Potholes and Resident Boulders on a Contemporary Limestone Shore (Sal Island, Cabo Verde Archipelago, Northeast Atlantic Ocean) Free

Potholes typically are associated with fluvial settings that entail the erosion of deep holes in exposed river bedrock due to the grinding action of pebbles caught in eddies (Thomas and Goudie, 1985, p. 386). In an early systematic survey by Elston (1917), this type of fluviatile feature was designated as normal but was distinguished from giant potholes because of glacio-fluvial action involving a greater energy flux related to the extreme volume of meltwater runoff. The occurrence of marine potholes earned a brief mention in the same study, citing concretions eroded from coastal sandstone entrained in tidal pools (Elston, 1917). The boring action of a grinder against bedrock was subsequently tested in an elaborate laboratory setting that induced stream flow directed through a glass tube against a concrete base that modeled vortex patterns (Alexander, 1932). The type of fluid mechanics performed exclusively on river potholes continues to attract investigators (Ji, Li, and Zeng, 2018; Ji et al., 2019).

Fluviatile potholes were discussed in a review paper by Wentworth (1944, p. 118), which also considered the equivalent development of potholes in coastal settings “where bedrock is exposed in the zone of wave action or at points where returning spray and surf are the agents of erosion.” The same article shows a shallow marine pothole eroded in reef limestone on the Hawaiian island of Oahu (Wentworth, 1944). The island proved to be a rich environment for further study by Abbott and Pottratz (1969), wherein 140 marine potholes were described from coastal bedrock, including calcareous beach rock, tuff, and basalt found mostly along Oahu’s S and SW shores.

Following the description of marine potholes on Oahu, a notable increase in documentation of marine potholes occurred based on examples elsewhere around the world (Abbott and Pottratz, 1969; Wentworth, 1944). These represent widespread localities in Japan (Sunamura, 1978), South Africa (Green et al., 2023; Miller and Mason, 1994), the Caribbean Islands (Blanchon and Jones, 1995; Sielski et al., 2017), China (Wang et al., 2013), the Canary Islands (Galindo et al., 2019), Spain and Italy (Pappalardo, Blanco Chao, and Pizzo, 2022), and Mexico’s Gulf of California region (Johnson and Callahan, 2023). The coastal shores studied in such places include a wide range of igneous rocks, such as granite, diorite, basalt, and andesite, but also sedimentary rocks such as shale, sandstone, limestone, and especially beach rock.

The effectiveness of the grinder as a rock tool against coastal bedrock is controlled by differences in rock hardness; however, basalt cobbles are known to be effective grinders within potholes in basaltic bedrock (Abbott and Pottrartz, 1969), and limestone blocks likewise are found to be effective in potholes surrounded by the same kind of limestone (Johnson and Callahan, 2023). Similarly, the reported elevation of contemporaneous marine potholes above mean sea level ranges from the intertidal to as much as 12 m above mean sea level (Galindo et al., 2019; Johnson and Callahan, 2023). Clearly, the reported variation in pothole elevation is a function of tidal range, which may be greater or lesser in different regions, and also affected by the reach of storm waves.

This contribution reports on a single locality with multiple potholes in the limestone of a dissected marine terrace bored by resident limestone grinders on the north coast of Sal Island in the Cabo Verde Archipelago off the western coast of Africa. The study site is unique to the island, being located between two small headlands at the extreme northern end of the island where the foreshore rapidly drops off to the 50 m isobath with basalt as the prevailing rock type. This part of the island experiences near constant wave action induced by the NE trade winds that form a dynamic segment of the greater North Atlantic gyre but also seasonal swells that originate from the western Atlantic as well as rare tropical storms. The goals of this study are to (1) reconstruct the geomorphology of a former sea cave eroded in limestone bedrock, (2) evaluate the physics of wave action required to derive limestone blocks that serve as pothole grindstones, and (3) review the region’s marine hydrology to better understand what kinds of events best account for the physical evidence recorded by potholes.

The African tectonic plate extends outward beyond the confines of its core landmass to meet divergent plate boundaries at seafloor spreading zones in the Atlantic, Southern, and Indian Oceans. Volcanic islands in the NE Atlantic Ocean are associated with tectonic activity in the Azores, Canary, and Cabo Verde Archipelagos as well as the lone islands of Ascension and St. Helena (Figure 1a). The Cabo Verde Archipelago comprises more than a dozen principal islands and several islets dispersed in a semicircular pattern that occupies an anomaly called the Cabo Verde Rise above the 4000 m isobath (Figure 1b; McNutt, 1988; Ramalho et al., 2010). By tradition, the archipelago is divided into leeward and windward islands in the context of strong NE trade winds. The so-called leeward islands extend from Brava in the SW to Maio in the E, but the north-facing shores of those islands are still affected by the trade winds. The windward islands stretch nearly 300 km west to east from Santo Antão to Sal (Figure 1b). Sal is one of the smaller islands, having an area of 216 km² with a circumference of 83 km and elevations up to 406 m formed by Quaternary volcanos (Figure 1c).

The enclosing 50 m isobath approaches closest to the island’s north end, where strong surf meets the dominant basaltic shores due to impact of seasonal Atlantic swells arriving from the NW as well as from the NE generated by the steady trade winds (Figure 1c; Bernardino, Rusu, and Soares, 2017). Above the basalt, parts of the north shore include a limestone terrace with fossil corals dated roughly to a mid-Pleistocene age (Reeb et al., 2024). The NE trade winds that reach Sal from the Canary Islands commonly register between 5 and 6 on the Beaufort Scale, which is equivalent to between 10.5 to 48.3 m/s (Johnson, Ramalho, and Silva, 2020; Stuut et al., 2005). Surface swells from this source are reported to have a near-shore amplitude of 3.5 m. The mean tidal range around the island is on the order of 1 m, although the effective tidal range along the northern coast is accentuated by the almost constant wave surge caused by the trade winds.

The latitude and longitude of the study site were determined using a Garmin Montana 650 device. The drone employed to produce aerial images was a DJI Mavic 2 Pro. A standard meter tape was used to measure the dimensions of the most prominent potholes and resident limestone boulders. Extensive photographs were recorded to document the site’s physical layout. At 1.83 m, the body height of the tallest participant (G.M. Martins) was used as a scale from photographs to estimate cliff heights otherwise too steep to descend on foot.

Well-formed marine potholes are bored into the floor of an arcuate-shaped breach cut landward into the limestone sea cliff against a Pleistocene marine terrace on the north shore of Sal Island between Ponta Norte and Ponta (Figure 1, marked by a star: 16°50′39″ N, 22°56′06″ W). An aerial view captured by drone from an oblique angle provides the physical context for a large reentrant having the shape of a natural amphitheater on the north-facing sea cliff (Figure 2). The outer bench edge of the feature sits approximately 8 m above mean sea level, which also corresponds to a prominent change in lithology from basalt to overlying limestone (Figure 2, dashed line). Limestone exposed in the slope directly below the amphitheater is discolored to a grayish tone by marine biofilm in the splash zone that extends laterally for about 20 m from side to side. A 6 m step marks an abrupt rise from the floor of the amphitheater to the edge of the enclosing marine terrace. In total, the seaward edge of this Pleistocene terrace sits about 14 m above mean sea level. The floor of the natural amphitheater covers about 250 m². Figure 3 shows a topographic sketch map that summarizes the observed spatial relationships among potholes and fallen limestone blocks on the amphitheater floor.

Three basic variations in the progressive development of potholes preserved on the amphitheater floor are distinguished by the sketch map (Figure 3). From the perspective of mechanical development, the best example is represented by an oval-shaped pothole in the NE quadrat (Figure 3, pothole 1). This feature is 2.8 m in length and 1.6 m in width, with a length-to-width ratio at nearly 2:1 having a maximum depth of 1 m (Figure 4a). Seen from above, the boulder appears black against the contrasting beige tone of the limestone floor. On closer examination, it was found not to be basalt but rather limestone thoroughly darkened by marine biofilm (Figure 4b). The pothole is the prime example from the amphitheater floor with a resident grinder found to be subspherical in shape with a measured diameter of 0.70 m (Figure 4b). The limestone walls around the pothole exhibit fine layering coated dark gray by the biofilm but also etched white by inlaid salt (Figure 4b). The lower walls of the pothole are crusted by a broad band of salt. The resident limestone boulder sits in a salt brine.

Another distinctive pothole is located in the NW quadrat of the amphitheater (Figure 3, pothole 2). Overall, this is a larger feature, the outer perimeter of which is also oval shaped, occupying a meter-deep depression more than 5 m in length and 2.5 m in width. No stone grinders were found in the depression. Two separate brine pools are clearly visible at the bottom of the depression, each surrounded by a crusted salt rim (Figure 5). A view of the structure at ground level shows that the salt band is enclosed by a darkly discolored limestone rim that envelopes at least two more outlying pits at a shallower position (Figure 6a). On close examination of the salt halo, large grains of NaCl are visible and found to incorporate the shelly debris from barnacles (Figure 6b).

A third depression is located in the SW quadrat of the basin (Figure 3, pothole 3). This pit is large and somewhat deeper than the others (Figure 3, potholes 1 and 2). The depression is occupied by a large limestone boulder and two smaller limestone fragments that were captured as rock falls (Figure 7). Fractures in the perimeter wall above the pit exhibit the outline of an impending boulder waiting to be loosened and plucked by wave activity that has already enlarged cracks on three sides (Figure 7, dashed line). Small potholes between 5 and 35 cm in diameter are arrayed around the pit, many of which are defined by salt fillings (Figure 8).

Foremost, the natural amphitheater carved into the side of a north-facing sea cliff exhibits primary evidence of enlargement by wave erosion that exerts progressive hydraulic pressure against preexisting cracks and fissures in the limestone walls (Figure 7). Limestone is undermined beneath bedded caprock, and massive blocks tumble into the structure from above (Figures 2 and 3). Some of the blocks are so large that it may be surmised they represent the late stage of a former sea cave unroofed by ongoing landward erosion. It is fortuitous that smaller limestone blocks plucked from the amphitheater walls become trapped in preexisting depressions on the floor. When waves spill into the amphitheater, it can be expected to shift those smaller limestone blocks back and forth, causing abrasion of limestone against limestone that works to enlarge a pit.

Seawater episodically flooding the floor of the amphitheater is evident from the widespread filling of both large and small potholes by salt (Figures 5 and 8). Sea salt is generally known to crystalize when approximately 90% of the seawater in an enclosure under aerial exposure undergoes evaporation. The secession of various natural salts derived from sea water proceeding to NaCl has been known for a long time (Ochsenius, 1888). Some of the larger potholes at the study site were observed to retain a liquid brine.

Evidence for pothole enlargement on the amphitheater floor is best exhibited by the lone example that retains a rounded boulder (Figure 3, pothole 1). The boulder comprises much of the same limestone as the surrounding pothole, so erosion may be likened to a mortar-and-pestle style of grinding action. As both pestle and the mortar comprise similar material, both can be expected to gradually degrade over time through the generation of carbonate sand. As any given pothole is enlarged by frictional wear, the nominative grinder becomes smaller in size and effectiveness. The radius of the resident boulder (0.70 m) may be applied to the equation for the volume of a sphere to calculate its mass, where mass (m) equals volume (v) multiplied by rock density (d). The limestone boulder is subspherical in shape (Figure 4b), and a rough calculation yields a value >40 kg.

Secondary evidence of pothole development implies the merger of multiple potholes supported by the dimensions of the larger feature in the amphitheater’s NW quadrat (Figure 3, pothole 2). In particular, this structure features two distinct depressions, both of which retain separate brine pools around which salt halos have crystalized (Figure 5). Notably, smaller additional depressions at a slightly higher level occur on the periphery (Figure 6a). Although no rock grinders are retained within the structure, it is likely that each of the larger potholes were initiated independently and that the rim separating them was gradually worn down by progressive erosion in the same style of mortar-and-pestle attrition.

Application of this formula requires that the existing subspherical grinder resident in a pothole be restored to its larger, rough shape before dislodgement. Such a quasicubic shape would be defined by horizontal layering and the spacing of vertical joint patterns. Wave impact against a cliff face is expected to introduce some tensile stress, with seawater introducing hydraulic pressure along horizonal partings, but with a much greater compressive stress by head-on wave impact (Herterich, Cox, and Dias, 2018). Hydraulic plucking of individual blocks in the surf zone can be expected to increase with the rising energy of wave shock against a sea cliff. A crude cube of limestone measuring 0.70 m on each of three planes will have a volume of 343,000 cm³. Assuming an average density of 2.5 g/cm³, the block could be expected to have a mass of 80 kg, which is twice the amount of the resulting boulder before its corners and rough edges are trimmed by subsequent abrasion in turbulent water. By this formula, removal of a limestone block from a cliff subject to wave shock would necessitate a wave with a breaking height of 4.75 m. Subsequent movement of that particular block from place to place on the amphitheater floor might easily require waves with less carrying energy.

Various factors control the size of waves as they approach land, among them fetch, wind speed, the nature of the sea floor close to shore, and the potential funneling effect of an embayment. Cliff retreat and the formation of coastal boulder deposits on the Aran Islands and around Galway Bay in western Ireland are known to be affected by winter storms in the North Atlantic that generate waves as high as 15 m at a water depth of 50 m as they approach the Irish coast (Erdmann, Scheffers, and Kelletat, 2018). Extensive ridges are formed by boulders equal to or far exceeding the 50 kg mass due to wash-over events at high cliff faces commonly 30 m in height. The displacement and landward transport of much larger limestone blocks in this region have raised the question of tsunami-generated waves in earlier Holocene times (Scheffers et al., 2009). Technically, hurricanes are uncommon at high latitudes, where extratropical storms are generated by weather fronts with extreme contrasts in air temperature on opposite sides of an advancing line (Pielou, 2001, p. 42). However, typhoons in the western Pacific Ocean at midlatitudes are known to affect cliff retreat and the development of coastal boulder deposits of similarly large size in the Philippines (Kennedy et al., 2017).

Reviewed by Johnson, Ramalho, and Silva (2020), the region of the Cabo Verde Archipelago at a latitude 15° north of the equator is well known as the origin of hurricanes that typically form under conditions of unusually warm surface waters (26–27°C). During an average September, the North Atlantic hurricane season may spawn as many as a dozen disturbances that often dissipate in the central, extratropical parts of the ocean. Only a few of such disturbances develop each year as Category 1 hurricanes reaching wind speeds greater than 119 km/h; however, the 2020 hurricane season proved to be unusual with conditions that affected weather conditions farther afield (Klotzbach et al., 2020; Probst et al., 2021). The season began early with two named tropical storms (Arthur, Bertha) registering wind speeds between 63 and 118 km/h. A record number of six hurricanes crossed the Atlantic to reach North and Central America. On 7 September 2020, Tropical Storm Rene introduced strong winds and heavy rain to the northern islands of the Cabo Verde Archipelago, including Sal. Although the storm failed to reach the intensity of a Category 1 hurricane, a rating of 11 on the Beaufort Wind Scale falls just below Category 1 as a major storm, with wind speeds between 103 and 118 km/h capable of generating waves toping 16 m. Such a level of wave intensity is more than three times the wave height calculated to loosen blocks from the limestone cliffs on the north shore of Sal (see previous discussion).

Although hurricanes are commonly spawned around the southern part of the Cabo Verde Archipelago, they typically build in intensity only after leaving the region. On a more regular basis during the winter months, ocean swell originating in the western Atlantic reaches the Cabo Verde northern islands, including Sal. Locals on Sal refer to the lavarias as a washing-out time when high surf hits the shore, often causing major coastal flooding. Such occasions are unrelated to storms and take place under sunny skies. No known estimate of wave heights related to these semiannual events is available, but even at a rating well below 11 on the Beaufort scale, the resulting wave shock against the north coast would exceed that associated with the prevailing NE trade winds with wave heights in the range of 3.5 m.

Haloclasty represents physical weathering that occurs when salt crystals grow and expand to degrade rocks after saline water seeps into cracks and undergoes evaporation. Clearly absent from riverine potholes bored in swift-running fresh water, this type of corrosion is a distinct possibility in coastal settings exposed to sea water. The process, however, remains poorly studied (Doehne, 2002). The presence of salt in virtually all potholes from the study site on the north coast of Sal includes many less than 35 cm in diameter (Figure 8), where the work of a small grinder stone is likely to be negligible. In this case, haloclasty could be a contributing factor in the initiation of pothole formation as a result of the crystallization and thermal expansion of salt crystals. This style of weathering may even be exacerbated by the climate characteristic throughout the Cabo Verde Archipelago with elevated temperatures.

The relationship of clast-size mobility as a function of wind speed and the generation of wave heights at sea is hypothetical because eye-witness verifications made by coast watchers is unlikely under dangerous conditions. During observations by the research team exploring the north coast of Sal Island in July 2024, wave shock in the intertidal zone was evident but limited to the influence of the NE trade winds (Figure 2). Wind-induced swells related to those trade winds are sufficient to induce the growth of a marine biofilm at the study site but insufficient to functionally overtop the 8 m high outer lip of the natural amphitheater (Figure 2). A gale that conforms to an 8 on the Beaufort scale will generate waves as much as 7.5 m in height; Tropical Storm Rene that struck Sal Island in 2020 may have been in that league. No basalt grind stones were found at the study site. They are unlikely to be introduced given the 8 m interval between the top of the basalt layers in the intertidal zone and the outer lip of the overlying limestone in the natural amphitheater. Thus, the source of grind stones responsible for the creation of potholes is effectively limited to the plucking of limestone blocks from the feature’s surrounding walls.

Another aspect still under investigation is that even rarer tsunami events are responsible for the formation of boulder fields elevated on the Pleistocene terrace at the north end of Sal Island. Evidence for the impact of a mega tsunami on Santiago Island as a result of a volcanic flank collapse on Fogo Island some 68,000 to 73,000 years ago is well constrained as to those leeward islands in the Cabo Verde Archipelago (Cornu et al., 2021; Ramalho et al., 2015). Such events, however, would long predate the erosional activities responsible for the erosion of a sea cave at the study site under consideration in this contribution.

Description of marine potholes in the scientific literature remains relatively uncommon compared with potholes more typically found in riverbeds. Direct and indirect observations based on deductive evidence leads to the following conclusions in relation to well-developed potholes on the floor of a natural amphitheater incised in a limestone cliff face on the north shore of Sal Island in the Cabo Verde Archipelago. Foremost, the floor surface is found to cover approximately 250 m² with an arcuate shape eroded roughly 6 m below the edge of an enclosing Pleistocene limestone terrace. The seaward edge of the structure drops another 8 m to a contact with layered basalt near the intertidal zone. Direct evidence from a resident limestone boulder 0.70 m in diameter occurs within a meter-deep pothole bored into the limestone surface. Indirect evidence suggests that limestone boulders also bored potholes that eventually coalesced into a single, enlarged structure with two or more depressions. A mortar-and-pestle style of erosion is envisioned, such that both are made of the same material. The enclosing mortar expands in shape, whereas the pestle diminishes in size during the process of abrasion. The resulting sand product is displaced by the circulation of seawater at the same time. Ranging in diameter from circular 5 cm dimples to much larger depressions up to 5 m in size, all potholes retain salt rims, and some retain a liquid brine. Using available hydrodynamic equations, it is possible to estimate the effective height of wave swell at 7.5 m necessary to pluck limestone blocks from the limestone cliff face enclosing the natural amphitheater. Extremely large blocks that now lay on the amphitheater floor are related to the collapsed rim of the Pleistocene terrace. Such a large erosional indentation in the cliff face on the northern coast of Sal Island is best explained by the gradual enlargement of a sea cave that eventually became unroofed by instability.

SPA and PM acknowledge contracts with project M1.1. A/INFRAEST CIENT/A/001/2021—Base de Dados da PaleoBiodiversidade da Macaronésia, funded by the Regional Government of the Azores. SPA also acknowledges his current FCT/2023.07418 CEEECIND research contract with BIOPOLIS. This work benefited from FEDER funds, through the Operational Program for Competitiveness Factors—COMPETE, and from National Funds, through FCT (UIDB/50027/2020, POCI-01–0145-FEDER-006821, UIDB/00153/2020, LA/P/0048/2020), as well as through the Regional Government of the Azores (M1.1.a/005/Funcionamento-C-/2016, CIBIO-A; M1.1.A/INFRAEST CIENT/A/001/2021; M3.3.B/ORG.R.C./005/2021). Finally, this work received support from FEDER funds (85%) and from the Regional Government of the Azores (15%) through the project M1.1.A/INFRAEST CIENT/A/001/2021—Base de Dados da PaleoBiodiversidade da Macaronésia.