as defined by the last few hundred years only) are moderate (tens of centimeters). Although greater magnitude of uplift occurs farther south along the coastline, with a maximum observed in 2012 at Playa Carrillo (Protti et al., 2014), Tamarindo was selected for more intense study because the size of the wetland appeared more likely to yield prolonged subsidence stratigraphy that would be suitable for recording relative sea-level changes.
The ria-type (fluvial-formed), tide-dominated Tamarindo Estuary consists of a wide, sandy channel that is hypersaline during the dry season and isolated from the coast by a narrow, flat-topped peninsula that rises ∼2.7 m above mean sea level (AMSL) (Figure 2a). Elevations reported here are based on surveying with level, rod, and laser range finder relative to mean sea level, defined as the midpoint between zero tide and mean high tides at Puntarenas, Costa Rica, the nearest tidal station. This peninsula is a beveled beach ridge, or chenier ridge, and is composed of semi-indurated, well-sorted, coarse, shell-rich sand (Figure 3). The estuary gives way to a broad, ecologically zoned mudflat wetland with a well-developed mangrove ecosystem (Figure 4). The mangroves grade from bank-forming (prop-root) Rhizophora trees (red mangrove) that surround the channel (Zone A; Figures 2b–c and 4) into brushier salt-resistant Avicennia (black mangrove) that extend into the broad mudflats beyond (Zones B and C; Figures 2b and 4). Clear mudflat patches, or salt flats, occur at supratidal to mean–high tide levels beyond the Avicennia zone (Zone D; Figures 2b, 2d, and 4). Zone D is hypersaline in the dry season (October through March) but can become flooded by fresh or brackish water during heavy rains of the wet season. The estuary itself varies in character. Along path BB′ (Figure 2a), the estuary consists of a broad, shallow, sandy-gravelly-shelly bottom with average local marine salinity (∼32 parts per thousand [ppt]) (Figure 2e). Salinity gradually increases upstream in the channel due to evaporation, resulting in hypersaline conditions (>35 ppt) during the dry season. The channel abruptly narrows and transitions to a deeper, muddy bottom that is highly reduced and organic rich at the location shown in Figures 2a and 2e.
The transitions from subtidal reduced mud in the estuary channel, to lower-tidal Rhizophora peat, to upper-tidal Avicennia peat and mud, to oxidized mud of the upper salt flat (Figure 4) are consistent with the expected zonation of mangroves as lying between mean sea level and highest astronomical tide (Belperio, Harvey, and Bourman, 2002; Ellison, 2008; Grand Pre et al., 2012; Ramcharan and McAndrews, 2006; Rowe, 2007; Saintilan and Hashimoto, 1999). Although a detailed topographic survey along the profile in Figure 4 has not been completed, observations show that the supratidal salt flat (Zone D) is flooded by tidal sheetflow only at spring tide conditions. A leveling line survey from Zone D to the outer beach via roads did imply a lower elevation than highest tide. This suggests the possibility that tidal currents are slowed by the narrow estuary channel and drag in the mangrove, and they do not reach as high elevation in the tidal flats as along the open shoreface. It is also possible that Zone D elevation is slightly lower than the more seaward zones (e.g., Zones 2 and 3), either due to subsidence associated with shrinking during the dry season or due to formation of retarding levies via either sedimentation or anthropogenic activity related to prehistoric saltworks in the estuary.
The Tamarindo Estuary contains geomorphic evidence for Holocene sea-level change. A stand of dead and dying nonmangrove trees at the northern edge of the estuary suggests recent (but pre-2012) relative sea-level rise, possibly associated with post-1950 interseismic subsidence. The lagoon barrier implies earlier relative sea-level fall. The 2.7 m AMSL beveled beach ridge may have been deposited during a mid- to late-Holocene highstand. Although regional records suggest that sea level stabilized near present levels by about 5000 years BP (Caballero et al., 2005; Curray, Emmel, and Crampton, 1969; Fleming et al., 1998; Sirkin, 1985), sea level just prior to this may have been a highstand of several meters above modern levels. This is suggested by global records of sea level in the tropics and models of sea-level change that take into account ocean siphoning near the equator due to glacio-isostatic adjustment (Milne and Mitrovica, 2008; Mitrovica and Milne, 2002). One radiocarbon age obtained from a shell in this sand of 3850 ±130 years BP (calibrated age; see Table 1, Figure 3) does imply that the beach ridge postdates the glacio-isostatic highstand. However, storm overwash or tectonic uplift may have also contributed to the height of this beach ridge.
The upper Tamarindo mudflat was cored in 2010 and 2011, using both gouge (2.5 cm diameter) and Russian-type (6 cm diameter) borers, to define recent depositional history and test for cyclic relative sea-level changes that may relate to earthquake history. This study hypothesized that sea-level change would be recorded by lateral migration of tidal mangrove zones. Specifically, coseismic uplift might result in emergence of the reduced subtidal muds of the estuary channel, over which mangrove peats might form. Coseismic uplift in temperate estuaries has been documented by freshwater peats developed over tidal mud with very sharp contacts (Atwater et al., 2004; Sawai et al., 2004). Similarly, oxidized salt flats and paleosols might form atop mangrove peats that were uplifted too far out of the intertidal zone. The primary goals for this coring effort were as follows: (1) assess whether coherent stratigraphy existed in this heavily bioturbated environment; (2) quantify depositional history and assess whether the age of deposits would overlap with the late-Holocene earthquake history; (3) test whether alternating facies are present that could indicate a history of relative sea-level changes; and (4) document lithostratigraphy and qualitatively interpret facies changes (recognizing that quantitative relative sea-level history would require future biostratigraphic work).
The second site, Playa Carrillo, is located along the central Nicoya coast (Figure 1b), above the locked zone of the megathrust (Feng et al., 2012) and centered in the zone of greatest coseismic uplift (0.7 m) during the 2012 earthquake (Protti et al., 2014). The site consists of a small wetland in a valley that may have once hosted the Ora River (Morrish et al., 2009). Topography suggests that the Ora River was captured by a coastal tributary eroding headwardly from the SW. Although the current drainage area of this wetland is small, the local depositional setting seemed promising for recording the recent relative sea-level change. The wetland surface was 1.2 m AMSL when cored in 2011 (prior to coseismic uplift) and thus prone to shallow flooding during monthly spring tides, and it was separated from the foreshore by an aggrading berm sequence (beach ridge) lying ∼2.5 m AMSL. One reconnaissance core was obtained from this site.
All cores from these sites were described in the field. Age control was established using 15 accelerator mass spectrometry (AMS) radiocarbon ages using both Beta Analytic and the Center for Applied Isotope Studies at the University of Georgia (UGA) (Table 1). Ages are reported as calibrated ages relative to 1950 and were calibrated using IntCal04 (for Beta Analytic ages; Reimer et al., 2004) and IntCal09.14c (for UGA ages; Reimer, et al., 2009). Radiocarbon ages were obtained on plant/wood fragments or bulk organic material (Table 1). The wood fragments typically consisted of reddish-brown pieces up to several centimeters in diameter that appeared detrital and lacked surface features typical of roots. Although it is possible that large wood fragments can be transported large distances and thus have long residence time prior to burial, this study interprets these wood fragments to have been derived locally from dead mangrove limbs that were transported only a short distance prior to deposition in the mangrove swamp. Therefore, one can expect the residence times at the surface prior to burial to have been minimal relative to the burial ages. This interpretation is speculative, however, given that the possibility of unique identification of the fragments as sourced from proximal mangrove trees does not exist and that there is no corroborating taphonomic evidence for how long they could survive prior to burial. Given that ages from wood fragments overlap with bulk sediment ages, and given that the burial ages are mostly old, these assumptions seem reasonable for our reconnaissance efforts.
This study focused investigations at Tamarindo at the northern tip of one peripheral estuarine channel, where an easily accessed transition in mangrove zones occurs over several hundred meters (Figure 2b). Fourteen cores were collected, ranging from 2–4 m depth, from the different mangrove eco-zones. Logs of representative cores from each zone are shown in Figure 5. The recovered stratigraphy was variable and generally could not be correlated between widely (>10 m) spaced cores. Repeat cores at individual locations ∼1 m apart, however, did reveal coherency. Such lateral variation is not uncommon for mangrove wetlands, where lateral variations in compaction, depositional environment, and preexisting topography are typical (Belperio, Harvey, and Bourman, 2002; Berkeley et al., 2009; Dura et al., 2011). Primary depositional features were also poorly preserved in cores, likely due to bioturbation and slow burial (see below).
The open tidal flat (Zone D) exhibits the most variable and laterally discontinuous stratigraphy (Figure 5). The upper meter consists of interbedded mud and sandy mud, which could have been deposited in an evolving salt flat above mean tide level. These horizons are heavily bioturbated, partially oxidized, gradational, and complex. Below these layers, a thicker layer of hard, greenish-gray, silty clay occurs, which was interpreted to be a deposit layer in a submerged tidal flat. This transition at ∼1 m depth may indicate minor emergence (relative sea-level fall) in the more recent depositional history.
Stratigraphy in the vegetated Avicennia and Rhizophora mangrove eco-zones (Zones A, B, and C) is more continuous, generally consisting of gray and brown silty clay in ∼25 cm thick, bioturbated layers overlying a prominent transition to soft, silty peat at ∼1–1.4 m depth (Figure 5). The dark peat contains large, red wood fragments and other herbaceous debris with an estimated organic content of more than 20%, which we interpret to be derived from prop-root type mangrove trees of the Rhizophora eco-zone. This is consistent with observations and descriptions in other similar environments (Grand Pre et al., 2012). Prop-root type mangroves generally form peats when above mean high-tide level, suggesting a history of relative sea-level rise over the span of the cores (Belperio, Harvey, and Bourman, 2002; Dura et al., 2011; Ellison, 2008; Engelhart et al., 2007; Grand Pre et al., 2012; Ramcharan and McAndrews, 2006; Rowe, 2007; Saintilan and Hashimoto, 1999). Interbedded with the peat are 10–15 cm thick horizons of featureless gray-green mud, which could not be correlated between cores. The alternation of peat and mud is more pronounced below 1.5 m depth and continues to the maximum core depth of 4 m. The lack of extensive organic debris and the reduced nature of these muds is consistent with deposition in a consistently submerged environment well below mean tide level, such was observed in the modern environment in the base of the estuary channel. The alternations between mud (subtidal) and peat (higher than mean tide) may indicate cyclic relative sea-level change during the longer history of submergence recorded below 1 m depth. In contrast, the lack of gray mud and peat in the upper meter of cores implies deposition above the prop-root eco-zone in the marginal mangrove system (e.g., Zone D). This interpretation is consistent with the degree of partial oxidation and complexity resulting from bioturbation in the upper meter, which are suggestive of periodic subaerial exposure.
Calibrated AMS radiocarbon ages (relative to 1950) are shown on cores and plotted in the lower right of Figure 5. Errors on ages (not shown) are generally ±100 years for the main calibrated peak (specific ranges provided in Table 1). Ages range from 675–7290 years BP. The antiquity of the upper meter of sediment is corroborated by a pottery shard from 0.7 m depth in Zone D (Core 1), which displays markings typical of pottery, tools, and artworks common to the middle first millennium (AD) pre-Columbian Costa Rican culture (Fernandez and Alvarado, 2006). Ages correlate roughly with depth, implying an average depositional rate of ∼0.4 mm/y, although there are individual ages in Cores 2 and 4 that fall out of depositional sequence (Figure 5; Table 1). This may indicate that inheritance is a problem in some ages obtained from wood fragments. Given the overall trend and the overlap between ages measured on wood fragments and bulk sediment, however, the age control for these cores should provide at least an approximation of overall Holocene depositional history of the estuary. Nonetheless, the weakness of the trend and the apparent age variation between cores suggests complexity that would require additional dating to fully characterize. For example, given the uncertainty in the age-depth trend, it is possible that deposition slowed toward the present following a more rapid interval of early to mid-Holocene accretion (Figure 5).
The reconnaissance core and radiocarbon ages from Playa Carrillo indicate a different depositional history than observed at Tamarindo (Figure 6). Although Carrillo currently has a much smaller drainage area and should thus have lower sedimentation rate, the basal date of 470 years BP suggests a depositional rate of 3.7 mm/y, seven times faster than at Tamarindo. The stratigraphy of this site is also different, consisting mainly of brown sand in the upper 1.2 m, with a mixture of gray sand and muddy sand mixed with peat below. The sands contain abundant shell fragments, including well-preserved Turritella gastropods (sea-snail shells) that are common in the swash zone of the foreshore beach environment. The sand horizons were generally not thin or well enough sorted to indicate deposition by tsunami. These observations imply a transition from a generally submerged intertidal mangrove environment to sand produced by rapid infilling by overwash from the beach across the storm berm (e.g., Wang and Horwitz, 2007) or sand in-wash transported from the estuary during highest tide (Figure 6).
The age of the deposits at the Tamarindo Estuary likely preclude this site from being useful for recording events over the past few thousand years. The average depositional rate of 0.4 mm/y implies that the transition in facies associated with frequent megathrust ruptures (as known from the historical record) may occur over just a few centimeters of stratigraphy, which could be too narrow to be preserved given the extent of bioturbation. The signal-to-noise ratio may thus be too low to constrain ground-level fluctuations associated with the most recent (including historic) interplate earthquakes.
The earlier history preserved in this estuary appears to record regional relative sea-level changes. The occurrence of peats below 1 m depth suggests relative sea-level rise on the order of 2–3 m between 7300 and 4500 years BP. Regional sea-level rise during this same period was likely on the order of several meters, followed by sea-level stabilization by ∼5000 years BP (Caballero et al., 2005; Curray, Emmel, and Crampton, 1969; Fleming et al., 1998; Sirkin, 1985) or even a highstand at about the same time (Milne and Mitrovica, 2008; Mitrovica and Milne, 2002), suggesting that the submergence of these soil horizons may have been due to regional sea-level change and that the coast during this period did not experience appreciable tectonic uplift or subsidence. A similar pattern has been observed in numerous other tropical mangrove environments, where coring has revealed early to mid-Holocene in-filling associated with regional sea-level rise (i.e. a transgressive lithologic sequence) followed by slower deposition since 5–7 ka (Belperio, Harvey, and Bourman, 2002; Dura et al., 2011; Grand Pre et al., 2012; Rowe, 2007). The transition in the cores from reduced mud to oxidized, bioturbated mud in the salt flat (Zone D) may in turn represent relative sea-level drop following a mid- to late-Holocene highstand, with transition from a tidal to a supratidal environment. The record at Tamarindo thus implies no major permanent tectonic uplift during the middle to late Holocene, because it would need to have been balanced by a comparable magnitude of deposition-related subsidence that would not be expected from such a small basin.
The cyclic relative sea-level changes indicated by the peat-mud alternations below 1 m depth could, however, reflect moderate uplift/subsidence events related to paleoearthquakes and the megathrust seismic cycle that were imposed on the transgressive sequence. Because these alternations could be formed by other dynamic factors in the mangrove depositional environment, such as channel migration, it cannot yet be concluded that they represent paleoearthquakes. Additional work is necessary, including more-closely spaced cores to rule out lateral depositional changes, higher resolution dating, and micropaleontological biostratigraphic analysis (e.g., Hemphill-Haley, 1995; Horton et al., 2007; Dura et al., 2011; Sawai, 2001; Sawai, Horton, and Nagumo, 2004; Sawai et al, 2004; Shennan et al., 2014; Sherrod, 1999). For example, Grand Pre et al. (2012) were able to demonstrate that a peat-mud transition represented a solitary early-Holocene coseismic subsidence event in a similar mangrove environment along the Sunda megathrust in Indonesia. A potential problem at this study's site is that the possible recurrence time based on historical observations (∼50 y) is so short that the marsh depositional environment may not have time to reset before the seismic cycle is repeated; determining whether it is requires further testing. Another limitation at Tamarindo is that there are no obvious sedimentary indicators of tsunami inundation, whereas tsunamis are key aspects of paleoseismic records at other sites of similar depositional environment (e.g., Ramirez-Herrera et al., 2014).
Due to its younger depositional history, Playa Carrillo is better suited for recording more recent relative sea-level changes, possibly connecting stratigraphy to the historical earthquake record. It was interpreted that the upper 1.2 m of sand at Carrillo was deposited recently (in the past 100 y?) via overwash from the beach zone (e.g., Wang and Horwitz, 2007) or via the estuary during storms. Additional ages (e.g., from Layer 1; Figure 6) would be required to better constrain the timing of this deposition. Infilling and progradation of sand may have been facilitated by interseismic subsidence, as the beach ridge gradually lowered relative to the high-water levels of storms and highest astronomical tides. Beneath the sand, the bioturbated, gray, muddy sand horizon (Layer 1) may be a buried soil horizon. Although not a peat, the wood fragments may indicate deposition in the high mangrove zone (i.e. above mean high tide), whereas this horizon was at mean sea level prior to uplift in the 2012 earthquake. It is possible that this layer was deposited after emergence associated with the penultimate event (i.e. 1950). The peat-mud alternations below this could in turn represent earlier uplift events followed by gradual subsidence and burial. If true, some previous events may not be preserved, given that the historical record documents three events in the past 200 years prior to 2012 (Protti et al., 2014), whereas the deepest muddy peat in the core (Layer 3) was deposited ∼480 years BP. Better age control of these couplets is required to further test this hypothesis. It is also uncertain how far back this record may go, given that we did not attempt deeper cores at this site.
These investigations should be considered reconnaissance in nature, given the limited spatial distribution of cores, uncertainties in age constraints, and the lack of biostratigraphic analysis. Preliminary results suggest that viable paleoseismic sites may be present along the Nicoya segment of the Middle America Trench, but that they are likely rare and have complex and fragmented records. Due to slow deposition, Tamarindo does not preserve a record of the most recent events, although older peat-mud alternations may represent mid-Holocene megathrust earthquakes. In contrast, historical events may be recorded by younger sediment at Carrillo, but the record may be fragmentary (i.e. missing events) and it is unclear how far back it extends. The Carrillo site illustrates the potential of localized depositional settings for recording a paleoseismic signal, suggesting that the search for sites should go beyond obvious, large estuaries and wetlands. This is important, given the relative rarity of large wetlands along the peninsula, due to tectonic emergence, wave energy, and coastline and drainage basin geometry. One site not yet investigated which has potential given similarity to Carrillo is the Bongo River Estuary (Figure 1b). Likewise, one major wetland not yet explored that could have a potential paleoseismic record is the estuary of the Tempisque River in the Gulf of Nicoya, although this may be too far from the trench to experience significant earthquake-related relative sea-level change.
The complexity of deposits that were observed in this study shows the importance of including biostratigraphic analysis in future work. It is well known that biostratigraphic proxies have the potential to document higher-resolution records of relative sea-level changes. Vertical zonation of intertidal facies at the decimeter scale in both temperate and tropical environments can be deciphered using local assemblages of diatoms (Hemphill-Haley, 1995; Sawai, 2001; Sawai, Horton, and Nagumo, 2004; Sherrod, 1999), foraminifera (Belperio, Harvey, and Bourman, 2002; Berkeley et al., 2009; Horton et al., 2007), or pollen (Ellison, 2008; Engelhart et al., 2007; Rashid et al., 2013; Rowe et al., 2007). These paleoenvironmental indicators can be highly sensitive to coastal gradients in salinity and other variables, enabling construction of transfer functions that quantitatively relate proxy assemblages to depositional water elevation (Grand Pre et al., 2012; Horton et al., 2007). Application of these techniques to Nicoya sites where suitable stratigraphy is present, such as at Carrillo, may prove fruitful for constraining the paleoseismic history of this megathrust.
This study has identified potential paleoseismic sites along the Nicoya segment of the Middle America megathrust at Tamarindo and Playa Carrillo. Due to the nature of the Nicoya coastline, potential sites with suitable subsidence stratigraphy are rare. At Tamarindo, estuary sedimentation related to overall early-Holocene sea-level rise records lithostratigraphic changes that may relate to minor relative sea-level change associated with subduction zone earthquakes. Late-Holocene events do not appear to be preserved, however, given slow deposition after ∼5 ka. Younger stratigraphy is exhibited at the Playa Carrillo Estuary, including peat-mud alternations that may represent relative sea-level changes associated with subduction zone earthquakes over the past 500 years. More coring, environmental analysis, and chronostratigraphy are required to constrain these events, however. In addition, biostratigraphy is required at both sites to enable quantitative interpretation of relative sea-level history. Although these results suggest a paleoseismic record is preserved along the Nicoya coastline, this record appears fragmented and will require careful additional work at numerous sites to patch together a coherent history.
The following people and organizations are thanked for assistance with this work: Alyssa Durden, Craig Cunningham, Illiya Smithka, Shawn Morrish, Lewis Owen, Bibi Santidrian, Gabriel Blanco, Ken Lacovara, Marino Protti, Mario Boza, Las Baulas National Marine Park, and the Leatherback Trust. Brian Atwater and anonymous reviewers are thanked for their reviews of an earlier version of this manuscript. Funding for this research was provided by MARGINS/NSF: OCE0948290.