Globally, high-magnitude (> 8 Mw) earthquakes often occur along subduction zones, such as in Chile in 1960, Sumatra in 2004, and Japan in 2010. The Guerrero segment (GS), an approximately 200 km-long part of the Mexican subduction zone (MSZ), is known for low seismic activity and is often referred to as the Guerrero Seismic Gap (Fig. 1a, b). A proposed plausible rupture of the entire gap could generate an earthquake of Mw>8.41. An earthquake scenario of such magnitude would affect the Mexico City metropolitan area with more than 22 million people. A catastrophic tsunami accompanying this earthquake could cause significant damage to coastal communities such as Acapulco, among others. Recent studies suggest that slow slip events (SSEs) may prevent the occurrence of large earthquakes in the GS2,3. In turn, the rheology on this segment of the MSZ promotes slow slip over fast slip, with earthquake generation at the plate interface3. However, this hypothesis is based on an incomplete assessment of tsunamigenic earthquakes historically and in the late Holocene. Indeed, there are other subduction zones where large earthquakes and SSEs have occurred in the same or neighboring sectors of subduction zones, such as the Japan subduction zone4,5,6.
Here, we present geologic evidence that reveals a ca. 2000-year history of large tsunamigenic earthquakes and demonstrate that Mw>8 tsunamigenic events occurred in the past in the GS, with a long and variable recurrence (> 700 years). Our data suggest a likelihood for their future occurrence. Furthermore, we discuss possible relations between slow and rapid slip, revealing that SSEs of the MSZ do not yield sufficient strain release to mute large earthquakes or prevent catastrophic earthquakes and tsunamis from occurring in this segment and neighboring segments.
Coastal and offshore morphology
The Guerrero segment of the MSZ, between Acapulco and Petatlán (Figure 1a, b), has been defined as a morphotectonic zone8 characterized onshore by a series of coastal lagoons, sand bars, beach ridges, alluvial plains, and rocky promontories7. The morphology offshore shows a narrow shelf, 7–12 km in width8, and a trench extending ~70 km from the coastline. Several morphologic features characterize the rough offshore topography, e.g., popped-up mountains, seamounts, ridges, and escarpments, which are associated in other regions with shallow megathrust earthquakes and tsunamis9,10. The coast near Acapulco is wave-dominated and microtidal, ranging from 0.6 m to -0.3 m11.
We focused on a low coastal alluvial plain in search for evidence of historical and prehistorical tsunamis and their triggering earthquakes (Fig. 1c). Previous studies infer late- and mid-Holocene earthquakes and associated tsunamis7,12. Here, we describe sites with excavations at a shoreline-normal transect (Fig. 1c; Supplementary material Supplementary Fig. S1) with a focus on the A1 and A2 sites. The A1 site is located on a lowland, covered by grasses, at the edge of an ancient estuary partially fringed by mangroves, and ~2 m above mean sea level (amsl). Seaward, the landscape is represented by a series of beach ridges and swales, swamps, mangrove marshes, sand dunes up to 5 m high, and a steep beach with cups. Site A2 is located along the same transect as site A1 and in a similar geomorphic setting but with a slightly higher up slope.
Geologic evidence of large earthquakes and tsunamis
We discovered robust evidence for past large tsunamigenic earthquakes that provides a needed context to evaluate the potential for future equivalent events in this sector of the MSZ. The local geology and topography indicate that the study site is currently in an estuary at a distance of ~800 m from the shoreline and 2 m above sea level, providing a higher preservation potential of tsunami deposits.
Three sand units are revealed by stratigraphic logs for sites A1 and A2 that may be associated with past tsunami events. Sand unit 1 in log A2 is apparently an artifact produced by landscape denudation from recent human activities and yields a near modern 14C age of 1989–1992 cal AD (Fig. 2; Supplementary Table S1, S2, S3). Radiocarbon ages are inverted with depth and may reflect burrowing and bioturbation13. Sand unit 2 is observed at both the A1 and A2 sites (Fig. 2; Supplementary Fig. S2) and contains unique taxa of marine diatoms that are absent in units above and below (Fig. 3). These diatoms are not well preserved and are scarce, and laboratory procedures have been used to enhance diatom detection14. Furthermore, the elemental composition of this unit shows an increase in Na, Mg, Ca, Br and Ba, indicating a marine influence (Supplementary Fig. S3). The clay unit below also shows abundant diatoms indicative of a brackish environment, while the clay with sand unit above shows a significant decrease in brackish diatoms, from 60% to 10%, suggesting a land-level change, probably with coastal coseismic uplift (Fig. 3).
Theanisotropy of magnetic susceptibility (AMS) was used to characterize the sediment fabric, confirm the occurrence of a high-energy event, and reconstruct the flow characteristics of a suspected tsunami. AMS results show magnetic fabric (MF I) with a significant degree of anisotropy P and anomalous orientation of K1 (not perpendicular to the shoreline) in sand unit 2 (Fig. 4), suggesting deposition in a high-energy environment during a tsunami inundation event. River inundation is excluded by the presence of marine diatoms and geochemical salinity indicators. This interpretation is consistent with a significant increase in magnetic susceptibility at the bottom of unit 2, reflecting swash processes sorting for the heavier magnetic grains, with a linear decrease upward, as depositional energy slackens, as normal grading of the unit (Fig. 4). The different mechanical properties of sediments (particle size and shape) likely explain why K1 lacks reorientation perpendicular to the MF flow direction, despite exhibiting a higher degree of anisotropy. MF II from the clay unit below is typical of sedimentary MF of a low-energy depositional environment with mixed horizontal K1 and K2 components and a low degree of anisotropy P15. Radiocarbon ages from sand unit 2 at both A1 and A2 sites yielded similar ages of 1955–1956 cal AD and 1954–1955 cal AD, respectively. Instrumental records report three earthquakes and tsunamis in the late 1950s to early 1960s that may be related to this sand deposit. The M 7.8 earthquake produced a tsunami near Acapulco in 1957, although tide gauge data at the Acapulco port did not confirm a permanent uplift produced by this earthquake16. The 11 May 1962 (M 7.1) and 19 May 1962 (M 7.0) tsunamigenic earthquakes (Fig. 1) produced permanent uplift of 15±3 cm and 7±3 cm during these events in the Acapulco region16. Sand unit 2 is likely associated with the 1957 earthquake and tsunami.
Sand Unit 3 is observed on both A1 and A2 and shows a sharp basal contact and flame structures indicating syndepositional soft-sediment deformation. This unit also includes broken shells and fragmented and whole marine diatom valves (Fig. 3), suggesting marine inundation by a tsunami. Sediments above sand unit 3 are composed of clayish silt with predominantly brackish diatoms (> 40%), while sediments below are composed of fine sand with less fragmented diatom valves and few brackish diatoms (< 5%). This sudden change in the environment suggests land subsidence. Based on the analysis of diatoms in the current topography and coastal environments at the site (Figs. 5, S4), the drop in land level could be equivalent to a change from a beach ridge elevation to an estuary or a swale elevation, likely a subsidence in the range of ≥ 1 m. We suggest that a large earthquake, Mw>8, produced the observed coseismic coastal subsidence and triggered a tsunami, flooding at least 800 m inland. A slight reorientation tendency of K1 to the perpendicular position and a slightly higher degree of anisotropy P suggest that sand unit 3 was deposited in a higher-energy environment17,18,19,20. The age of this event through OSL dating of quartz grains indicates that this earthquake and tsunami occurred between 1240 and 1370 AD (Supplementary Table S3). The earliest historical documents written in Spanish record earthquakes that occurred in the 15th century21 and the oldest historical tsunamigenic event recorded on the GS date to 1537 AD22. Thus, historical records do not, but our results using different proxies demonstrate that a large tsunamigenic earthquake occurred on the Guerrero coast between 1240 and 1370 AD.
Sand unit 4 is composed mainly of fine sand and shows a sharp basal contact with a black sandy silt below. It contains shell fragments and brackish and marine diatoms, with some plant debris, suggesting a marine flooding event. The unit above contains fine sand and scarce but predominantly brackish diatoms (< 4%), more than in the sediments below sand unit 4 that show brackish to marine diatoms increasing downward in the slice, indicating a change in environment, probably a minor land-level change, i.e., minor uplift. The time frame of this event is not conclusive, older than 1240–1370 AD (Fig. 2).
Finally, the lowest deformed sand unit 5 with sharp basal contact, shell fragments, and marine diatoms (although diatoms are scarce) suggests probable marine inundation. Deformation of this unit indicates an apparent liquefaction event, most likely produced by an earthquake before 590–666 Cal AD and after 485–359 Cal BC (Fig. 2).
Coastal land-level changes and variable recurrence of large events
In summary, we provide evidence of four tsunamigenic earthquakes, although one of these events has a nonconclusive age (sand unit 4). These four earthquakes produced tsunamis, although only event 3, marked by sand unit 3, produced permanent and remarkable deformation, i.e., coastal subsidence. The other three events apparently produced a slight uplift or minor to nonpermanent deformation. The inland extent of tsunami flooding is a minimum estimate based on the surface distribution of associated deposits. However, given the high velocity and destructive pathways of a tsunami, the source of sediment and preservation of deposits, the 800 m inundation limit is a minimum estimate23,24,25,26. The time frame of these events shows that recurrence is highly variable for the past ca. 2000 years (Fig. 2). However, this record has unknown incompleteness and is a maximum estimate of average tsunamigenic earthquakes. Nevertheless, the “1300 AD” event (unit 3) is likely a remarkable event that produced considerable coastal deformation (≥ 1 m subsidence) and may reflect a rupture closer to the trench. Most instrumentally recorded events on the MSZ rupturing near the coast produced coastal uplift for earthquakes in 1962, 1985, and 202016,27,28. However, other large near-trench earthquake events (within 20 km) causing seismic slip in the shallow portion of the MSZ, such as Colima 1995 and Jalisco 1932, produced coseismic coastal subsidence29,31,32. Earthquake ruptures near the trench cause vertical coseismic displacement and uplift, generating tsunamis33. Near-trench events in the GS are relatively common, such as the 2002 Mw6.7 near-trench tsunami earthquake, which produced a limited tsunami response (< 10 cm)34. Tsunami modeling of a hypothesized near-trench Mw8.4 earthquake produces tsunami amplitudes over 9 m at the coast and inundation over 1 km inland in the study area. This model fits long-term evidence of a Mw>8 earthquake and a tsunami that flooded > 1 km inland (Fig. 6).
Does slow slip inhibit large earthquakes?
Geophysical studies in the GS are based on a limited 113-year time series and measurements of SSEs in the last ~20 years35,36,37 of instrumental seismic observations. SSEs are hypothesized to reflect trapped fluids, although there is limited evidence of fluids associated with the Guerrero segment and other fault zones. Furthermore, the concept that the rheology of this segment favors slow slip over fast slip2,3 remains unconstrained. Megathrust earthquakes have ruptured on the same or neighboring sectors of subduction zones where slow slip occurred4,5,6,38,39. For instance, two slow slip events in the Japan subduction zone preceded the 2011 Tohoku earthquake, indicating the coexistence of slow slip and earthquakes in the same portion of the subduction zone, such as the Mw9.1 Tohoku earthquake4. Other examples of SSE´s preceding large earthquakes were observed in Chile and Mexico in 2012 and 2014. In Chile, an SSE that lasted 2 weeks anticipated the Mw8.2 Iquique earthquake40. In Mexico, the SSE initiated four months prior to the 2012 Mw = 7.4 with the Ometepec earthquake moving toward the earthquake source region41. Additionally, a Mw7.3 earthquake was preceded by an SSE in the neighboring region of the earthquake source in Guerrero42. However, the phenomenon of SSEs and the occurrence of large earthquakes are still not well understood. Furthermore, geodetic observations in GS have a relatively short period to accurately determine the seismic coupling state and include limited observations of > 8 Mw earthquakes, providing uncertainty in these estimates43. Thus, the interplate locking state should be known to infer the occurrence of the potential of a large earthquake6. Submarine topographic features offshore Guerrero, such as seamounts and ridges8, might also promote an increase in interplate coupling and consequently tsunamigenic earthquakes10. Geologic and historical evidence of large earthquakes (Mw> 8), such as those presented here for the GS, provides records that span hundreds to thousands of years, providing minimum estimates on and evaluating large (> 8 Mw) earthquakes and tsunamigenic event potential44,45,46,47. Through combined long-term and instrumental observations and a deeper understanding of the relationship between SSEs and rapid slip (earthquake), earthquake and tsunami future occurrences might be better forewarned.
Long-term record of earthquakes vs. short instrumental geophysical observations
Our results show geologic and historical evidence of past large earthquakes coupled with tsunamis in the last 2000 years. A potential large tsunamigenic earthquake (M > 8) in ~1300 AD, i.e., is substantially larger than that recorded instrumentally and observed during the past ca. 110 years. The ages for the four tsunami-indicative sand units are younger than the mid-Holocene sea-level highstand, and sea-level records for the past 2 ka show little to mild changes in sea level, < 1 mm/year, until the late mid-19th century when sea level increased48. Previous work on this segment of the MSZ also provided evidence for apparent older earthquakes, probably 3 tsunami events during the past 4600 BP years7,12. Evidence of another large earthquake that produced land subsidence accompanied by marine inundation, most likely a tsunami, by c. 3400 yr BP, was reported on the Guerrero coast7. The neighboring Oaxaca segment shows historical and geologic evidence of a Mw8.6 earthquake and tsunami in the 1787 flooding > 500 km alongshore and up to > 6 km inland49.
Recent geologic evidence provides new insights into extraordinary catastrophic events in the Mexican subduction zone. Recurrence of large events might be in the range of centuries. Our observations also indicate that indeed subduction zones might have variable rupture modes. The GS and other segments on the MSZ and other subduction zones require combined long-term and instrumental observations to forewarn eventual catastrophes. Our findings offer evidence to prepare communities for earthquake and tsunami hazards.