Reef island evolution in a turbid‐water coral reef province of the Indo‐Pacific

Coral reef islands are vulnerable landforms to environmental change. Constructed of largely unconsolidated reef‐derived sediments, they are highly sensitive to variations in metocean boundary conditions, raising global concern about their future resilience and stability in the face of increased natural hazards, sea‐level rise and anthropogenic climate change. This study examines the evolution of an inshore turbid reef island from the southern Pilbara region of Western Australia (Indo‐Pacific) using detailed analyses of island chronostratigraphy (composition, texture) and geochronology (21 in‐situ radiometric dates) from Eva Island. Downcore, composition of island‐grade (reef‐derived) sediments were homogenous, dominated by molluscan (37%–42%) and coral (32%–37%) constituents. The 14C radiometric dating of island sediments, beachrock and coral microatolls identified five stages of island formation across changing sea‐level regimes over the mid to late Holocene: (1) limestone platform accretion at ca 6,000 cal yr BP, coinciding with reef decline or ‘give‐up’ on neighbouring Exmouth Gulf reefs; (2) sand cay (i.e. core) initiation and vertical aggregation at ca 5,000 cal yr BP during the point of sea‐level regression to current levels; (3) major accretion and lateral progradation of the island between 3,500 cal yr BP and 2,500 cal yr BP including the modification of island shorelines through alongshore reworking of sediment; (4) lateral accretion and minor expansion to the north and formation of beachrock pavement between 2,500 and 650 cal yr BP; and (5) planform adjustment (erosion of the north‐west island) and backstepping under stabilised sea levels over the past 650 years. While this model is comparative to studies on island formation following incremental sea‐level fall following the mid‐Holocene highstand, it demonstrates active landform readjustment under stabilised sea levels over the past 2,000 years, probably the influence of local‐scale metocean boundary conditions within climate windows across the mid to late Holocene period (i.e. independent of sea‐level fluctuations). Importantly, while sediment production rates are predicted to be lower in turbid‐water reef systems than clear water, Eva Island shows no change in carbonate producers (i.e. proportion of mollusc and coral) over the course of island building, indicating the carbonate factory has not experienced significant adjustments in reef ecology, but has remained stable despite low water quality.


| INTRODUCTION
Coral reef islands are low lying sedimentary landforms that form atop reef platforms or atoll rims.Reef derived biogenic sediments constitute the dominant island grain type, which are transported lagoonward through waves and currents, and accumulate as hydrodynamic energy begins to dissipate (Dawson et al., 2012;Ford et al., 2020;Perry et al., 2011;Yamano et al., 2014).The biogenic sediments that comprise reef islands are predominantly derived from their adjacent coral reefs and consist of the skeletal remains of calcifying reef biota such as coral, crustose coralline algae (CCA), molluscs, foraminifera, echinoids and Halimeda (Bonesso et al., 2022;Kench et al., 2005;Liang et al., 2022;Woodroffe, 1992;Yamano et al., 2000).Reef islands provide a habitable landform for coastal is land nations (e.g.atoll nations) and endemic/threatened flora and fauna (e.g.sea turtles, sea/shorebirds), however, they are highly vulnerable to projected climate change im pacts, particularly sea level rise (SLR) and changing wave climate.These threats may increase the risk of landform destabilisation and erosion, rendering them uninhabitable (East et al., 2018;Ford et al., 2020;Storlazzi et al., 2018).
Globally, numerous studies have established the onset and development of reef islands over the last ca 6,000 cal yr BP, with sea level fall during the mid to late Holocene considered as the critical driver (Dickinson, 2009;Ford et al., 2020;Liang et al., 2022).Reinforcing this island development model is the precondition that the reef plat form, or lagoon infill has reached sea level before island accretion can initiate.For example, reconstructions of is land evolution from the Indian Ocean (e.g.Cocos [Keeling] Islands; Maldives Archipelago), Pacific Ocean (e.g.Kiribati, Central Pacific) and Torres Strait (e.g.Bewick Cay, Great Barrier Reef), demonstrate that island construction initi ated following the mid Holocene highstand (i.e.into the Holocene regression), the point at which accommodation space for vertical coral reef growth becomes constrained in response to a fall in sea level (Kench et al., 2012;Woodroffe et al., 1999;Woodroffe & Morrison, 2001).However, emerg ing geological evidence from the south west Pacific (e.g.Jabat Island, Marshall Islands) and eastern Indian Ocean (e.g.Maldives Archipelago) suggest island formation has occurred prior to (i.e.latter stages of Holocene SLR; Kench et al., 2005) and during the mid Holocene highstand (East et al., 2018;Kench et al., 2014).Collectively, these studies reinforce the variability in island building processes and local effects, demonstrating formation under a variety of sea level scenarios and reef growth (or lagoon infill) histo ries (Kench et al., 2020).
This study examines, for the first time, island build ing processes of an inshore turbid water reef island lo cated within Exmouth Gulf in the southern Pilbara region of Western Australia, using a combination of (1) the point of sea level regression to current levels; (3) major accretion and lateral progradation of the island between 3,500 cal yr BP and 2,500 cal yr BP includ ing the modification of island shorelines through alongshore reworking of sedi ment; (4) lateral accretion and minor expansion to the north and formation of beachrock pavement between 2,500 and 650 cal yr BP; and (5) planform adjust ment (erosion of the north west island) and backstepping under stabilised sea levels over the past 650 years.While this model is comparative to studies on island formation following incremental sea level fall following the mid Holocene high stand, it demonstrates active landform readjustment under stabilised sea levels over the past 2,000 years, probably the influence of local scale metocean bound ary conditions within climate windows across the mid to late Holocene period (i.e.independent of sea level fluctuations).Importantly, while sediment produc tion rates are predicted to be lower in turbid water reef systems than clear water, Eva Island shows no change in carbonate producers (i.e.proportion of mollusc and coral) over the course of island building, indicating the carbonate factory has not experienced significant adjustments in reef ecology, but has remained stable despite low water quality.

K E Y W O R D S
coral reefs, Holocene, Indo Pacific, reef islands, sea level, sediment chronostratigraphic and (2) accelerator mass spectrometry (AMS) 14 C radiocarbon age dating analyses.Unlike global reef island counterparts (e.g.Maldives Archipelago), the Pilbara inshore islands are situated within a region of high spatial and temporal variability in turbidity, fringed by shallow turbid coral reefs and exhibit landmass eleva tions to and exceeding 18 m above mean sea level (amsl) and volumes of 4.7 million m 3 (e.g.Long Island; Bonesso et al., 2020).While prior research has focussed on island building within clear water (oligotrophic) coral reef set tings, no previous studies have reconstructed island devel opment within turbid water reef ecosystems, highlighting a critical knowledge gap.Turbid water systems are sen sitive and have been typically synonymous with lower carbonate production rates compared to clear water sys tems.Yet, as clear water coral reefs trend towards these environments with climate change (Cacciapaglia & van Woesik, 2015;Heery et al., 2018;Zweifler et al., 2021), it is important to understand turbid systems and the im plications for reef island development.An improved un derstanding of contemporary turbid water systems may, therefore, help enhance future management of present day clear water systems and their island landforms under future perturbations in environmental conditions.Thus, this study provides the first detailed reconstruction of reef island evolution in Western Australia, and insights into the onset and timeframes of island formation in a turbid water reef province of Australia, and more broadly, the Indo Pacific region.

MORPHOLOGY, CLIMATE AND OCEANOGRAPHIC CONDITIONS
Island accretion and evolution was examined for Eva Island (21°55′19″ S, 114°25′55″ E), an inshore sand cay located in the Exmouth Gulf, at the southern extent of the Pilbara region, Western Australia (Figure 1A through D).The island planform shape is roughly circular with a total area of 14.7 ha (0.147 km 2 ), volume of 559,427 m 3 , and a maximum elevation of 8.6 m above sea level (Bonesso et al., 2020).Island geomorphology is characterised by a mobile spit, vegetated foredunes (i.e.low lying vegetation [Spinifex sericeus] <1 m in height) and a central basin/ depression.Remnant underlying beachrock is present on the north northwestern periphery and extending off the southern edge of the landmass (i.e.historical island foot print; Figure 2A through F).Surficial sediments collected from the island landmass (beach, foredunes, central basin) and adjacent limestone reef platform/sub reef environ ments indicate that Eva Island and reef is composed of un consolidated biogenic carbonate fragments rich in mollusc (34% of reef and island sediments, respectively) and coral constituents (31% of reef and 27% of island; see Bonesso et al., 2022).
The North West Cape region of Western Australia's Pilbara region has a dry arid climate, with very hot humid summers (average temperature between 36 and 37°C from November to April) and cool mild winters (average temperature between 28 and 29°C from May to October).Rainfall is low (ca 260 mm/year), commonly associated with tropical cyclone/tropical low activity during the first 6 months of the year (with approximately three tropical cy clones annually within the Exmouth region) (Cartwright et al., 2021;Cuttler et al., 2020).The semi enclosed waters of the Exmouth Gulf experience a semi diurnal tidal re gime, with an average tidal range of 1.8 m.The Gulf expe riences seasonal and inter annual variability in wind and wave regimes influenced by: (1) long period wave energy (peak period ca 7-10 s) ocean swell waves that propagate into the Gulf from the north northwest during the Austral winter, and (2) short period, locally generated wind-waves (peak period ca 5 s) from the south southwest during the Austral summer (Cuttler et al., 2020).These factors, in addition to south west flowing tidal currents (Dufois et al., 2017), increase sediment resuspension and result in high spatial and temporal turbidity throughout the year, with mean monthly turbidity (Total Suspended Material, TSM) ranging between 0.75 and 5.7 mg/L (Cartwright et al., 2021).Seasonally, turbidity patterns in the Gulf are higher during the winter (coinciding with the inten sification of easterly trade winds and swell waves), with TSM fluctuating between 1 and 2 mg/L at Eva Island (Cartwright et al., 2021).

| Core recovery, logging and analysis
Four sediment cores (hereafter: PC2-PC5) were collected at Eva Island using a manual sand auger in October 2021, re covered at 20 cm intervals to a maximum depth of approxi mately 5 m below the island landmass surface (i.e.foredune ridge, island basin; Figure 3).Following collection, 100 g of sediment was sub sampled from each 20 cm interval along the four cores PC2-PC5 (n = 95 samples, total) for composi tion (i.e.biogenic and siliciclastic) and grain size analyses (i.e.mean grain size [μm] and sorting [σ]).All sediment samples were rinsed in freshwater and dried at 60°C for 48 h prior to processing and laboratory analyses.Standard wet sieving protocol was performed to separate the sam ples into seven fraction sizes (>4,000, 2,000, 1,000, 500, 250, 125 and 63 μm) in accordance with methodology outlined in Bonesso et al. (2022).Sediment texture (average grain [μm] size and sorting [σ]) was calculated in GRADISTAT© (Blott & Pye, 2001; Table 1).Biogenic and siliciclastic com position of the 95 sediment samples was determined using randomised point counts under a Nikon SMZ445 binocular microscope (Belgium Metrology©, Belgium) by identifying a minimum of 200 grains per sample from the 1,000 and 500 μm fractions.These two size fractions were chosen for identification as they represented the dominant sand grade material (by % mass).Sediment was classified into the fol lowing categories: coral, molluscs (univalve gastropods and disarticulated bivalves), echinoderms, CCA, foraminifera, bryozoans, Halimeda, siliciclastic (quartz), limestone, other reefal calcifier (i.e.crustaceans and sponge/coral spicules) and unknown.
Site PC1 is an exposed foredune on the north west periphery of the landmass (i.e.exposed cross section ex hibiting depositional sequences) from which only pristine mollusc constituents were sampled for AMS 14 C radiocar bon age dating (Figure 3).No sediment textural or compo sitional analyses were performed for PC1.

| Radiocarbon ( 14 C) age dating of coral microatoll, beachrock and island sediments
To ascertain relative timing of island accretion and evo lution, 14 C radiocarbon age dating was employed for reef derived sediments (Ford et al., 2020) (Callow et al., 2018;May et al., 2017May et al., , 2021) ) and the resulting DEM and orthomosaic were exported to Fledermaus© (Quality Positioning Services, QPS), with the orthomosaic draped over the DEM to allow 3D visualisation of the island and adjacent reef flat.The photogrammetry derived DEM was used for qualitative observation only (see Supplementary Material for full details on drone imagery processing for Figure 1D).the Chronos 14 Carbon Cycle Facility (Mark Wainwright Analytical Centre), University of New South Wales (UNSW), Sydney, Australia.A total of 21 samples were ob tained from island sediment, beachrock and coral micro atolls (Porites sp.), including 12 intact gastropod shells from sediment cores (Code: PC), three shell fragments extracted from exposed beachrock deposit (Code: BR), three fos sil coral microatoll specimens (Code: CO, Porites sp.) and three intact bivalve shells from the exposed island foredune deposit (Code: E_NW).To ensure accurate and robust 14 C radiocarbon ages are obtained, the most pristine, intact and least durable reefal constituents were selected to minimise the effects of the lag between death and deposition with low likelihood of alteration during storage and/or transport prior to deposition (Kench et al., 2011).Details of the sam ple type and depth relative to MSL are shown in Table 2.  Samples were pre treated following the protocols out lined for carbonates in Turney et al. (2021).To remove any surface contamination, samples were cleaned by physical abrasion and prolonged sonication in distilled water and then surfaces were etched with 0.1 M HCl to remove the outer layer of the sample, before being rinsed with etha nol and oven dried at 70°C (Turney et al., 2021).Samples were pulverised with a vibratory mill in a tungsten car bide grinding bowl with a single disc.Pre treated sam ples were converted to CO 2 by reaction with 85% H 2 PO 4 (1 mL 85% v/v) at 40°C and flushed through a phosphorus pentoxide (Merck SICAPENT®) trap with helium gas to be transferred into the AGE3 system (Ionplus®, Switzerland) (Turney et al., 2021;Wacker et al., 2010).The CO 2 was con centrated in a zeolite trap, heated to 420°C and released as pure CO 2 into the graphitisation reactor tube where it was reduced to graphite over an iron catalyst at 580°C within the AGE3 system (Turney et al., 2021;Wacker et al., 2010).Radiocarbon ( 14 C) measurement was carried out using a MICADAS (Ionplus®, Switzerland) accelerator mass spectrometer (Wacker et al., 2010), following sample mea surement procedures outlined by Turney et al. (2021).The 14 C results are reported as calendar years before present (hereafter: cal yr BP) following correction for the reservoir age.The ages were calibrated using the online software OxCal© v4.4.4 (Ramsey, 1995), the Marine20 calibration curve (Heaton et al., 2020) and the local marine reser voir effect (ΔR) for the region (around Exmouth Gulf) of −61 ± 14 years.

| Sediment texture and composition of reef island cores
Medium to very coarse sand (250-2,000 μm) dominated all four island core samples (PC2-PC5, 94.8%, on av erage) with gravels and silt sized material accounting for 4.2% and 0.9% respectively (Figure 3).Downcore variation was across the island cores, with upper island units generally consisting of medium sands, from which coarse and very coarse sand layers occur with increasing depth from the surface (452-1,797 μm range across all cores).Sorting (μm) decreased downcore, with moderately well sorted upper units and poorly sorted lower units (1.48-2.94μm range across all cores).Skeletal composition of island sediments was ho mogenous downcore (i.e.no variation between upper and lower sedimentary units) and was dominated by mollusc (37%-42%) and coral (32%-37%) with subordi nate fractions of CCA (16%) and foraminifera (4%-6%; Table 1).4A; Table 2).Dating of disarticulated shell frag ments from exposed beachrock pavement (Code: BR, n = 3) fringing the north northwest periphery of the island landmass recorded an age range of 1,045-2,522 cal yr BP (Figure 4A).

| Island sediment ages
Along Profile 2 (Code: P2) of the island landmass, the old est date on intact gastropod clasts is located at the base of sediment core PC3 (Depth: 4.6-4.75m) with an age of 4,871 cal yr BP (Table 2 and Figure 4C).Dating of gastro pod clasts at 3.2-3.4m and 4-4.2 m depth from PC3 show a 2,000 year time lag in deposition, recording ages of 2,929-2,768 cal yr BP respectively.An age determination of 3,222 cal yr BP was recorded at the base of sediment core PC4 (Depth: 4-4.2 m), followed by an age inversion at a depth of 2.6-2.8m and 1-1.2 m, with sediment clasts dating at 2,504-3,383 cal yr BP respectively.Sediment core PC5 also showed evidence of age inversion across the core profile, with the youngest age determination at the base recorded at 2,812 cal yr BP (Depth: 4.8-5 m), followed by 3,327-3,407 cal yr BP at a depth of 3.4-3.6m and 2-2.2 m, respectively.In summary, age determinations across PC3, PC4 and PC5 are suggestive of deposition between 5,000 and 2,500 cal yr BP (Table 2 and Figure 4C).Gastropod clasts extracted from PC2 on the northern pe riphery of the landmass recorded radiometric ages between 1,813 cal yr BP (Base of core, 5-5.2 m) and 1,370 cal yr BP (Depth: 3.2-3.4m).Notably, no chronostratigraphic age in version was evident.PC1, located at the exposed foredune (Location NW, Profile 1, Figure 4B) recorded the youngest radiometric ( 14 C) ages, ranging between 775 cal yr BP (Depth: 3-3.2 m depth) and 661 cal yr BP (Depth: 1-1.2 m depth).

| DISCUSSION
Five conceptual phases of reef island development have been identified based on a combination of sedimentologi cal, geomorphological and geochronological datasets for the inshore sand cay Eva Island, in the southern Pilbara region of Western Australia, Indo Pacific.

| Phase 1: Scenario of reef accretion of Eva reef flat and context within the Exmouth Gulf, Western Australia
There are limited island reef core data from the Exmouth Gulf that can determine when reef flats/limestone reef platforms became established and thus provide the foun dation for island development.Yet, insights from neigh bouring fringing reefs along the Exmouth Gulf's western coast may be used to infer this record (Figures 5A and 6A).A reef growth study by Twiggs and Collins (2010) off the Exmouth Marina (Figure 1B) shows reef develop ment had initiated in Exmouth Gulf by at least 7,500 cal yr BP, coinciding with the onset of the Holocene high stand, with the reef continuing to aggrade vertically until around 6,500 cal yr BP.Between 6,500 and 5,800 cal yr BP the reefs are postulated to have entered 'give up' mode due to a lack of accommodation space when sea levels stabilised across the mid Holocene highstand and began to fall by the end of this stage (SL at 5,800 cal yr BP = 2 m; Twiggs & Collins, 2010).To place the develop ment of Eva Island's reef into context, comparison has been made with a composite Holocene sea level curve de veloped by Twiggs and Collins (2010).On the limestone reef platform at Eva Island, the oldest coral fossil mi croatoll (Porites sp.) was measured at 5,926 cal yr BP (ca 6,000 cal yr BP; Figure 4A), coinciding with the stage of reef decline or 'give up' on the Exmouth Gulf's west coast (Twiggs & Collins, 2010).The elevation of the Porites microatoll is −0.40 m relative to present MSL (Table 1), which is 1.6-2.1 m above the surveyed height of living Porites from the 'live coral patch reef' at the edge of the limestone platform (mean elevation range sourced from Bonesso et al., 2022).The age determination and eleva tion of the coral microatolls on Eva suggest the limestone reef platform established by at least 6,000 cal yr BP and subsequently became emergent as sea levels fell; therefore, island development on Eva's platform could have initiated by 6,000 cal yr BP.

| Phase 2: Island core initiation and vertical aggradation
The formation of Eva Island as a sand cay had initiated by 4,871 cal yr BP on the eastern side of the limestone reef platform (Figures 4C,5B and 6B).Significantly, Holocene sea level had begun to regress and stabilise to current levels during this phase (Collins et al., 2006;Twiggs & Collins, 2010).Notably, sediments during this preliminary stage of island core formation are characterised by poorly sorted coarse sands rich in mollusc/coral constituents, probably sourced from the surrounding limestone reef platform and delivered to the island core by the focussing effect of local hydrodynamic processes (Gourlay 1988).

| Phase 3: Sustained island accretion and planform adjustment
Sustained accretion between 3,500 and 2,500 cal yr BP allowed lateral progradation of Eva Island, which estab lished much of the island landmass (Figures 4C, 5C and  6C).Across this 1,000 year window (post island core formation) sediment composition of the mid and upper island units remain dominated by mollusc and coral constituents, reflecting no apparent or significant eco logical adjustments as a result of relative sea level fall (i.e. a fall in sea level was not fatal for limestone reef platform biota).After 2,500 cal yr BP, there is evidence that Eva Island probably underwent planform adjust ment, with date inversion at PC5 suggestive of the re working of an existing reservoir of island sediments.According to Liang et al. (2022), date inversion may be indicative of active reworking of island grade material through a combination of alongshore flux and recircula tion, particularly on an island's lateral margins.At Eva Island, this of material may have been caused by local wind wave regimes, whereby prevailing wind generated waves from the south southwest, coupled with potential pulse high energy events (i.e.historical tropical cyclones; Nott, 2011) resulted in shoreline ad justment and recirculation of island sediments across the then present shoreline (Cuttler et al., 2020;Wu et al., 2021a).Alternatively, bioerosion (e.g.sea urchin, serpulids, parrotfish) and mechanical erosion (i.e. in creased wave exposure) of the limestone reef platform (as sea level fell) may have released sediment that was previously (at higher sea levels) incorporated into the reef framework (Cuttler et al., 2019).This older 'rel ict' material could then have been reconstituted and transported towards the beach, deposited on top of con temporary sediments already deposited on the island shoreline, resulting in older radiocarbon ages overlain atop of younger.

| Phase 4: Stable accretion and lateral progradation towards the north
Over a 1,000 year window from 2,000 cal yr BP, Eva Island began to laterally prograde northward, through stable in cremental accretion (Figures 5D and 6D).Sequence stra tigraphy of radiocarbon ages between 1,813 and 1,370 cal yr BP (Accretion event at core PC2; Figures 4C, 5D and  6D) are suggestive of little to no planform adjustment or extensive sediment reworking across this timeframe.In addition, geomorphic stability during this period is rein forced by the formation of beachrock pavement (Liang et al., 2022;Vousdoukas et al., 2007) along the north ern sectors of the island shoreline between 2,500 and 1,000 cal yr BP.A latter sequence of island accretion is characterised by the limited progradation of the island margin to the north northwest between 1,000 and 650 cal yr BP (Accretion event at core PC1; Figures 4B, 5D and  6D).Exposed foredune stratigraphy at core location PC1 at Eva's north west margin, indicate subtle geomorphic change, with the youngest recorded radiometric age dates ranging from 775 to 661 cal yr BP (Figures 4B and 5D).

| Phase 5: Planform adjustment under stabilised sea levels
At a timepoint between 650 cal yr BP and the current day, the north west sector of the island underwent erosional scarping, undercutting the north west foredune, and back stepping to its present position (Figures 5E and 6E;Bonesso et al., 2020).Contemporary erosion is evident via the scarped foredunes with a >30° angle of repose (i.e. an identified ero sional signature; see Bonesso et al., 2020).Yet, a remnant island footprint on the LiDAR DEM; Figure 4A) is visible, suggesting that the foredune was progressively eroded back over time to its present position within the last 650 years.This may have been the result of changes in sediment supply (i.e.sediment deficit) from that side and/or changes in physical forcing (wind/wave regimes) over this climate window (Cuttler et al., 2020;Wu et al., 2021b).

| Eva Island in the context of global reef island evolution models
The model of evolution at Eva Island reports reef island building processes in response to and independent of (i.e.within the last 1,000 years) sea level processes across the mid to late Holocene.Current models of reef island evolu tion from the Indo Pacific have shown varying stages of sea level adjustment over the Holocene.These include reef island formation: during a fall in relative sea level (e.g.Warrabar Island, Torres Strait; Woodroffe et al., 2007), post a fall in relative sea level (e.g.Mba Island, New Caledonia; Yamano et al., 2014), during the Holocene highstand (e.g.Cocos [Keeling] Islands; Woodroffe et al., 1999), dur ing a rise in relative sea level (i.e.Holocene transgres sion; e.g.South Maalhosmadulu Atoll, Maldives; Kench et al., 2005) or during present day sea level (e.g.Tepuka Island, Tuvalu; Kench et al., 2014).The model of evolu tion for Eva Island is unique when compared to its global counterparts in that island development is sustained over both sea level fall (i.e.4,800-1,000 cal yr BP) and during stabilised sea levels (i.e.within the last 1,000 years from which sea levels have shown no further reduction).This is supported by: (1) the heterogenous age structure of is land sediments, suggesting sustained accumulation, with on going production and delivery to the landmass, (2) homogenous composition of sediments (mollusc/coral rich), which may reflect little to no change in historical reef ecology over time (despite sea level fall and reduc tion in accommodation space), thus resulting in sustained carbonate production and supply, and (3) the continual delivery of sediments to the island landmass via possible intermittent high energy pulse events (i.e.tropical cy clone activity), which have been widely reported across the Pilbara region (Nott, 2011).Significantly, under sta ble sea level conditions, there is evidence of planform re adjustment that may be the result of changing metocean conditions (i.e.within climate windows across the mid to late Holocene), which occur independently of sea level fluctuations.These results suggest that Eva Island, and more generally islands within the region, are highly sensitive to changes in metocean conditions, due to role in island accretion.

| Eva Island development in the context of carbonate production within turbid-water reef settings
Previous studies have highlighted the importance of sus tained carbonate sediment supply in the nourishment of low lying reef islands (Perry et al., 2011).Sediment sup ply is determined, in part, by local metocean boundary conditions (e.g.wind/wave climate) but also sediment production rates (i.e.skeletal remains of reef calcifiers and/or reconstituted relict framework material; Browne et al., 2021).Quantifying sediment production in reefal systems is complex, but particularly difficult in turbid water (and remote) systems due to the fieldwork condi tions (Zweifler et al., 2021).While numerous studies have highlighted lower carbonate sediment production rates on turbid water reefs than clear water reefs (Mallela, 2007;Mallela & Perry, 2007), there is also evidence that turbid reefs may exhibit sustained or higher carbonate sediment production, with comparable proportions of key in-situ and direct sediment producers.For example, a study by Browne (2011) on two inshore turbid reefs on the central Great Barrier Reef (Paluma Shoals and Middle Reef) re ported a high proportion of mollusc (ca 25%) and coral (ca 32%) skeletal fragments within the sediment record.Interestingly, the composition of sediment constituents (>60 samples) was relatively uniform both within and between reefs, as such, it was proposed that this mix of constituents could represent a sediment signature for tur bid coral reefs.Here, comparable proportions of molluscs (34%) and corals (27%-31%) were observed in the surfi cial sediments at Eva Island (Bonesso et al., 2022), and in the Holocene sediment record.The temporal stability of sediment constituents suggests that the carbonate fac tory (i.e. the reef) has not experienced any major shifts in reef ecology due to significant environmental changes and that these reefs have been consistently turbid over the course of island building.In addition, recent assessments of in-situ coral carbonate framework production for Eva Island (3.8 kg/m 2 /year; see Dee et al., 2020), using the census based carbonate budget methodology (see review by Browne et al., 2021), are comparable to many clear water reef localities globally, reinforcing that turbid water systems may support stable and productive carbonate fac tories.These results may have significant implications for clear water systems which transition towards more turbid environments, whereby clear water systems with turbid tolerant calcifying communities (both molluscs and corals that are adapted to high sediment loads) may exhibit sus tained island sediment budgets and therefore, heighten the resilience of these systems.

| CONCLUSION
This study highlights a unique record of reef island evo lution within a turbid water coral reef province within the Indo Pacific.Results highlight that island forma tion is linked to incremental sea level fall following the Holocene highstand, but more notably, under stable sea level trajectories, probably attributed to variation and frequency in metocean boundary conditions over the last 2,000 cal yr BP.The heterogenous ages of island sediments along with homogenous sediment composi tion constituting mollusc and coral rich sediments is suggestive of sustained island building processes (pro duction and accumulation) over Eva's accretionary evolution.Yet, these result highlight that reef island evolution is complex and site dependant and may be influenced by local metocean factors (wind, wave, cur rent and high energy event exposure), particularly dur ing stabilised sea level trajectories.Underpinning these temporal and spatial differences in island building is fundamental in evaluating susceptibility to future sea level change and to better inform on future trajectories of reef island landforms under global environmental change scenarios, including declines in water quality conditions.

ACKNO WLE DGE MENTS
This project was funded by an Australian Research Council (ARC) DECRA Fellowship DE180100391 awarded Nicola Browne of Curtin University of Technology, Perth, as part of the island resilience project (2018)(2019)(2020).The authors acknowledge Jarrod Cooper (School of Biological Sciences) and the Minderoo Exmouth Research Lab (MERL) for as sistance in the field within Exmouth Gulf, including boat ing operations, equipment logistics and geological core sampling and recovery.Thank you to the technical staff at the 14 C Chronos Carbon Cycle Facility for the processing of samples for radiocarbon dating, and to Professor Sean Ulm at James Cook University for providing the local res ervoir effect (ΔR) for Exmouth Gulf (personal communi cation).We are grateful to Agisoft LLC for providing the software Agisoft Metashape© that was used to process the drone surveys presented in this study.Likewise, we are thankful to Quality Positioning Services (QPS) for provid ing the software Fledermaus© that was used to visualise the derivative products (Figure 1) from the drone data processing.
using AMS at F I G U R E 1 Location of (A) study site Eva Island (highlighted in blue) and location within the Exmouth Gulf; (B) Exmouth Marina (white star) in the western Exmouth Gulf (C) Exmouth Gulf Region in Western Australia and (D) 3D Digital Elevation Model (DEM) model (top) and drone photograph (below) of Eva Island.Photogrammetry data to create orthophoto mosaics (1.9 cm) and Digital Elevation Model (DEM, 3.3 cm) was collected over Eva Island in November 2021 using a Remotely Piloted Aircraft (RPA, Phantom 4 RTK).Data were processed in Agisoft Metashape© Geomorphic observation during field work at Eva Island showing; (A) aerial view of island landmass; (B) manual sand auguring to extract cores within island basin; (C) exposed beachrock pavement at the northern periphery of island landmass; (D) coring of fossil coral microatoll (Porites sp.) on northern limestone platform of island; (E) scarped foredune (with coastal spinifex grass) and location of core PC1 and (F) aerial view of scarped foredune, beachrock pavement, limestone platform, patch reef and reef crest zones.Aerial drone photography was performed using a DJI Phantom 4 RTK, flying at a maximum altitude of 120 m in October 2021.

F
I G U R E 3 Percentage (%) of reef derived constituents downcore for island cores PC2-PC5.Core locations are shown overlaid on airborne bathymetric LiDAR DEM of Eva Island (0.5 m resolution, derived from Bonesso et al., 2020).T A B L E 1 Summary of island core sediment composition (skeletal) and texture (grain size, sorting) from Eva Island.

F
I G U R E 4 (A) Island core locations (yellow dots), radiocarbon sample sites (black dots), and region of remnant island footprint as shown on airborne bathymetric LiDAR DEM of Eva Island (0.5 m resolution, derived from Bonesso et al., 2020); (B) island cross section along Profile 1 (P1) showing radiocarbon sample ages (black dots) for PC1 core on NW periphery of island landmass and (C) island cross section along Profile 2 (P2) showing radiocarbon sample ages (black dots) for cores PC2-PC5.

F
Cross section schematic model for the accretion of Eva Island, Pilbara region of Western Australia (Indo Pacific) over the past 6,500 cal yr BP.

FIGURE 6
FIGURE 6 Aerial schematic model for the accretion of Eva Island, Pilbara region of Western Australia (Indo Pacific) over the past 6,500 cal yr BP.

Core ID Skeletal Composition (%) Sediment Texture Coral Mollusc CCA Foraminifera Grain size range (μm) Average grain size (μm) Sorting range (μm) Average sorting (μm)
Note: CCA is synonymous for Crustose Coraline Algae.Average composition (% ± SD) of reefal constituents are reported alongside both mean grain size (μm ± SD) and sediment sorting values (μm ± SD).Range of grain size and sorting are denoted as the minimum and maximum values across each core (PC2-PC5).
Summary of sample information and radiocarbon ( 14 C) ages including the conventional age, 1σ range of calibrated age and calibrated age with median probability.