Boulder-Cobble Beaches, Eastern Tasman Bay, New Zealand Free

Boulder and cobble beaches, spits, and barriers are vital natural structures that dissipate wave energy and can act as natural barriers to impede tsunamis and storm surges. Further understanding of their formation help comprehend their behavior and potential morphological changes as sea level rises and storm intensity increases in the future (Rubinato, Heyworth, and Hart, 2020).

Throughout the world, boulder and cobble beaches are commonly associated with mid-high, energy-wave environments (Carter and Orford, 1984; Lorang, 2000). Boulder-cobble beaches at Nelson (Boulder Bank) and Cable Bay have traditionally been interpreted as spits (Figure 1; Bruce, 1962; Hockstetter, 1864; Johnston, 1976, 1979, 2001); however, incident wave energy is too low to transport boulders and cobbles persistently along these beaches by typical spit-forming processes such as longshore drift (Hartstein and Dickinson, 2006). In addition, Pleistocene glaciation did not occur in this area of New Zealand, so glacial processes could not have emplaced or reworked the clasts (Dickinson and Woolfe, 1997). During this period, storms may have been more intense, but wave energy was constrained by a limited fetch and the shallow nature of Tasman Bay, which would have been subaerial during glacial periods of low sea level (Clement, Whitehouse, and Sloss, 2016; Hartstein and Dickinson, 2001).

This paper describes two boulder-cobble beaches in eastern Tasman Bay that have not been previously documented (Figure 1). Geomorphologically, the boulder-cobble beach at Greville Harbour is a barrier that blocks the entrance to the inner part of a drowned valley (Figure 2). The boulder-cobble beach at Motuanauru Island is geomorphologically a spit (Figure 2). Unfortunately, the geomorphological terms for these boulder-cobble beaches imply a genetic origin for their formation, which is not consistent with the data.

This paper also develops a conceptual model of the beaches at Greville and Motuanauru. This model is not consistent with the traditional interpretation for the origin of the Nelson Boulder Bank. The Greville and Motuanauru Beaches are smaller and younger than the Nelson Boulder Bank, which is larger and may have developed over the last million years to its present form. Thus, the two smaller beaches may provide analogues for the much larger beach at Nelson.

Any model concerning the origin of boulder-cobble beaches and platforms in eastern Tasman Bay must account for (1) a local source of clasts, (2) rounded shape of clasts, (3) poor sorting of clasts, (4) low sloping foreshore, and (5) low wave energy in Tasman Bay. In addition, the boulder-cobble clasts form an erosional surface that lies above poorly sorted gravels over bedrock. The clasts appear to be derived from this bedrock, which lies seaward and has become buried in its own gravels. Although modern coastal processes have modified these beaches, their overall shape and position are a product of wave quarrying in conjunction with the geologic structure and characteristics of the bedrock.

In eastern Tasman Bay, most studies on boulder-cobble beaches have concentrated on the Nelson Boulder Bank, which is a barrier, and Cable Bay, which is a tombolo-shaped barrier and provides an analogy to the boulder bank (Figure 1a; Dickinson and Woolfe, 1997; Hartstein and Dickinson, 2001, 2006). The Nelson Boulder Bank is a 13 km long gravel barrier that comprises a gravel ridge and a boulder platform. Its origin has been the subject of extensive debate with two possible models for its formation.

The first model suggests that the Boulder Bank formed by longshore drift because of its spit-like shape, single rock type, apparent source to the NE, mobile gravel ridge, and decreasing clast size along its length (Bruce, 1962; Johnston, 1976, 2001). The second model suggests the boulder platform is an erosional lag of boulders and cobbles that overlie a thick accumulation of poorly sorted gravels. These gravels were produced by wave quarrying of a former cliff, which ran the length of the barrier but now exists only at the head of the barrier (Dickinson and Woolfe, 1997; Hartstein and Dickinson, 2006).

At the Nelson Boulder Bank, a gravel ridge sits on top of the boulder platform. Although this ridge is mobile in the present wave environment, the boulder platform has a seaward slope of 1–3º and could not be active under the present or past wave environments (Dickinson and Woolfe, 1997; Hartstein and Dickinson, 2006). The surface of the platform is erosional and not depositional. This is further supported by the fact that no known mechanism can transport some of the large boulders (>2 m a axis and weighing several hundred kilograms) 13 km along shore (Dickinson and Woolfe, 1997; Hartstein and Dickinson, 2006). The second model suggests that the boulder platform formed essentially in situ with modification from wave activity.

The rounding of clasts has also been used as evidence for longshore drift (Bruce, 1962; Johnston, 1976, 2001); however, experimental evidence and field observations suggest that cobbles and boulders in streams and beaches can be rounded and reduced in size within short distances from their source. This is because of in situ abrasion when clasts are subjected to short-term fluctuations of lift and drag forces (Bartrum, 1947; Kelly, 1983; Schumm and Stevens, 1973). Thus, the clasts vibrate and rotate in place, becoming rounded.

The tombolo-shaped barrier at Cable Bay comprises two key geomorphological features (Figure 1a): a gravel barrier made of cobbles and pebbles that blocks the Delaware estuary that is situated some 700 m landward from the mouth of the bay and a boulder lag that extends from the mouth of the bay to the tombolo-shaped barrier. The boulder lag was left behind as sea level rose and wave energy within the bay could no longer move them shoreward (Hartstein and Dickinson, 2001, 2006). Both the boulder lag and barrier are composed of granodiorite that is only found at the mouth of the bay.

In this paper, the terms boulders, cobbles, and pebbles refer to particle size, as described by Blair and McPherson (1999). Gravels refer to a mixture of particles generally smaller than cobbles but larger than sand. The term clast refers to a rock fragment but is interchangeable with the term particle. The term beach is widely used but has little consistency in definition. The term beach is used here to include the accumulation of unconsolidated sediment that is limited on the seaward margin by low tide and by storm-wave deposition on the landward side (Davis, 1984).

Early studies on boulder beaches attempted to extrapolate from studies on sand and gravel beaches and looked at details rather than broad patterns of sedimentation and morphology (e.g.,Bartrum, 1947; Hills, 1970; Shelley, 1968). Doubts concerning the validity of such extrapolations were fully confirmed in a comprehensive study of boulder beaches by Oak (1984) along the coast of New South Wales in Australia. Oak (1984) proposed that boulder beaches demonstrated certain unique sedimentary characteristics that distinguished them as fundamentally different from pebble and cobble beaches. Hence, relationships established in the many studies of sand and gravel beaches were often not applicable to boulder beaches.

The slope of a beach is considered by many as the primary index of morphological response to wave action (McKenna, 2005). Generally, the slope of the beach-face is controlled by two factors, wave intensity and sediment size, where size is the more critical variable (Bascom, 1951; Shepard, 1973). Most of these studies, based on sandy beaches, found that beach-face slope increased as particle size increased. Coarse-sediment beaches, which are highly permeable, form steep foreshores because the slope-reducing backwash is diminished by percolation. Continuing the trend into boulder-sized particles shows that they have beach-face slopes less than 15° (Bujan, Cox, and Masselink, 2019). Furthermore, the average steepness of boulder beaches appears to decrease with increasing particle size (Bujan, Cox, and Masselink, 2019), which supports Oak’s (1984) supposition that boulder beaches are fundamentally different from pebble and cobble beaches.

All beaches, regardless of their sediment-size composition, reflect the most recent wave action that is competent in moving particles (Lorang, 2000). On beaches where not all waves are competent, the minimum competent wave energy increases as particle size increases (Oak, 1984). There is no reason to doubt that beach-face slope increases with particle size, but it does so only when waves are competent to transport all sizes of the available sediment. Thus, boulder beaches are unlikely to attain the steep slopes predicted from sandy beaches because wave energy is not competent to permit the development of a steep equilibrium beach profile (Oak, 1984).

Most boulder-cobble barriers have moved landward under transgressive (relative rise) postglacial, sea levels. Transgressive sea levels lead to a rollover of the barrier during storm events (Carter and Orford, 1984; Carter et al., 1987). In this process, boulders remain as evidence of earlier shorelines, whereas the main body of the beach face migrates inland tens of meters or more (Carter and Orford, 1984; Hartstein and Dickinson, 2001, 2006). Boulder barriers migrate primarily by washover and overtopping processes. Migration is landward and not lateral along the barrier because wave energy is competent enough to move only boulders landward. Percolation into the porous gravels reduces the strength of the backwash, which cannot move large clasts back into the swash zone to produce a saw-tooth (zig-zag) movement in the presence of wave-induced longshore currents (van Rijn, 2014). Therefore, large clasts can move only landward. Transgressive or landward migration of a boulder barrier may continue even in the absence of sea-level rise during large storm events (Carter and Orford, 1984).

The rocks that crop out around the areas of Greville Harbour and Motuanauru Island provide the source material for the boulder-cobble beaches. Maitai Group rocks (Permian 280–225 Ma) cover both areas and comprise low-grade metamorphic rocks with poorly to moderately well-bedded, sandstones, siltstones, and mudstones (Rattenbury, Cooper, and Johnston, 1998). These sediments are sourced largely from volcanic and sedimentary rocks, giving them a greenish color. The beds at both areas are steeply dipping, which means their lithology and mass characteristics change within short distances along the shoreline.

Within Maitai Group rocks, the Stephens subgroup crops out at Motuanauru Island (Rattenbury, Cooper, and Johnston, 1998). Clasts on the spit generally show thin (<1 cm) beds alternating between sand and mudstone. At Greville Harbour, the Greville Formation, which is within the Maitai Group, is exposed on both sides of the harbor (Rattenbury, Cooper, and Johnston, 1998). Clasts on the barrier are generally a fine- to medium-grained sandstone, which have either a brown or green tint.

The Maitai Group rocks, which crop out around the boulder beaches, are generally fractured and break into small clasts when attempting to extract samples. In contrast, the clasts on the beaches are hard and solid and difficult to break with a hammer. This suggests that wave processes are exceptionally efficient in winnowing fractured rock from unfractured rock in landslip debris and cliff faces. The ratio of unfractured or solid rock to fractured rock that has been removed is unknown. This ratio would be essential when estimating the amount of cliff retreat over time.

Acoustic surveys (side-scan sonar and echo sounding), topographic surveys (drone aerial photography), wave climate modeling, and clast counts were conducted on two boulder-cobble beaches at Greville Harbour and at Motuanauru Island.

Bathymetric data within Greville Harbour and around Motuanauru Island were collected using a single point echosounder (D390 single beam echosounder made by CHCNAV) coupled to hypack processing and navigation software. A series of survey lines (swaths) were established seaward and landward of both coarse gravel features. Each line was spaced 50 m apart. Bathymetric data were converted into XYZ data and converted into ArcGIS files to compile detailed bathymetric maps of both study areas as well as to provide high-resolution data for the numerical modeling.

Surveys of the seabed were undertaken using a CM800 Side-Scan Sonar unit. The system was operated at high frequency (325 kHz) with a swath width of 50 m per channel. A total of 15 perpendicular and eight parallel sonar swaths were taken at Greville Harbour. At Motuanauru Island, seven perpendicular and five parallel sonar swats were collected, each spaced 75 m apart. The beam depression of the side-scan sonar was set to 19°, and the transducer was embedded into the hull of a 9.5 mm mono-hull aluminum vessel that was used to conduct the survey. To accurately position the survey, an OmniStar differential positioning system was used. Processing of the side-scan data was undertaken using hypack mosaic software from which the files were converted into Arc GIS format.

Results from the side-scan sonar survey were verified using a remote video camera (Deep Blue Pro, Splash Cam). The video camera was lowered on a cable from the surface to provide images of the seabed. Several towed transects were made at each site to identify the different areas of backscatter identified on the sonagraphs. In addition to the transects, 10 camera drops were also made at random locations to further determine seabed type. Seabed features such as boulders, bed rock, sand, and mud were noted.

A topographic survey was conducted at both beaches using a fully automated DJI Air 2S drone with DJI software (https://www.agisoft.com). The drone was flown at a height of 60 m along parallel transects across both gravel features. Drone images were postprocessed with Agisoft Metashape, a Structure-from-Motion photogrammetric program (Westoby et al., 2012). Agisoft Metashape processes digital images and generates three-dimensional spatial data, which can be georeferenced and orthorectified. Three nearby benchmarks were used for the georeferencing process. The software used the drone images to build dense point clouds and produce a digital elevation model. Both bathymetry and digital elevation model data then were compiled in GIS software (ArcGIS; https://www.agisoft.com) to build composite of bathymetric-topographic maps of the survey areas (Figure 2). Topographic resolution was less than 30 cm, after postprocessing, whereas bathymetry resolution was less than 50 cm.

To calculate the wave climate within Tasman Bay—with a focus on storm conditions and the potential of these waves to move boulder-sized clasts along both beaches—numerical modeling software was used (Mike 21FM; DHI, 2025). A regional model that included much of the Tasman Sea was coupled to a local model (Tasman and Golden Bays), which covered an area of approximately 4900 km². The local model had a minimum mesh resolution of 25 m within both the Greville Harbour and Motuanauru Island study areas and a maximum resolution of 3600 m at the outer boundary where Tasman Bay enters the Tasman Sea.

Bathymetry was obtained from nautical charts sourced from Land Information New Zealand (https://www.linz.govt.nz/) and bathymetric surveys conducted within Greville Harbour and around Motuanauru Island at a resolution of <1 m, as described previously. The wave model was run for a period of 25 years with the results presented from a 14-day period (16 March–9 April 1998). To capture an exceptionally large wave event, this period included Cyclone Yalli, which was one of the largest cyclones to hit Tasman Bay over the last 50 years (Hartstein and Dickinson, 2006). Waves across the domain were modeled by wind data from Climatic Forecast System Re-analysis (https://www.weather.gov/ncep/), which is a weather forecast model developed by National Oceanic and Atmospheric Administration and National Center for Environmental Prediction that covers the entire globe over a period of 33 years from 1979 to 2011 and extended up to 2022 (for this study).

To determine the size distribution of clasts on the surface of the boulder-cobble beaches, 50 m transect lines were established at 17 sites on the Greville barrier and at seven sites on the Motuanauru spit (Figure 3). At each site, a 50 m tape was laid out, and only clasts that lay directly under 1 m spacings were counted (Figure 4a). For each clast, the long (l), intermediate (i), and short (s) axes were measured, and the geometric mean size (l × i × s)^0.33 was calculated. When clasts were too big to move by hand, the dimensions were estimated. The tape measure was re-positioned three to four times at each station so that 150–200 clasts were counted. Mean clast size for each site on the profile was calculated as the first moment, and sorting was calculated as the second moment (Folk, 1980).

This method determines size distribution from area percentage and is comparable to standard sieve methods, which determine size distribution from weight percentage (Adams, 1977). For clasts larger than about 20 mm, roundness was estimated from a chart and averaged at each site (Krumbein, 1941). Although the dimensions from Krumbein (1941) were strictly measured from 0 to 1, the decimal point was left off; for simplicity in the field, roundness was counted from 1 to 10.

Several pits were dug to about 40 cm below clasts on the surface. Because of side-wall collapse and water infiltration, digging deeper than about 40 cm was not possible. Clasts from the deepest part of the pit were shoveled into a 10 L bucket and were measured and counted. Clasts larger than −4 phi were counted and determined as volume percentage. Clasts and particles smaller than −4 phi were sieved, and modes were determined by weight percentage using Folk’s (1980) moment method. A mixture of shells was collected from the lowest part of the pit at site 475ss to obtain an average radiocarbon age (Supplementary Material 1).

Similar methods and data collection were used at both areas; however, each area had similar but different characteristics, and results are provided in separate sections.

In Greville Harbour, the boulder-cobble barrier separates the outer harbor from the inner harbor (Figure 1b; Supplementary Figure 1a). A narrow (75 m wide) channel, 5–6 m deep, cuts the barrier into northern and southern segments. A narrow gravel ridge, 2–3 m higher than the boulder-cobble platform, extends about 200 m southward from the northern spur of the north barrier (Figure 2a; Supplementary Figure 1b). With the exception of this gravel ridge, both north and south barriers are covered by water at high tide and exposed only at low tide. The tidal range is 3 to 4 m (LINZ, 2024).

The northern barrier, narrower and longer than the southern barrier, has a distinct bend about half way along its length (Figure 2a). This bend area is low in elevation, and it is the first section of the northern barrier to flood with the rising tide. A small outcrop of bedrock is exposed on the seaward edge of the barrier where it attaches to the spur of the ridge at the north (Supplementary Figure 1b).

The southern barrier is shorter, wider, and flatter than the northern barrier. From where it attaches to the spur at the south, it generally slopes downward to the north. At its northern end near the channel, a low hollow that appears as a pond occurs at low tide.

Clasts on the barrier comprise the Greville Formation, which crops out in the cliffs to the north and south of the barrier (Rattenbury, Cooper, and Johnston, 1998). Although outcrops on the cliffs are fractured and weathered, clasts on the barrier are solid and not fractured. A Zingg plot of clast dimensions on the barrier shows that most of the clasts fall into the disk or platy quadrant of shapes.

The average sorting, roundness, and size of the clasts shows small discernable trends in spatial distribution across the barrier (Table 1; Figure 3a–c). The largest clasts (average of 360 mm diameter) occur on the seaward slope of the northern barrier (525 m) near the channel cut (Figure 3c). Clasts on the southern barrier are generally smaller than those on the northern barrier. The surface of many clasts shows evidence of salt weathering such as pitting, fracturing, spallation, and small cavities of tafoni-like textures (Figure 4b). Plots of sorting, roundness, and size of clasts showed no apparent correlations, and hence statistical tests were not made.

A medial gravel ridge, which sits on top of the boulder-cobble platform, extends about 200 m southward from the north cliff face (Supplementary Figure 1b). This ridge is above midtide and remains exposed at high tide. Clasts on the ridge are significantly smaller and slightly better sorted than those on the platform.

At sites 475 and 750, small pits were dug to about 40 cm deep (Supplementary Figure 2a,b). Both pits showed larger, better sorted clasts on the surface but smaller, poorer sorted clasts in the subsurface. Digging deeper than about 40 cm was not possible because the sides of the pit caved in and filled with water. Fragments of mussels, bivalves, and gastropods collected at the bottom of pit 475ss were mixed and crushed and gave a whole sediment radiocarbon age of 2633 ± 26 BP or calibrated at 1 sigma, 297 BC to 125 BC (Supplementary Materials 1).

Side-scan sonar and seafloor bathymetry indicate that in Greville Harbour, the boulders and cobbles on the barrier extend seaward several hundred meters and were observed at depths greater than 12 m (Figure 4c; Supplementary Figure 3). No break occurs in slope between the subaerial beach face and the submarine boulder-cobble slope extending off shore. The northernmost profile (A-A′) shows a seaward slope of about 2.6°, whereas the southernmost profile (F-F′) is less than 1° (Supplementary Figure 4).

Greville Harbour is sheltered in most directions, other than a small window to the W/NW (Supplementary Figure 1a). During the modeling period (Cyclone Yalli), winds were observed to generate a maximum wave height approximately 1.0–1.5 m directly seaward of the existing boulder beach. Higher waves are observed (above 3 m) at the mouth of Greville Harbour (Figure 5a,b). Using previous gravel entrainment observations and boulder measurements and calculations made by Lorang (2000) and Hartstein and Dickinson (2006), a maximum boulder entrainment of 0.2 m was calculated on the existing boulder beach (slope of 3°). At the mouth of Greville Harbour, a 3 m wave could entrain boulders between 0.3–0.6 m in diameter, assuming a beach slope of between 2–8°. Except for the southern end, waves impinge most of the barrier at a perpendicular angle; however, wave height decreases southward along the barrier (Figure 5b).

The Motuanauru Island spit is a long, narrow ridge of boulders and gravels that extends from a steep, spur ridge almost 300 m to the SE (Figure 2b). At low tide, it is about 125 m wide at the base of the ridge but narrows to the SE as it slopes below sea level. The spit slopes about 1° to the SE.

Clasts on the spit comprise the Stephens subgroup, which crops out on Motuanauru Island (Rattenbury, Cooper, and Johnston, 1998). Although outcrops on the island are fractured and weathered, the clasts on the spit are solid and not fractured. Similar to the clasts on the Greville barrier, a Zingg plot of clast dimensions on the spit shows that most of the clasts fall into the disk or platy quadrant of shapes.

The average size, roundness, and sorting of the clasts show small discernable trends in spatial distribution over the spit (Figure 3d–f). The largest clasts are boulders both on the north and south sides of the spit about 150 m from the base of the spur ridge (Figure 3f).

The medial ridge, which is about 1–2 m above the north and south platforms of the spit, comprises mostly poorly sorted gravels that are significantly smaller than clasts on the adjacent beaches (Figure 4d). Sorting of ridge gavels becomes poorer with distance from the base of the ridge (Figure 3d). Plots of sorting, roundness, and size of clasts showed no apparent correlations, and hence statistical tests were not made. Smooth outcrops of the Stephens subgroup are exposed on the flanks of the medial ridge as well as between boulders on both the north and south beaches (Figure 4d). In places, a thin layer of boulders and gravels rest on the rock outcrops.

Side-scan sonar and seafloor bathymetry indicate boulders and cobbles on the spit extend seaward below low tide for at least several hundred meters, and boulders were observed at depths of 5 m in the waters around Motuanauru Island (Supplementary Figure 1d). With the exception of the south side of the spit along Profile A-A′, no break in slope occurs between the subaerial beach face and the submarine boulder-cobble slopes extending offshore. Bathymetric profiles have slopes between 3.0° and 1.4°, which generally become less steep landward to the SE (Supplementary Figure 4).

The Motuanauru gravel and boulder beach is sheltered by the island itself and a rocky outcrop to the NW. Model results indicate a maximum wave height of approximately 2.0 m along the gravel beach and up to 3 m on the seaward side of Motuanauru island (Figure 5c). This is a similar wave height to what was observed at Greville Harbour. Assuming a beach slope of 3.5°, the maximum size boulder that can be entrained along the gravel beach is approximately 0.25 m; however, on the seaward side of Motuanauru Island, boulders up to 0.4 meters could be entrained. Direction arrows show that waves appear to refract around the island and impinge at an oblique angle on the spit (Figure 5c).

On most beaches in the world, the wave energy sorts particles, and then these fall on a plot of beach-face slope vs. size (Bujan, Cox, and Masselink, 2019); however, on Greville and Motuanauru beaches, the boulders and cobbles show no sorting by waves, and they do not fall on the statistical curve of Bujan, Cox, and Masselink (2019) for more than 2100 beaches throughout the world (Figure 6). Although few data points for boulder-cobble beaches are found on the Bujan, Cox, and Masselink (2019) plot, all these beaches in this size range have seaward slopes of between 10° and 30°. None of the beaches from Bujan, Cox, and Masselink (2019) have flat seaward slopes of between 1° and 3° (similar to those at Greville Harbour and Motuanauru Island).

The fact that boulder-cobble beaches in this study do not fall on the curve from Bujan, Cox, and Masselink (2019) suggests that the beaches are unique and have not formed through normal beach processes, where wave energy acts to sort particles based on size and density. This supports the observation that the boulder-cobble beaches are erosional and that these clasts are immobile in the present wave environment. Although large clasts on this erosional platform are left behind, the small clasts are moved shoreward by waves. At Greville, these clasts end up either in the medial ridge on the north end of the barrier or they are swept behind the barrier. Movement of clasts along the barrier is nil because wave backwash is too low to return the clasts to the swash zone. At Motuanauru, the small clasts are swept into the medial ridge and along the spit. Medial ridge gravels from both beaches plot close to the curve from Bujan, Cox, and Masselink (2019), suggesting that they are partly influenced by waves (Figure 6).

It is unclear why the two boulder-cobble beaches in this study differ, with one forming a barrier and the other a spit. Antecedent geology and topography may be the main reasons for the difference, but the direction of wave impact on the beaches also plays a role. Although antecedence sets the stage, wave energy modifies the relic topography. At Greville, wave impact is perpendicular to the beach, which forms a barrier in the middle of the estuary. At Motuanauru Island, wave impact is parallel to or at low angles to the beach, resulting in the formation of a spit. Thus, antecedence and wave energy are closely linked in the origin of these beaches.

Despite these differences, the boulder-cobble beaches at Greville Harbour and Motuanauru Island have a number of similar features. From these similarities, common processes that lead to a model for understanding were developed regarding how the Greville barrier and Motuanauru spit formed.

Both beaches have medial ridges of gravel that are above high tide. Particles on these ridges are generally smaller than those on the boulder-cobble platforms. These platforms slope 1–3º seaward and are fully submerged below midtide level. The platforms extend seaward in the case of Greville Harbour and laterally at Motuanauru Island for several hundred meters before they are buried by fine-grained sands and silts (Figure 4c). At this point the slope of the platform is obscured by the flat-lying sands and muds. The angle of the platform slope appears to increase with increasing wave energy. For example, at Greville Harbour wave energy is highest at the north end of the barrier (Figure 5b); this coincides with the steepest slope (2.6°) of the boulder-cobble platform (Supplementary Figure 4).

Clast characteristics are similar on both beaches. The average size, roundness, and sorting of the clasts show only minor discernable trends in spatial distribution over the beaches (Figure 3). In addition, no meaningful correlations between size, roundness, and sorting of clasts occurred; however, at Motuanauru spit, clasts on the medial gravel ridge are significantly smaller than clasts on the adjacent boulder-cobble platforms (Figure 4d). At Greville Harbour, clasts on the medial ridge are only slightly smaller than those on the boulder-cobble platform.

On both beaches the clasts appear to be locally sourced because they are similar to rocks, which crop out on the adjacent cliffs. With few exceptions these clasts are sub-rounded; therefore, clast rounding does not relate to distance of travel. This suggests that the rounding of clasts occurs by wave tumbling and vibration in place, as described by Schumm and Stevens (1973) for rivers and by Bartrum (1947) for beach boulders. On the Greville barrier, many clasts show evidence of salt weathering, indicating little abrasion in the swash zone and stability on the erosional surface (Figure 4b). Zingg plots for clasts on both beaches show that the majority are disk or platy in shape. This most likely reflects the sedimentary bedding of the source rocks.

Both beaches are associated with bedrock. At Greville, vertically dipping sedimentary rocks crop out at the NW end of the barrier (Supplementary Figure 1b). At Motuanauru spit, bedrock crops out below many of the clasts eastward from the base of the cliff for approximately 100 m (Figure 4d).

To build a conceptual model for the origin of these boulder-cobble beaches, similar features, as described, were used and then linked to sedimentary processes, which are complementary but not identical. Clasts on the boulder-cobble platforms of these beaches are not evenly distributed or sorted. This suggests that sorting in a wave environment is not the dominant process of deposition for the boulder-cobble beaches, which do not fall on the curve for particle size vs. foreslope angle of Bujan, Cox, and Masselink (2019). It was assumed that many clasts on the Greville and Motuanauru beaches are moved shoreward with the direction of breaking waves during high-energy events.

The estimated age of shells in the subsurface at Greville barrier (site 475) suggests that about 40 cm of sediment has accumulated in the last 2200 years, which is less than 2 cm every 100 years (Supplementary Materials 1; Figure 2a). This sediment was eroded on the seaward face of the barrier and moved shoreward. Clasts that could not be moved shoreward by breaking waves are left behind as a lag on the surface of the beach gravels. Although the radiocarbon age gives a crude rate for deposition, the rate of erosion is largely unknown.

Both boulder-cobble beaches being antecedent to the topography and geology is crucial in developing the conceptual model of origin. Antecedence has been found to control the evolution of sand-rich barrier islands globally; however, antecedence has not been reported to control the evolution of coarse clastic beaches (Cooper et al., 2012; Dillenburg et al., 2000; Gal et al., 2021).

The basic premise here is that the shapes of the Greville Harbour barrier and the Motuanauru spit indicate that the beaches originated from the erosion of the ridges to which they are now attached (Figure 4d; Supplementary Figure 1b). In addition, the beaches would form where the bedrock geology is hard enough to form resistant boulders and cobbles. Much of the erosion results from wave erosion at the base of the resistant bedrock ridges. Scree and landslip debris at the base of the ridges are reworked by waves. Clasts in the debris are rounded in situ, and as the ridge continues to retreat, more and more gravel accumulates to bury the bedrock. Thus, as ridge erosion continues, the gravels become thicker and thicker, and bedrock becomes buried deeper as more and more gravels accumulate on top of it.

In this model, waves not only erode the base of the ridges, but they also round and re-work the bedrock debris at the base of the ridges. The relatively low-energy waves of Tasman Bay transport the clasts shoreward or in the direction of the breaking waves. The distance that the clasts are transported depends on the size of the clasts and the energy of the waves; however, intense salt weathering of clasts at the Greville barrier suggests that transport is minimal (Figure 4b).

The platform-like shallow slope of both boulder-cobble beaches continues 100 to 200 m offshore before it becomes buried by sands and muds at approximately 10 m deep. Clement, Whitehouse, and Sloss (2016) found that paleo sea level around New Zealand at −10 m occurred at 9 ka. This suggests that beach-forming processes were active from at least 9 ka. Thus, the Tasman Bay wave environment formed and reworked the boulder-cobble platform during Holocene sea-level rise.

A resistant bedrock ridge once closed off the inner estuary of Greville Harbour (Figure 7a). As this ridge eroded from both the landward and seaward sides, it was breached by a stream that provided an outlet for the inner valley. The stream would have been active during low stands of sea level (120–20 ka) and is now seen as a channel between the north and south boulder-cobble barriers. A problem with this model is the unknown rate of erosion and, hence, the amount of time taken to erode the ridge. Such erosion may have occurred through several eustatic cycles of sea-level rise and fall.

During Holocene sea-level rise, clasts are continually moved up the beach or shoreward and cover pre-existing gravels on the platform. Waves pushed most clasts landward, overtopping the bedrock ridge and depositing them on the lee side of the ridge (Figure 7b). This forms a wave-cut platform with an erosional lag of large clasts (boulders and cobbles), which could not be moved landward by waves. Thus, the bedrock ridge becomes buried in its own debris as the barrier migrates landward.

Clasts that could not be moved shoreward remain on the submerged platform seaward of the barrier (Figure 7b). At low stands of sea level, soils would have developed on the exposed boulder platforms, and they might have appeared as wave-cut terraces. The poor sorting of particles on the platform is evidence of reworked scree deposits from the eroding ridge. Rounding of the clasts took place in the wave zone from in situ rolling and vibration.

The boulder-cobble spit at Motuanauru Island appears to have formed from the erosion of an east-trending ridgeline (Figure 8). This ridge eroded westward toward the center of the island. Active erosion of the spit appears to occur on the south-facing part of the island where a bedrock platform is forming at the base of the cliff. Clasts eroded from this platform are moved eastward or along the spit by breaking waves moving from west to east. Erosion of this platform provides the sediment that buries the former ridge.

Unlike the Greville barrier where waves break perpendicular to the barrier, waves eroded the retreating spit at an angle from both the north and south sides. Boulders and cobbles, which could not be moved by waves, formed platforms during Holocene sea-level rise. Because waves approached from two directions, profiles of the submerged platform at Motuanauru are more symmetrical than those at Greville where waves approached from only one direction (Supplementary Figure 4).

Eastern Tasman Bay has a number of boulder-cobble beaches that are unusual in the world because they occur in relatively low-energy wave environments. In addition, they are exposed only at low tide and have shallow-dipping foreslopes. Sitting on top of these boulder-cobble beaches is a gravel ridge, which is above high tide and comprises clasts smaller than boulder-cobble. The boulder-cobble beaches of the Greville Harbour barrier and Motuanauru Island spit do not plot on the curve of Bujan, Cox, and Masselink (2019) of 2144 worldwide beaches. Thus, sorting and deposition of clasts in a wave environment is not the dominant process by which these boulder-cobble beaches formed.

Antecedent topography and geology played a major role in the origin of the boulder-cobble beaches at Greville Harbour and Motuanauru Island. These beaches are remnants of eroding ridges, which have been reworked and smoothed by waves during Holocene sea-level rise. The beaches are erosional platforms where clasts on the surface are those that are left behind and not entrained by breaking waves. The poorly sorted distribution of clasts on the beaches reflects the landslip debris from the eroding cliffs. The beaches are part of a shore platform that slopes and extends offshore several hundred meters until buried by bottom muds and sand. The platform was mostly shaped and formed during Holocene sea-level rise.

The geomorphology and clast characteristics of the beaches in this study are similar to those for the Nelson Boulder Bank. The beaches at Greville and Motuanauru are smaller and probably younger than the Nelson Boulder Bank, which is much larger and may have developed over the last million years during multiple changes in sea level to its present form. If the beaches at Greville and Motuanauru are analogues to the Nelson Boulder Bank, then the Boulder Bank is antecedent to an ancient sea cliff, which has eroded and been modified in the wave environment.

Several questions remain from this study. The ridges, to which the boulder-cobble beaches are attached, comprise extremely fractured and weathered rock, yet clasts on the beaches are rounded and unfractured. This suggests that most of the fractured material has been winnowed and lost. Without knowing the ratio of fractured to unfractured rock, it is difficult to estimate the rate of cliff retreat.

The reason that the boulder-cobble beaches form a shallow-sloping platform with an elevation of approximately mean-tide level remains unclear. This platform is cut and smoothed by wave processes during rising sea level, but what determines the slope of the platform is not understood. Empirically, the steeper the wave-cut platform the higher the wave energy, but this observation lacks data and quantification.

Tasman Bay is relatively small and shallow, yet a major difference occurs between beaches in western Tasman Bay and eastern Tasman Bay. Beaches in the west are mostly sandy, whereas beaches in the east are rocky and comprise large clasts. Differences between western and eastern Tasman Bay include rock type, wave climate, and vertical land movement, all of which may give rise to two extremely different types of beaches.

This research was undertaken as follow-up to research on the Nelson Boulder Bank and Cable Bay in the late 1990s. Grace Wrong (ADS Environmental Services) and Julia Martin (Victoria University of Wellington) drafted excellent figures. Dane Dickinson (Weora, Ltd.) flew the drone photography and provided field assistance; the drone was provided by Kevin Norton (Victoria University of Wellington). The principal authors provided funding for the project. We thank four anomalous reviewers, all of whom made especially constructive comments.