Introduction

The physical structure of a coral reef consists of calcium carbonate deposited by scleractinian corals, calcifying algae and other encrusting organisms. These sources of skeletal carbonate make major contributions to reef-building, but the net carbonate budget of a reef is additionally influenced by erosional processes, including physical, chemical and biological erosion (Perry et al. 2012). Biological erosion (or bioerosion) is typically the dominant erosional process in coral reefs and is defined as the destruction and removal of deposited calcium carbonate by the direct action of an organism (Neumann 1966). In most reef settings, external bioerosion driven by grazing parrotfish and sea urchins has been identified as being of major importance (Bellwood et al. 2003; Chazottes et al. 2002; Perry et al. 2015), but endolithic bioerosion can also be quantitatively important in some locations (Tribollet and Golubic 2005). Endolithic bioerosion occurs when substrate is removed from the inside of coral skeletons both mechanically and/or chemically by a variety of i) macroboring taxa, including eroding species of worms, sponges and bivalves; and ii) microboring organisms such as unicellular algae, fungi and cyanobacteria (Hutchings 1986). These taxa infest both live (Le Campion-Alsumard et al. 1995; Holmes et al. 2000) and dead coral substrate (Tribollet et al. 2002), resulting in substrate degradation (Le Campion-Alsumard et al. 1995), the release of sedimentary carbonate (specifically from endolithic sponges) (Neumann 1966; Rützler and Rieger 1973; Carballo et al. 2017), and increased susceptibility to physical fragmentation (Scott and Risk 1988).

Endolithic bioerosion processes are thus of significant relevance on coral reefs, but the abundance and activity of endolithic bioeroding taxa, and the resulting rates of erosion, can vary significantly over temporal and spatial scales (Silbiger et al. 2016, 2017; Tribollet and Golubic 2005). Endolithic erosion intensifies in dead coral substrates, as more organisms colonise the substrate and the macro- and microendolithic communities develop and mature (Davies and Hutchings 1983; Le Campion-Alsumard et al. 1995). Many endolithic macroborers are filter feeders (Smith et al. 1981), and eutrophication and high water turbidity have been shown to increase substrate infestation rates through enhanced food availability (Chazottes et al. 2002; Silbiger et al. 2014, 2017). However, high sedimentation can have negative impacts on some macrobioeroders through smothering and shading (Macdonald and Perry 2003). Light levels are a major control on the abundance and composition of microendolithic communities (Gektidis 1999; Perry and Macdonald 2002), as many euendolithic species are phototrophic. Simulated ocean acidification scenarios have also been shown to accelerate both macrobioerosion (DeCarlo et al. 2015) and microbioerosion (Reyes-Nivia et al. 2013) through a weakening of the carbonate substrate. These interacting environmental controls result in marked differences in endolithic erosion rates between locations, but a more complete understanding of the causes and implications is hampered by the relative paucity of quantitative data on erosion rates and the taxa driving erosion.

Here we contribute to this increasingly important topic through the quantification of bioerosion rates, and the contributing taxa, in the remote Chagos Archipelago, Central Indian Ocean. The remoteness and protected status of the archipelago mean that the confounding effects of direct human disturbance to reefs, including eutrophication, can largely be excluded (Hays et al. 2020). Based on experimental deployments of carbonate blocks across two different depths, we provide the first macro- and microendolithic bioerosion rate data for the region. We also assess the relative contributions of different organisms (macro-, micro-, and external bioeroders) to total bioerosion rates, data which enhances not only our understanding of variations in bioerosion across locations but can also improve local reef carbonate budget estimates.

Study area and methods

To quantify endolithic bioerosion rates, we deployed experimental carbonate blocks on stainless-steel racks at depths of 5 m and 10 m (n = 7 per depth and exposure period) on the fore-reef of Ile Poule, Peros Banhos Atoll, in the Chagos Archipelago, Central Indian Ocean in May 2018 (Fig. 1a). The northern atolls of the archipelago have been uninhabited since the 1970s, reducing direct human disturbance to reefs to a minimum (Hays et al. 2020). Ile Poule is a low-lying sand island and its fore-reef slopes gently towards a terrace at ~ 10 m depth, before dropping off steeply at a depth of ~ 12 m. Our experiments were thus confined to the upper fore-reef slope habitats with differences between the 5 and 10 m depth sites (apart from depth) largely being evident ecologically as a subtle transition towards a higher prevalence of more tabular Acropora colonies at 10 m depth. A distance of about 50 m horizontally separated deployment sites. Blocks (approximately 9 × 5 × 4 cm) had been previously cut from recently deceased Porites lutea colonies collected from nearby shallow fore-reef locations in the same atoll, which experience the same environmental conditions. Only blocks showing no visible signs of internal erosion were utilised. Carbonate blocks were recovered after ~ 1 year (300 days) or ~ 3 years (1084 days) of exposure (Fig. 1b). After recovery, blocks were photographed, soaked in diluted bleach overnight to neutralise organic components, dried and prepared for macro- and microbioerosion analysis.

Fig. 1
figure 1

a Geographical map of the Chagos Archipelago in the Central Indian Ocean and location of study sites (5 m and 10 m water depth) on the fore-reef of Ile Poule in Peros Banhos Atoll. Satellite image of Ile Poule derived from Google Earth Pro V 7.3.2.5776 (March 5, 2023); Image: 2023 Maxar Technologies. b Experimental carbonate blocks deployed on a stainless-steel rack and examples collected after one or three years of exposure. Note higher crustose coralline algal cover on 5 m blocks and extensive external erosion on blocks deployed for three years

Full size image

Quantifying macrobioerosion rates

Macroendolithic erosion rates were quantified using non-destructive computer tomography (CT) scanning of experimental blocks, with preparation following the approach of Silbiger et al. (2014). CT scanning images the internal and external structures of solid objects, resulting in a 3D array of object densities. We scanned three or four experimental blocks from each depth and time period (total n = 13), avoiding blocks that had large accretions of molluscs or coral colonies on the surface. Scans were taken with a Nikon XTH 225ST (voltage = 160 kV, current = 230 µA) at 0.5° increments, the voxel resolution being 50 µm3. Scans were then exported to the open-source image processing software ImageJ (Abramoff et al. 2004), where image histograms were equalised to enhance contrast prior to analysis. The Trainable Weka Segmentation (TWS) plugin (Arganda-Carreras et al. 2017) was then used to quantify the volume of bioerosion within each block by segmenting and summing the area of void space/low density areas (borings) in each slice and multiplying by slice depth (50 µm) (Fig. 2a). Macrobioerosion rates were then calculated following Silbiger et al. (2014) using Eq. 1:

$${\text{Macrobioerosion}}\;{\text{rate~~}}\left( {{\text{kg~m}}^{{ - 2{\text{~}}}} \;{\text{yr}}^{{ - 1}} } \right) = \frac{{{\text{Vol }}~ \times ~\rho _{{{\text{bulk}}}} ~}}{{{\text{SA }}~ \times \;{\text{Time}}}}$$
(1)

where:

Fig. 2
figure 2

a Segmented CT scan of a carbonate block deployed at 10 m depth for three years showing macroborings visualised in 3D—the view is top-down onto the upwards facing block surface. Worm borings are visible throughout the block (red arrow), mollusc borings in the lower right corner of the image (yellow arrow) and sponge borings in the upper block region (orange arrow). b SEM image (magnification of 2000 x) of the exposed top surface of a carbonate block deployed at 10 m depth for one year, showing many small microborings, and few larger microborings. c SEM image of a resin embedded sample (side view at 450 x magnification) from a block deployed at 5 m depth for one year, revealing the resin cast boring traces of euendoliths

Full size image

Vol = Volume lost through internal bioerosion (cm3), ρbulk = Skeletal bulk density of blocks, determined using the Archimedes principle (Bucher et al. 1998) on wax-dipped substrate fragments = 1.63 ± 0.11 g cm−3, SA = Surface area of all six exposed sides of the block minus area of screw holes (cm2), and Time = Deployment time (d)/365.

In addition to the quantification of total macrobioerosion rates, we also calculated the relative contribution of the three dominant taxa (worms, sponges, molluscs) based on borehole morphologies (Glynn and Manzello 2015). Ten equidistant virtual slices along the length of each block were isolated from the segmented image stacks produced by TWS, and erosional traces were attributed to one of the three macroborer groups (Fig. 2a) based on visual guides in Glynn and Manzello (2015).

Quantifying microbioerosion rates

For the analysis of microbioerosion rates, two 1 cm3 cubes were cut from the upper facing central surface of blocks that had been deployed for one year (n = 5 blocks per depth). Three-year blocks were not analysed for microbioerosion as the surface was covered in thick crustose coralline algal (CCA) crusts (up to 2.8 mm thick) which prohibited comprehensive imaging of surface borings. However, small differences in microendolithic erosion rates over time were expected because previous studies have shown microendolithic communities to reach maturity within one year of exposure (Chazottes et al. 1995; Le Campion-Alsumard et al. 1995; Gektidis 1999; Tribollet and Golubic 2005). Cubes were soaked in diluted hydrogen peroxide (~5%) for 24 h and lightly brushed to remove any surficial filaments prior to analysis. One randomly selected cube from each block was used to quantify the surface area removed by microborers. Surfaces were imaged using a TESCAN VEGA3 scanning electron microscope (SEM) at a magnification of 2000 x, with the view field set to 104 µm (15 images per sample) (Fig. 2b). The surface area removed was calculated by counting all boreholes in each image and multiplying the number in each of three measured size classes (1–2 μm, 5–8 μm, 10–15 μm diameter) by the respective mean cross section of borings. The second cube was used to quantify microborer penetration depth and to allow identification of microendolith trace types based on resin embedding methodologies (Golubic et al. 1970; Wisshak 2012). Embedding in low viscosity resin (Taab TRANSMIT LM) took place under controlled vacuum for ~ 3 h. Cubes were then cured in the oven (20 h at 70 °C) before one side surface was polished on a grinding plate (succession to 1200 grit paper) until the carbonate was exposed. Resin cubes were then etched for 20 s in 10% HCl, rinsed and again dried in the oven at 30 °C. SEM images (magnification of 150 x) were taken from the exposed surface down to the limit of microendolith penetration (Fig. 2c), and seven depth measurements were taken per block at equidistance along the exposed surface to calculate the average depth of penetration per endolithic trace-type size class (Carreiro-Silva et al. 2009). The identification of the microbial organisms and their boring traces followed the descriptions of Le Campion-Alsumard (1979), Radtke (1993), Vogel et al. (2000), Radtke and Golubic (2005) and Wisshak et al. (2005) and are discussed in detail in the results section.

The rate of microbioerosion for each endolithic trace type size class was determined following Tribollet (2008a) (Eq. 2) and then summed to derive a total.

$${\text{Microbioerosion~rate~~}}\left( {{\text{kg~m}}^{{ - 2{\text{~}}}} {\text{ yr}}^{{ - 1}} } \right) = ~\frac{{S\; \times \;D{\text{p }} \times \rho _{{{\text{micro}}}} \times 10^{{ - 3}} }}{{{\text{SA}}\; \times \;{\text{Time}}}}$$
(2)

where:

S = Surface area removed (cm2), Dp = Mean depth of penetration (cm), ρmicro = Skeletal microdensity of blocks, determined using the Archimedes principle (Bucher et al. 1998) on clean (as opposed to wax-dipped) substrate fragments = 2.80 ± 0.03 g cm−3 (note: microdensity is used here as euendolithic filaments bore into the microstructure of the skeleton, not in void spaces), SA = Total surface area of substrate examined in SEM images (m2), and Time = Deployment time (d)/365.

Quantifying external erosion and accretion rates

External erosion rates (n = 13) were calculated as the difference in block volume before (calculated from linear measurements of block length, width and height) and after deployment (total CT scan block volume minus accretion volume), multiplied by the bulk density of coral blocks (1.63 ± 0.11 g cm−3) (Enochs et al. 2016). Measurements pre-deployment were conducted in the centre of blocks, but as not all blocks were completely rectangular, external erosion rates have higher uncertainties than macro- and microendolithic erosion rates. External volume loss was normalised to only five surface areas (top and sides), as the lower surface was considered inaccessible to larger parrotfish which are the main agents of external bioerosion in the Chagos Archipelago (Lange et al. 2022).

Accretion on blocks (n = 13) (mostly CCA but also accreting coral recruits and worms) was quantified semi-manually using the Segmentation Editor plugin in ImageJ. In every 20th slice through each image stack, CCA crusts or accreted organisms were selected using the ‘Selection Brush’ tool and then thresholded from the background of the scans, with the 'Interpolate’ feature used to interpolate the selected features to consecutive slices. The number of voxels identified as accretion was multiplied by voxel size (50 µm3) to yield the volume of accretion on each block, and then multiplied by the bulk density of local CCA crust fragments (1.44 ± 0.08 g cm−3), determined using the Archimedes principle (Bucher et al. 1998).

Statistical analysis

Statistical analyses were carried out using the software PAST version 4.11 (Hammer et al. 2001) and R v4.3.0 (R Core Team 2021). A series of two-way analysis of variance (ANOVA) was conducted to investigate the effects of exposure time and depth on rates of bioerosion (macrobioerosion and external erosion), accretion, and relative contributions by macroborer taxa. As microbioerosion data were only collected for one-year samples, a one-way ANOVA was carried out to investigate the effect of depth on microbioerosion rates. Prior to these analyses, the data were tested for the basic assumptions of homogeneity of variance and normality and were log transformed if necessary. Spearman’s correlation analyses were conducted to explore relationships between macrobioerosion and external erosion, macrobioerosion and accretion, and between external erosion and accretion after one and three years of exposure, following Tribollet et al. (2002). Microbioerosion could not be correlated with other variables as SEM analysis was not always conducted on the same blocks as CT analysis (due to thick CCA crusts on some blocks which inhibited imaging of surface borings), and averages across sites and depth did not yield a sufficient number of data points for correlation analyses.

Results

Bioerosion rates and the contribution of different groups at both depths and over time are visualised in Fig. 3 and detailed in Table 1. Macrobioerosion rates varied between 0.051 ± 0.003 and 0.086 ± 0.026 kg m−2 yr−1 (Fig. 3a, Table 1), with depth and exposure time having no significant effects on macrobioerosion rates (Table 2). Worm % contribution to total macrobioerosion was significantly higher at 10 m depth (p = 0.03), and sponge % contribution was significantly higher (p < 0.01) at 5 m depth (Table 2). Mollusc % contribution did not significantly differ across depth and there was no significant change with exposure time for any of the macroborer contributions (Table 2).

Fig. 3
figure 3

a Mean (± SE) macrobioerosion rates (kg m−2 yr−1) at 5 m and 10 m depth after one and three years of exposure, with colours indicating the contribution by different borer groups. b Mean (± SE) microbioerosion rates (kg m−2 yr−1) at 5 m and 10 m depth after one year of exposure. The presence of thick CCA crusts prohibited the quantification of microbioerosion rates after three years of exposure. c External bioerosion rates (kg m−2 yr−1), calculated as the difference between pre-deployment and post-deployment block dimensions. d Accretion rates (kg m−2 yr−1) of CCA crusts and other calcifying organisms on block surfaces. Please note the use of different scales on y axes of figures

Full size image
Table 1 Bioerosion rates in experimental blocks quantified using CT analysis (macrobioerosion, external erosion, accretion) and SEM analysis (microbioerosion) and relative contribution of taxa to macrobioerosion
Full size table
Table 2 Results of the two-way ANOVAs comparing rates of macrobioerosion, microbioerosion, external erosion, accretion and % contribution of the macroboring groups across the different depths and exposure times
Full size table

Microbioerosion rates after one year of exposure varied between 0.187 ± 0.028 at 5 m and 0.313 ± 0.049 kg m−2 yr−1 at 10 m depth (Fig. 3b), though differences were not statistically significant (p = 0.09). The main microendolithic trace types found in samples were Scolecia filosa, characterised by thin, long, curving filaments of 1–2 µm diameter (Fig. 4a), and Eurygonum nodosum, characterised by filaments of 5–8 µm in diameter with short lateral branches and characteristic lateral swellings that contain nitrogen fixing heterocysts (Fig. 4b). The trace type Ichnoreticulina elegans, characterised by dichotomous branching filaments of 5–8 µm in diameter (Fig. 4c), was also common in samples, although less abundant than S. filosa and E. nodosum. Rare traces of Fascichnus frutex were found in samples from both depths (Fig. 4d), these are differentiated by their shorter galleries with larger diameters of 8–15 µm and with frequent bifurcations and rounded terminations. S. filosa, E. nodosum, F. frutex and I. elegans are the trace types of the cyanobacteria Plectonema terebrans (Radtke 1991), Mastigocoleus testarum (Schmidt 1992), Hyella gigas (Radtke 1991) and the chlorophyte Ostreobium quekettii (Radtke and Golubic 2005), respectively.

Fig. 4
figure 4

a SEM image (2000 × magnification) of the boring trace type Scolecia filosa, characterised by thin, curving filaments (1–2 µm in diameter). b SEM image (1000 × magnification) of the trace type Eurygonum nodosum, characterised by filaments of 5–8 µm in diameter, with short lateral branches and characteristic heterocysts (white arrows). c SEM image (1000 × magnification) of the trace type Ichnoreticulina elegans, characterised by dichotomous branching filaments 5–8 µm in diameter. d SEM image (1000 × magnification) of the trace type Fascichnus frutex, characterised by filaments 10–15 µm in diameter with frequent bifurcations

Full size image

External erosion rates varied between 0.468 ± 0.105 at 10 m after one year of exposure and 1.117 ± 0.163 kg m−2 yr−1 at 10 m after three years of exposure (Fig. 3c). External erosion rates increased significantly over time (p = 0.02), with depth having no significant effect on external erosion rates (Table 2). External erosion was mainly caused by grazing parrotfish, as visible from frequent bite scars on the substrate (Fig. 1b). Accretion rates on the block surface varied between 0.028 ± 0.014 at 10 m after one year of exposure and 0.381 ± 0.027 kg m−2 yr−1 at 5 m after one year of exposure (Fig. 3d). Accretion rates were significantly higher at 5 m compared to 10 m depth (p = 0.001) but did not vary significantly with exposure time (p = 0.06) (Table 2).

The correlation analyses between macrobioerosion and external erosion, macrobioerosion and accretion, and external erosion and accretion after one year of exposure did not yield any significant correlations (Table 3). After three years of exposure, external erosion was negatively correlated with accretion rate (p = 0.003).

Table 3 Results of Spearman’s correlation analyses between macrobioerosion—external erosion, macrobioerosion—accretion and external erosion—accretion in each block after one or three years of exposure
Full size table

Discussion

This study provides the first macro- and microendolithic bioerosion rate data for the central Indian Ocean. By assessing the contributions of different endolithic taxa, and the impacts of time and water depth on bioerosion rates, we contribute to a better understanding of this poorly quantified but critically important reef carbonate cycling process (Brandl et al. 2019). Rates of macrobioerosion in the Chagos Archipelago ranged from 0.066 to 0.086 kg m−2 yr−1, with evidence of a community succession towards sponge dominance after three years. These rates are both above and below those reported from other Indian Ocean reefs, probably reflecting variations in key marine environmental parameters including nutrients and turbidity regimes. For example, our calculated rates are higher than those reported in La Reunion across a nutrient gradient (0.020 ± 0.007 kg m−2 yr−1; Chazottes et al. 2002) but are lower than those at inshore sites in Exmouth Gulf, Western Australia, which are characterised by frequent high turbidity events (0.091 ± 0.01 kg m−2 yr−1; Dee et al. 2023). Interestingly, our rates are very similar to those reported from a comparable open ocean reef setting on the Great Barrier Reef (0.070 ± 0.017 kg m−2 yr−1; Tribollet and Golubic 2005) and across four low nutrient sites in Palau (0.076 ± 0.039 kg m−2 yr−1; DeCarlo et al. 2015). Tribollet and Golubic (2005), reported a clear decrease in macrobioerosion rates with distance from shore, indicating that eutrophic conditions are favourable for macroborers, as many species are filter or detritus feeders (Chazottes et al. 2002). Additionally, much higher rates were quantified across the Indo-Pacific at high nutrient or very shallow coastal sites (0.229 ± 0.024 to 0.632 ± 0.132 kg m−2 yr−1) (Carreiro-Silva and McClanahan 2012; DeCarlo et al. 2015; Prouty et al. 2017).

Differences in macrobioerosion rates between locations may also arise from variability in community succession. Newly exposed coral substrate is known to be quickly colonised by pioneer borers, such as polychaete and sipunculid worms, whilst larger borers, such as sponges and bivalves, are typically slower to recruit (Kiene and Hutchings 1994; Chazottes et al. 2002; Tribollet and Golubic 2005). Ours is the first study in the Indian Ocean to report sponge borings in deployed substrates after only one year of exposure (Chazottes et al. 2002; Dee et al. 2023), and indeed we observed evidence of sponge erosion in every block examined. This contrasts with the observations of Chazottes et al. (2002) and Dee et al. (2023) who found no evidence of sponge and very few bivalve borings in Porites blocks deployed for one year. Boring sponges were however reported at one inshore site on the Great Barrier Reef after one year of exposure (Tribollet and Golubic 2005), although the reported contribution of 22% was relatively low compared to our study (Table 1). We do not know the reason for the rapid sponge infestation in the Chagos Archipelago, as sponge cover in the surrounding reefs is generally low (few %; Lange et al. 2022). In general, rates of sponge macroboring after three years are not unusually high compared to other studies (Tribollet and Golubic 2005). Whilst local environmental parameters certainly act as controls on rates of substrate infestation and erosion, an additional influence arises from externally driven erosion pressure. For example, Chazottes et al. (2002) reported fourfold higher external erosion rates at their study site (2.588 ± 0.544 kg m−2 yr−1) than that observed in the Chagos Archipelago (Table 1), and it is reasonable to hypothesise that this intense grazing pressure removed large amounts of the reef substrate and thereby caused lower reported rates of endolithic erosion. Endolithic bioerosion rates quantified in situ therefore always represent ‘residual rates’, because the amount of macro- and microbioerosion in the substrate removed by external erosion cannot be quantified (Chazottes et al. 1995). In support of this, Dee et al. (2023) found much lower external erosion rates (0.025 ± 0.01 kg m−2 yr−1) than those reported here, possibly explaining higher residual macrobioerosion rates. These observations point to important interacting dynamics between site-specific external eroder pressure and rates of macrobioerosion, although it has to be noted that at the local scale our macrobioerosion rates were not correlated with external erosion rates.

Rates of microbioerosion in the Chagos Archipelago ranged from 0.187 to 0.313 kg m−2 yr−1, and the community was dominated by cyanobacteria at both depths. Similar to macrobioerosion, these rates are greater than those found by Chazottes et al. (2002) in La Reunion (0.056 ± 0.01 kg m−2 yr−1), and as noted above this may reflect low residual endolithic erosion rates due to high external bioerosion pressure. Mwachireya et al. (2018) also found low microbioerosion rates in Tridacna shells at a pristine control reef on the Kenyan coast (0.069 ± 0.009 kg m−2 yr−1). However, this may reflect the short length of exposure (60 days), as samples should be exposed for at least one year to assess differences in microbioerosion intensity across sites (Tribollet 2008a). Chazottes et al. (1995) suggest that the highest rates of microbioerosion occur within the first two months of exposure, whilst Grange et al. (2015) found increasing microbioerosion rates within the first six months of exposure, followed by a stabilisation of rates.

The dominant microborers found in our study, Plectonema terebrans, Mastigocoleus testarum and Ostreobium quekettii, are consistent with those found by Chazottes et al. (2002), who additionally report some species of boring fungi, but no other taxa. In contrast, Mwachireya et al. (2018) found the boring traces of 14 different microborers (seven cyanobacteria species, three chlorophytes and four fungi) in Tridacna shell fragments exposed for just 60 days, although P. terebrans, M. testarum and O. quekettii were again the most common, with rare traces of H. gigas also being found. The cyanobacteria M. testarum and P. terebrans have previously been identified as pioneering euendoliths (Chazottes et al. 1995; Le Campion-Alsumard 1975; Le Campion-Alsumard et al. 1995), while the chlorophyte O. quekettii takes longer to establish but typically dominates the community thereafter (Grange et al. 2015; Tribollet 2008a). Although O. quekettii traces did not dominate microendolithic communities in our samples, their presence alongside P. terebrans suggests that the communities may be considered ‘mature’ after one year (Gektidis 1999), and had likely reached a stable community state. Annual rates of microbioerosion however might change over time and we regret that the analysis of three-year samples was not possible due to thick CCA crusts on most blocks. Microbioerosion rates may either increase due to accelerating infestation of substrate or decrease due to limited availability of substrate or increasing grazing pressure (Tribollet and Golubic 2005).

As previously discussed, intense grazing may cause low residual rates of endolithic bioerosion (Chazottes et al. 1995), which can lead to a negative relationship between microbioerosion and external erosion. In La Reunion, grazing was dominated by echinoids and correlated negatively with microbioerosion rates (Chazottes et al. 2002). On the other hand, positive correlations between microbioerosion rates and grazing rates have also been demonstrated (Chazottes et al. 1995; Grange et al. 2015; Tribollet and Golubic 2005). This makes sense as euendoliths are a food source for grazers (Clements et al. 2017) and in turn the removal of surface substrate can extend the depth of light penetration and consequently the depth of euendolithic infestation (Schneider and Torunski 1983). At our study site, sea urchins are rare, and none were found during ecological surveys in 2015, 2018 and 2021 (Lange et al. 2022). External erosion was instead caused by parrotfish, as visible from bite scars on the block surfaces (Fig. 1b). Interestingly, erosion rates calculated from the blocks were 4-times lower than those estimated from parrotfish abundance at the study site in 2021 (3.95 ± 1.75 kg m−2 yr−1; Lange et al. 2022). This indicates that the deployed substrate did not (yet) constitute the preferred food source for parrotfishes, which target cyanobacteria and other protein-rich microalgae within the surface layer of reef substrate (Clements et al. 2017). It is however reasonable to hypothesise that the nutritional value of experimental blocks would increase over time as the microendolith community matures, which is supported by the significant increase in external erosion rates over time in our study.

Similar to microbioerosion, both positive and negative relationships between macrobioerosion and grazing have been found previously. Tribollet and Golubic (2005) found a negative correlation between grazing and macrobioerosion across the GBR shelf, with very low grazing and high macrobioerosion rates inshore, but high grazing and low macrobioerosion rates at offshore oligotrophic reefs. The low grazing rates found at the inshore reefs are a result of lower parrotfish abundance at turbid inshore sites (Cheal et al. 2013) and likely coincide with favourable conditions for macroborers, while high grazing pressure at the offshore reefs result in low residual rates of macrobioerosion. This negative relationship has also been documented at a local scale by Sammarco et al. (1987) and Kiene and Hutchings (1994), where high levels of grazing prevented the settlement of macroboring larvae. While Chazottes et al. (2002) also found a negative correlation under high grazing pressure, moderate grazing pressure facilitated the accretion of CCA and settlement of macroalgae on blocks, providing protection for the recruitment of macroborers and leading to a significant positive relationship between macrobioerosion and external erosion rates. Contrary to studies mentioned above, no significant correlations were found between grazing and macrobioerosion in the Chagos Archipelago, albeit this may partly be due to the relatively low sample size. However, external erosion rates were negatively correlated with accretion rates after three years of exposure. This supports the findings of Chazottes et al. (2002) and is likely caused by the thick layer of calcifying algae, which are less palatable than turf algae for grazers (Bruggemann et al. 1994) and can also limit the colonisation by euendoliths (Tribollet 2008b).

Summary

This study provides the first rates of endolithic bioerosion for reefs in the remote Chagos Archipelago, data that is critical for quantifying local carbonate budget estimates and helps understanding of abiotic and biotic controls on reef accretion. External bioerosion was the dominant erosion process on deployed carbonate blocks and is likely to be even higher on more degraded reef substrate. Endolithic erosion rates are consistent with the limited data available from other open ocean oligotrophic settings, although sponges and molluscs have seemingly recruited faster than at other sites within the Indian Ocean. Microbioerosion was 2–5 times higher than macrobioerosion and was dominated by cyanobacteria. None of the bioerosion rates were significantly different across depths, but accretion of crustose coralline algae was higher at 5 m compared to 10 m depth. Further study is required to assess additional fine-scale environmental drivers of macro- and microbioerosion in the region, so that spatial variability can be better understood, and estimates of local carbonate budgets can be improved.