Megaboulder Movement by Superstorms: A Geomorphological Approach

This contribution aims to add to the ongoing discussion on marine transport mechanisms (storm waves/tsunamis) of extraordinary large boulders (megaboulders, see definition below) in the coastal landscape. Its focus is on the only site (northern Eleuthera, Bahamas) where these enigmatic landscape markers are undisputable of pre-Holocene dislocation age (Eemian, ∼125,000 years ago) and discusses the occurrence of significant changes in environmental parameters such as position of a cliff or loss of boulder mass by weathering over this long time span. Both factors are of less importance for Holocene boulder deposits.

In recent discussions about future climate change, analogues of the past, such as the climate conditions during the warmer Eemian interglacial period, are studied to understand the hazard of superstorms under a warming climate. Different scenarios were proposed by modelling field data from the Eleuthera megaboulder site (based, among others, on investigations of Hearty and Tormey [2017], Rovere et al. [2017a,b], and Mylroie [2016, 2018]). The published results are in agreement to identify storm waves as geomorphological agents yet differ in the estimate of the magnitude of such an event. Hearty and Tormey (2017) suggest the occurrence of superstorms without an analogue in modern times; however, Rovere et al. (2017a,b) suggest that the deposits are associated with so-called normal hurricane activity during a sea level several meters higher than today. Both authors dismiss tsunamis as a causative placement mechanism for the megaboulders. The applied models quantitatively base the position of the boulders in relation to former sea levels, modern distance to the sea, and modern size and mass, including possible errors of measuring. They do not explicitly quantify (1) the rate of recession/erosion of the place of boulder origin (a cliff during a high sea level in the Eemian)—yet these changes over time significantly influence the transport distance of the boulders—and (2) weathering type and rates of the boulders themselves reducing their form, size, and mass.

For boulders of extraordinary size, the term megaboulders is used frequently in existing literature. This terminology was introduced as megaclasts (and mégablocs by Viret [2008] for the Eleuthera boulders) for boulders of tens of tons to more than 100 tons in places or was simply applied to the largest boulders found in their specific study area (e.g.,Dewey and Ryan, 2017; Lorang, 2011; Noormets, Crook, and Felton, 2004; Scheffers et al., 2009; Williams, 2010; Williams and Hall, 2004). Future studies should reserve the term coastal megaboulder to boulders exceeding >200 tons and those undisputedly dislocated by marine forces against gravitation to an inland position. This category is very rare and is thus far reported from only six regions along the coastlines of the world (Table 1). The megaboulders located in the Sunda Strait are related to the tsunami of the Krakatoa volcanic collapse in 1883 AD, and the deposits in SW Japan stem from the historical Meiwa Tsunami of 1771 AD, whereas the others remain under discussion regarding the type of dislocation event (storms or tsunamis).

Despite decades of research undertaken on coastal boulders, no catalogue of accepted methods for field and laboratory analyses has been established. An inventory of widely accepted methods should include the compilation of data, such as size, density, and mass of boulders, or their origin, transport distance, relative age indicators, and the related sea-level history.

Furthermore, understanding of boulder transport and deposition has been limited by a lack of a clear differentiation between transported coastal boulders and boulders transported against the direction of gravitational force. This is an important distinction to ascertain. Boulders that are transported in a coastal landscape can be associated with fundamentally different dislocation processes other than wave-induced processes such as a cliff collapse or detachment driven by physical weathering and subsequent movement under the influence of gravity (examples found in Erdmann, Kelletat, and Scheffers [2018]).

Additional factors are the rather small number of case studies in which near-time surveys after extreme events have been undertaken and that the majority of scientific publications concerned with coastal boulders focus on deposits from the Holocene period, which is, from a geologic viewpoint, a rather limited time span to study such events.

The debate on the origin and dislocation processes of coastal boulders in different locations is challenging because terms and definitions to describe coarse sediments are not used consistently in different studies. According to Blair and McPherson (1999), the boulder category starts at 25.6 cm in particle length, and the threshold for the megaclast category is an axe length >4.1 m. As shown in Erdmann, Kelletat, and Scheffers (2018, their Figure 1), a maximum axe length is a problematic criterion because it may not accurately reflect the mass of a boulder. Depending on a boulder's form, a columnar boulder may differ from a cube with the same maximum axe length by a factor of 400 in mass. The calculation of mass for a boulder (volume × bulk density) requires an accurate value for bulk density, which is difficult to measure and often is calculated—in the best case—with density of samples from the outer surface areas of the boulder; in most cases, these are altered by weathering. This problem is of less concern for freshly relocated boulders in near-time inspection surveys. In this article, boulder mass is reported in tons.

The term “superstorm” has been used in the discussion on the Eleuthera boulders by Hansen et al. (2016), Hearty and Tormey (2017, 2018), Mylroie (2016, 2018), and Rovere et al. (2017b), to suggest that during the warmer climate of the Eemian storms with energies significantly above the intensity of category 5 Holocene tropical cyclones did occur. Kennedy et al. (2017) and May et al. (2015) used the term supertyphoon for Cyclone Hayan (a category 5 hurricane) impacting the Philippines in 2013 with the worldwide strongest record of sustained winds reaching 315 km/h and gusts up to 375 km/h. Near-time surveys reported that the largest boulders moved during this event were estimated to weigh 180 tons (not against gravitation, compare with May et al. [2015]) along with two boulders weighing 208 tons (Kennedy et al., 2017), for which transport distance and lift against gravitation remains unknown. Beyond the previously mentioned articles, the term superstorm has been used in a variety of media for an especially powerful storm, yet it is critical to note that the term does not align with an exact meteorological definition, which is in contrast to the specific terms hurricane or cyclone (a category 5 event on the Saffir-Simpson scale that has been in practice since the early 1970s). Shepherd and Maue (2014) provided a convincing argument in the Washington Post as to why the term superstorm is meaningless and should be avoided altogether.

This study aims to contribute to previous research focused on the giant boulders found on the northern Eleuthera Island in the Bahamas. Hearty and Tormey (2017) attributed transport and deposition of these boulders to superstorms during the peak of the Eemian period when air temperature was 2°C warmer and sea level was at its highest, causing storms to be of strengths not seen during the Holocene. Rovere et al. (2017a,b) found, through numerical modelling of Holocene hurricanes, that these events may have transported megaboulders if sea level was at least ∼3.5 m higher. Both studies exclude tsunamis as a possible mechanism of boulder transport because the origin and energy involved in such events in this environment remains unknown.

Studies undertaken over the past three decades have been able to attribute boulder transport to specific events, and most of them found strong storm waves as the only or most probable process for boulder movement onshore (e.g., Cox et al., 2012, 2015, 2018; Erdmann, Kelletat, and Scheffers, 2018; Etienne and Paris, 2010; Fichaut and Suanez, 2011; Hall, Hansom, Jarvis, 2008; Hall, Hansom, Williams, 2010; Hall et al., 2006; Khan et al., 2010; Paris, Naylor, and Stephenson, 2011; Scheffers and Scheffers, 2006; Suanez, Fichaut, and Magne, 2009; Williams, 2010; Williams and Hall, 2004). In all cases, boulder mass was clearly below the megaboulder class, as defined here and documented in Table 1. Goto et al. (2010) discriminated between boulders transported by typhoons and those by tsunamis at the same sites on islands of the Okinawa group north of Taiwan. The study found a clear envelope of storm-dislocated clasts on a ∼350-m-wide seaward belt of a reef platform and larger boulders that had been moved by the Meiwa Tsunami in 1771 AD up to >1000 m in a landward direction.

Findings by Richmond et al. (2011) supported the idea that boulders of more than a few tons transported by storms cannot be found further than an average of 50 to 100 m inland. Terry, Lau, and Etienne (2013) found a clear difference between large (>40 m³) boulders, mostly unaffected by a category 4/5 tropical cyclone, and those moved by a storm of the same energy with a maximum boulder mass of less than five tons.

In contrast to Holocene coastal boulder deposits (emplaced approximately during the past 7000 years with a high sea level) with a limited time frame for modifying processes such as weathering, field documents from the Eemian (about 125,000 years ago) imply that the problem of reconstructing environmental parameters such as position of the coastline or boulder dimensions during dislocation. This article intends to fill these gaps by presenting reasonable estimates for an Eemian position of the palaeo-cliff, determining the transport distance of the megaboulders and the original mass of boulders by using state of the art quantitative values for limestone dissolution. Both parameters (transport distance and mass) have been significantly changed over the past 125,000 years by cliff erosion and weathering, so these aspects should be evaluated in the discussion on superstorms and integrated into future modelling studies.

The Bahama archipelago developed on a more than 1000 km NW–SE-extending carbonate platform during Pleistocene sea-level high stands by currents, waves, and eolian deposition of fine (oolithic) sands, mud, and shell debris and coarse particles from marine organisms including corals. The position and origin of several strata (by waves, storm waves, strong rain splash with fenestrae, hillwash) remain under debate (e.g., Hearty, 1997; Hearty and Tormey, 2017, 2018; Mylroie, 2016, 2018), but agreement exists on their general age as Younger Pleistocene interglacial deposits terminating in the Eemian sequence, the latter dated with U/Th.

Landscapes of the Bahama Islands (Eleuthera, Cat Island, Long Island, and others) are rather similar with low irregular hills (see, e.g.,Figure 1 with a maximum elevation of 24 m asl south of the Bull and Cow boulders) or long chevronm shapes with gently inclined slopes. The highest elevation of the Bahamian archipelago is 63 m above sea level on Long Island. Cliff sections facing the open and deep ocean to the east are long and often show a straight alignment (maximum height around 20 m, plunging in water depths of 5–10 m). A submerged abrasional platform extends to the east, inclining to the 20-m isobath 300–400 m from the cliff (compare also bathymetry data in supplemental material in Rovere et al. [2017b]). This platform comprises coral base rock and is covered with living coral reef, coral rubble, and sand flats. The sheltered leeward (western) shorelines along the megaboulder section of Eleuthera Island are characterized with beaches of low-angle profiles that show no significant sign of abrasion and submerge into shallow foreshores covered with carbonate sands and mud.

The coordinates for the megaboulder section 600 m to the south of Glass Window Bridge can be found in Table 2, including elevations and bathymetry to the east. Topography and geology are documented in detail by Hearty and Tormey (2017) and in particular in the supplement material provided by Rovere et al. (2017b). North of Glass Window Bridge, a long train (>4 km) of partly imbricated boulders with masses of tens to more than 100 tons at 12–15 m asl are scattered over the inland part of a rocky and mostly bare cliff top up to 100 m wide.

These boulders, mentioned by Hearty and Tormey (2017) as modern and often activated analogues for the Eemian megaboulders, indicate extreme wave events. Incorporated into these boulder deposits are large (>1 ton) fragments of Acropora palmata, which yielded radiocarbon ages of around 3000 BP (Kelletat, Scheffers, and Scheffers, 2004) as their mortality and deposition age. The surface morphology of the boulders shows sand abrasion at their foot and karstification of their upper surface. These weathering features point to long-term stability of the boulder position with very limited activation by hurricane events since the time of their emplacement, as described in Kelletat, Scheffers, and Scheffers (2004) who argue therefore to one single extreme dislocation event (a tsunami) with higher energy compared with local hurricane waves of the last 3000 years. The Bahamas hurricane history is recorded since 1851 (Hurricane Dorian, 2019; Neely, 2019), documenting a hurricane passing by every 2 years and landfalls at parts along the >1000-km-long island chain every 4 years, therein one of category 5 and seven of category 4 that have 1-minute sustained winds up to 247 km/h. Hitherto, Hurricane Dorian (September 2019) with 910 hPa central pressure, sustained winds of 185 mph (295 km/h), and gusts up to 355 km/h was the worst national disaster in history of the Bahamas. The system made landfall at Abaco Island about 110 km north of the Eleuthera megaboulder site.

The scientific dispute that surrounds the position of megaboulders in cliff-top locations on north Eleuthera, Bahamas (Hearty, 1997; Hearty and Tormey, 2017; Rovere et al., 2017a,b; Viret 2008) is related to (1) the process of dislocation and whether it can be attributed to hurricane-induced or tsunami-induced waves and (2) whether a superstorm in warmer geologic periods or hurricanes comparable to those on modern intensity scale would have been responsible for the boulder dislocation and transport.

Hearty and Tormey (2017) identified that the combination of megaboulders, chevrons, and run-up deposits together in the field present an undisputable conclusion that superstorms must have been present during the Eemian climate when global temperatures were up to 2°C higher. Thus, much higher storm intensities, compared to those during the Holocene climate, occurred. Mylroie (2016, 2018), however, disagreed with a number of these arguments, particularly on the genesis of chevrons and run-up deposits, and Rovere et al. (2017a,b) argued for normal hurricanes during an Eemian sea level at least 3.5 m higher compared to modern conditions.

There is a general agreement that the two largest boulders, identified as Bull and Cow (No. 1 and 2 in Figure 1, see also Table 2), were deposited in their current locations by marine forces during the Eemian, the last interglacial period (MIS 5e, about 128,000 to 116,000 years ago), with a sea level between 6 and 9 m higher than today (Hearty and Tormey, 2017). To reconstruct the transport path of these boulders, a quantitative estimation of conditions during the Eemian period is needed, in particular the origin of the boulders from a cliff, the position of this cliff in the Eemian, and the mass of the boulders when dislocated.

This article discusses three additional large boulders, identified as boulders 3, 4, and 5, with the latter also referred to as Twins because they represent two fragments that were thought to have been originally transported as one boulder and are now split at their original site, probably by weathering of the exposed parts and bioerosion along sea level (Figures 1 and 2, Table 2).

The boulders found here are some of the largest boulders located in a coastal location in the world and dislocated against the direction of gravitational force (Tables 1–4). Their size, volume, and mass have been estimated by Hearty (1997), measured onsite by Viret (2008), and detected by the Structure-from-Motion technique by Rovere et al. (2017a,b).

Boulder density was calculated using small fragments taken from the outer part of the boulders. Viret (2008) used thin cuttings of samples to determine a porosity of between 7–20%. Using a mean porosity value of 12% and the density of the most frequently occurring minerals in the boulders (calcite and aragonite with 2.7 g/cm³), the bulk boulder density is estimated to be around 2.4 g/cm³. Rovere et al. (2017b, Supplemental Material) took eight small samples from the outer parts of Cow and Bull (between 26.75–121.61 g) and used the Archimedean Principle (samples tested wet and dry) to calculate pore volume as a percentage. Results for the wet samples ranged from a maximum of 1.98–2.58 g/cm³ to a minimum of 1.73–2.46 g/cm³. Porosity was calculated as being between 8.8% to 23.1%, with a mean of 13.7% for the eight samples. Consequently, the mean density for the eight samples was 2.33 g/cm³, although Rovere et al. (2017b) used the mean density as only 2.06 g/cm³ (compare with Table 3).

Rovere et al. (2017b, Supplement Material) quantified two uncertainties in boulder volume measurement: one from the precision of measuring techniques and another from possible loss of size and mass after deposition. The length of boulder axes was measured using a three-dimensional image of each boulder, produced using Structure-from-Motion and one- and two-dimensional terrestrial images and additional images from a drone. Uncertainties using this method were calculated as 19 cm for the x axis, 11 cm for the y axis, and 44 cm for the z axis. When taking these uncertainty values into consideration, the volume and mass of both the largest boulders (Cow and Bull) could have been up to 8% larger than calculated by Rovere et al. (2017b). The Cow, therefore, may have a mass of at least 404 tons, whereas the Bull may weigh around 963 tons.

Rovere et al. (2017b) tested whether, with an increase in axe length of 20% potentially disappeared over time through weathering, these boulders could have been dislocated and transported to their modern locations by storm waves. Using this increase in axe length, an increase in volume and mass of 58% was found (compare with Table 4) using the cuboid form constructed by Rovere et al. (2017b, see figure 3E), the Bull's volume rising from 449 m³ to 775 m³ and its mass from 1.076 tons to 1.860 tons, the Cow's volume from 186 m³ to 321 m³ and its mass from 383 tons to 661 tons.

To overcome estimations for differences in boulder size between the time of dislodgement and today's values, rates of limestone dissolution in the terrestrial environment were applied. These rates have been derived by different methods, in different latitudes, and over different time spans from field and laboratory experiments (Table 4, including references). The second parameter of importance for the evaluation of the type and energy of the boulder transport process is transport distance. Cliff retreat rates over time (affecting transport distance) have been calculated for the two time periods of high sea level and associated marine erosion—Eemian and the Younger Holocene from worldwide data (Table 5)—and for slope retreat under terrestrial conditions during time periods of lower sea levels during the last ice age. The following discussion aims to add a perspective on Eemian coastal processes in the NE Caribbean based on comparisons and rates of cliff retreat and limestone dissolution under terrestrial conditions.

In geomorphology, the reconstruction of forming processes over different timescales is of utmost importance. In coastal geomorphology, the reconstruction of regional sea-level variations is key to establishing a site-specific process history as sea level in relation to land topography, foreshore depth, and exposure determine the energy of marine hydrodynamics. This is in particular important, if a time scale covers totally different forming processes like a phase of highest interglacial sea level, followed by a period of lowest glacial sea level conditions, and ending with another high sea level in the modern interglacial.

Any case study of the Eleuthera coastal megaboulders have to reconstruct their history spanning two time periods in a (potentially) coastal situation. (1) The time span of the Eemian high sea level, which according to Hearty and Tormey (2017a; Figure 4) lasted around 17,000 years from about 134,000 to 117,000 years BP. If the boulder dislocation event occurred at the beginning of Eemian high sea levels, this long time has to be taken into account for the terrestrial phase of weathering; however, if the onshore transport occurred during the termination of the Eemian, the time span may have been only 1000 to 2000 years of surface exposure and weathering. (2) The second phase in which the boulders rested in a coastal landscape started with the mid-Holocene sea level high stand, transforming the coast to its modern geologic setting. For most of their time since dislocation, the Eleuthera boulder deposits were exposed to terrestrial subaerial conditions and weathering without being impacted by marine hydrodynamic processes—a time period spanning almost 117,000 years. This history makes the Pleistocene Eleutheran boulder deposits unique in a worldwide comparison.

For the following discussion, terrestrial dissolution rates are calculated over a time period of 100,000 years to arrive at absolute minimum values for mass (Table 4).

If the timing of megaboulder dislocation and transport in northern Eleuthera has been identified, a number of questions remain to develop a full picture of the processes by which these boulders were affected. These outstanding issues include the original size of the boulders in the Eemian period and the extent of weathering and the subsequent effect on boulder size since that period; their origin with regard to the most probably position of the Eemian cliff and the inland transport distance of the boulders to their present places; and the type and process of transportation, depending on the two quantitative figures mentioned above. Based on these findings, the following points are discussed.

According to Hearty and Tormey (2017 a, see figure 4), sea level during MIS 5e between 134,000 and 126,000 years ago and again from around 125 to 120 years ago was up to 6 m above modern levels, with a short (1000 year) time span to a maximum of 9 m between 120,000 and 118,000 before steeply dropping to below the modern sea level. The sea-level height at which dislocation of the boulders took place (and whether they were all emplaced by the same event or with time gaps of a thousand years or more) is unknown but remains an important parameter for reconstructing boulder dislocation.

The coastal topography is likely to be similar to the modern-day topography, with undulating low hills of sand and eolianite and an accentuated surface roughness through karstification of the resistant limestone bedrock from former interglacials. From the current cliff topography and stratigraphy, it is thought that the cliffs in the Eemian (MIS 5e) were of similar heights as today.

The exact boulder origin is important for potential transport against gravity. The boulder lithology is oolithic limestone dating from MIS 9 or 11, placed on less consolidated younger Eemian Oolithic sands as eolianite (Hearty and Tormey, 2017a). There is no evidence (marine organisms attached, borings) to suggest that they were derived from the intertidal. The most probable origin of the boulders is the cliff face, which means a lift against gravitation over a cliff edge of 10 m.

The modern cliff profile and submerged platform are erosional features; therefore, it is likely that the platform was at a higher elevation during the Eemian at the base of the eastward facing cliff. The bathymetry of the submerged platform is relevant for modelling potential wave energy (Rovere et al., 2017b), which also depends on water depth during the Eemian. If the foreshore platform was higher than it is now but sea level was also higher during the Eemian, water conditions would not have been significantly deeper.

To effectively estimate the distance over which the megaboulders in question have been transported, the position of the cliff during the Eemian, when sea level was higher, needs to be known. The minimum rate of cliff recession from this previous position to its modern location can be estimated using published rates of weathering and cliff erosion for different environments and lithologies (Table 5).

Following the time of boulder transport, cliff recession would have occurred over a short time span of 1000–2000 years in the MIS 5e with marine influence if the megaboulders have been broken off toward the end of the Eemian high sea level or up to 15,000 years over most of the Eemian period if the megaboulders would have been broken off the cliff at the beginning of the Eemian high sea level. To keep to minimum values for a potential transport distance of the megaboulders, calculations use the shorter time span. Another time period where marine erosion impacted the cliff existed for about 7000 years during the Younger Holocene period of high sea level. From the end of MIS 5e to the end of the first part of the Holocene (more than 110,000 years), any denudation and erosion at the old cliff took place under terrestrial conditions. To estimate the implication of these potential periods of erosion to the Eleuthera cliffs, several facts have to be considered.

The northern Eleuthera coastal lithology is defined by limestone, comprises oolites and eolianite from reef sand, and has been exposed to a minimum of 8000 years of bioerosion at the tidal margin during the Eemian and Holocene periods (1000–2000 in the end of Eemian, 6000–7000 in the Holocene), which may have contributed to cliff retreat. However, mechanical wave erosion is considered to be the main erosional process, which occurs at a mean global minimum rate of 0.5–5 cm/y (compare with Table 5) and is attributable to high energy waves and some salt weathering at the cliff faces in the Bahamas.

The rocks have a low resistance to weathering compared to old meta-sedimentary or crystalline lithologies attributable to poor cementing and a sedimentary texture.

The frequency of strong wave events is relatively high with at least 5–10 per century (i.e. around 500–1000 events over the study period), and, because of the latitudinal location of the study site, some of the events can be assumed to be hurricanes (Neely, 2019).

The highest minimum rates of long-term cliff retreat in limestones are found in England at around 1–5 cm/y and approximately 0.6–2 cm/y based on archaeological remains at Irish limestone coastlines (Table 5). A mean retreat of 2 cm/y along a marine cliff is a conservative value, which produces an estimate of 140 m of erosion over 7000 years, which is a viable distance, as it is less than half the width of the foreshore platform.

Since the Eemian period, estimations for the retreat of steep slopes (in this case a steep cliff, exposed to terrestrial forming conditions during low sea levels of the last glacial period) are rare and studies to compare to the Bahama environment lacking. The rock formations of Eleuthera are of relatively low resistance and relief energy and discharge areas influencing hill wash erosion are almost absent. Assuming erosion caused the cliff to change from a vertical 20 m cliff at the peak of Eemian sea level to a 20° slope (i.e. the top retreats but the toe remains stable), the cliff top is estimated to have receded by around 52 m, which would be the minimum erosional distance east since first contact between the cliff and the sea during peak sea-level height in the Holocene. Consequently, it is estimated that since the time of megaboulder deposition on Eleuthera, the associated cliff edge has retreated by at least 200 m. This results in a transport distance for Bull and Cow in the Eemian of more than 200 m and a transport distance of more than 400 m for the Twins (boulder 5) (compare also Figure 1B).

As the dislocated boulders are cemented to the ground surface where they are resting at least since Eemian times (Hearty and Tormey, 2017; Mylroie, 2016, 2018), any interim movement can be excluded. For the Twins, potentially a movement prior to the emplacement of Bull and Cow is possible but is not possible later because in this case the Twins had to pass the position of Bull and Cow. The Twins are resting on muddy sand, and if they have not been emplaced as the only boulders by a so far unknown event over a maximum distance, they most probably belong to the group of the other megaboulders. They have been emplaced in water depth of at least 6–9 m during the Eemian (see Hearty and Tormey [2017]), which is too shallow to allow storm waves of enough height and energy to move or activate them. The distance between Bull and Cow to the Twins (Figure 1) is 240 m, with a difference in elevation of their basement of about 18 m, which provides a medium slope inclination of 7.5% or about 3.5°. This slope angle is not sufficient to activate terrestrial slope process and movement of the irregular megaboulders. A reactivation during the Holocene can also be excluded, as the strongest Holocene hurricanes did not transport boulders of any size across the island into the leeward (western) waters of Eleuthera.

The Bull, Cow, and Twins (Boulder 5) boulders are known to have a pure oolithic limestone lithology and have limited cementation, as the age of sedimentation is most probably MIS 9 or 11. In particular, the shape of the Cow, with its mushroomlike profile looking from the west (Figure 2C), suggests that some erosion has taken place in the lower part since deposition, which is thought to be attributable to washover from hurricane waves and bores during times of higher sea level since the Mid-Holocene. Estimates of mass lost by erosion considers only mass loss through dissolution by rainwater in a terrestrial environment, which is the dominant process of weathering for all the upper exposed surfaces of the limestone megaboulders (e.g.,Figure 2A,D).

Dissolution rates of coastal limestone boulders have been studied by overprinting of bioerosive rock pools from the littoral (with secondary pool development in the original ones) over the last 20 years by Kelletat and Schellmann (2002) and Whelan and Kelletat (2002) for Cyprus; by Scheffers, Scheffers, and Kelletat (2005) from the Caribbean Islands of Grenada, St. Lucia, and Guadeloupe; by Kelletat et al. (2005) from Mallorca (Spain); by Scheffers and Kelletat (2006) from Barbados, St. Martin, and Anguilla; by Scheffers et al. (2008) from the Peloponnesus (Greece); by Mhammdi et al. (2008) from Morocco; by Scheffers et al. (2010, 2009); and by Erdmann et al. (2015), Erdmann, Kelletat, and Kuckuck (2017), and Erdmann, Scheffers, and Kelletat (2018) with documents from Ireland.

Additionally, some salt weathering, mechanical erosion by torrential rain during hurricanes, and washover by hurricane splash may have played a role, but they are not considered further here. In the calculation of mass loss by weathering only the upper surface of the boulders is considered, which is identified as the area of the boulder seen from above (e.g.,Rovere et al., 2017b, Supplemental Material, figure S5A). The upper surface areas of the boulders are 51 m² for Cow, 69 m² for the Twins, and 114 m² for Bull, which amounts to approximately 25–30% of their total surface.

The rate of dissolution in the terrestrial environment for pure limestone has been measured and subsequently calculated by several authors using different methods, including the use of microerosion meter, limestone tablets, elevation of karst tables under glacial erratics of known age of deposition, and the cosmogenic nuclide chlorine 36 (compare, among others, Furlani et al. [2009], Godard et al. (2015), Goldie (2005), Matsushi et al. (2010), Smith et al. (1995), Stephenson and Finlayson (2009), and Taborozi and Kazmer (2013); compare also Table 4). The mean rate over longer time scales is estimated to be between 0.02 to 0.03 mm/y, around 30–50 times lower than rates of bioerosion of the same type of limestone in a littoral environment (Erdmann, Scheffers, and Kelletat, 2018; Erdmann, Kelletat, and Kuckuck, 2017; Kelletat, 1988; Spencer, 1988; Trudgill, 1987; Trudgill and Crabtree, 1987; Trudgill et al., 1987). Additionally, Mylroie (2018) identified a tropical denudation rate of limestone of the order of 1 to 5 m per 100 ka, which agrees with those calculated by other studies. However, the rates of dissolution times are calculated for a period of 100,000 only. These estimates therefore represent only the minimum time of exposure to rainwater precipitation (Table 4).

If assuming the minimum surface size affected by dissolution and the minimum time in which weathering (dissolution) could have taken place, it is highly likely that during the time of dislocation in the Eemian, the boulders were at least double the mass they are today, which has major implications for modelling transport of these boulders and associated event intensities.

Based on the previous arguments from the Eleuthera boulder settings and reference to modern analogies for quantifying geomorphological processes in similar conditions and rocks, a tentative reconstruction of the Eemian situation suggests the megaboulders in question were significantly larger (approximately double in mass) when they were dislocated and have been transported further (over 200 m to over 500 m wide). Therefore, transport of the Eemian megaboulders by storm waves or their bores (as concluded by Rovere et al. [2017a] for a sea level several meters higher than today is unlikely because events of this magnitude have not been recorded hitherto in any ocean basin. Therefore, the validity of postulating superstorms as being responsible for the emplacement of the megaboulders (Hearty and Tormey, 2017) should be reconsidered. Typhoons in the western part of the Pacific have been documented to be increasing in intensity, e.g., in Haiyan (2013; compare with May et al. [2015]) with sustained winds of greater than 300 km/h, resulting in approximately 50% more energy compared to strong Atlantic hurricanes. However, no boulder dislocation against the direction of gravitational force comparable to the setting in Eleuthera has been described so far.

With regard to the position of all the megaboulders near Eleuthera's Glass Window Bridge, besides Bull and Cow (which are close to the modern cliff edge), boulder 5 (the Twins) offers a specific challenge in explaining its position 76 m from modern Twin Beach in shallow leeside waters. Its existence is a very strong argument against the storm and superstorm theories, as the two boulder fragments (formerly one with a mass of around 1000 tons during the Eemian) has been transported over or through more than 100 m of water with a depth of 5–10 m, depending on the position of the MIS 5e sea level. Hitherto, it has not been reported that a storm wave or a storm wave bore has the capacity to transport a 1000-ton boulder across a body of shallow coastal water over a >100-m distance at the sheltered exposition of an island barrier. Therefore, the argument is posed that a tsunami as a possible transport mechanism should not be discarded in the attempt to explain the history of the megaboulders. Also, boulders 3 and 4, close to sea level, may have been transported through shorter distances in coastal waters during the Eemian.

Only five other sites are found globally where boulders of a similar size to those on Eleuthera have been transported onshore by marine forces: Tongatapu, Rangiroa, central West Australia, Sunda Strait, and SW Japan (Table 1). Four of these sites are situated in the leeward (western) exposition of the specific geographic locations, where cyclones approach from the east. Thus, these sites cannot be directly compared to the setting of the megaboulder assemblage of Eleuthera; however, the only other marine force able to move boulders of the size in question over long distances and against the direction of gravitation are tsunamis. Therefore, tsunamis should not be excluded as a transport or emplacement mechanisms of these megaboulders, only by the argument that no similar events have been documented in the last centuries or because tsunamis are not known to occur within a given region.

Hearty (1997) and Mylroie (2018) propose a local bank margin failure generating a return wave as one of three potential mechanisms. This may explain the local appearance of a few giant boulders, but for the emplacement of very large coastal boulders along more than 100 km of the coastlines of Long island, Cat Island, and Eleuthera in the Holocene (Kelletat, Scheffers, and Scheffers, 2004), tsunamis should not be excluded as a possible mechanism until further conclusive evidence is provided.

For future comparative studies, a record of the following categories for each coastal boulder study location is proposed.

1. Coastal boulder characteristics: form and dimension; rock type and density; mass, orientation; other geomorphological and/or sedimentological features (pattern of boulder deposits, e.g., isolated, clusters, fields, trains, ridges, imbrication, sorted or chaotic, relative age indicators)

2. Coastal boulder transport process: hydrodynamics (waves, swell, bores, tsunamis, infragravity waves, freak waves); spatial parameters of boulder movement (place of origin; transport distance, elevation (prior and post event); type of movement (sliding/shifting, overturning/rotating, saltation)

3. Coastal boulder deposition environment: coastal environmental parameters (tides, wave climate, registered wave heights); exposure of location; foreshore conditions (bathymetry, profile, sediments); coastal profile; storm and/or tsunami history; sea-level history; neotectonic history

To summarise, hitherto research to constrain the history of the Bahamian megaboulders is in agreement in so far that the deposits have been dislocated and transported by marine forces against the direction of gravitational force onshore. This contribution argues that it is of importance to further constrain the environmental history of the Eleuthera megaboulders and embed this understanding in wave models before unambiguously a (super)storm event can be attributed as the event type. Further research on coastal boulder deposits should aim to collect a consistent catalogue of boulder characteristics and environmental parameters that can better inform comparative studies and wave models and importantly, reconstruction of palaeolandscapes and that sea-level variations need to be under better constraint in future studies.