For both tide gauges, we also used the Unified Tidal Analysis and Prediction (UTide) models51 to estimate MLWS and MLWN relative to lowest astronomical tide (LAT) for the 18.61-year period between 2004 and 2023. We build on earlier work at Siloso Point8 by re-calculating tidal datums using a longer water-level time series obtained from January 2021 to June 2024.
The mean 2023 HLG for the Biola reef unponded microatolls was 44.3 cm (± 2.7 cm, 1 s.d. of repeat measurements of the tide-gauge benchmark) above admiralty chart datum (ACD), equal to 11 cm above MLWS at the Raffles Lighthouse tide gauge (Fig. 3; Supplementary Table S2). The 2023 HLG for these corals is below the midpoint of MLWS and MLWN, which has been assumed to be the upper limit of unponded microatoll growth. The HLG of the unponded corals at Pulau Biola was, as expected, lower than the HLG of ponded Porites microatolls. The average elevation of the highest ponded microatolls' living rims (and dead microatoll centre) is 79 cm above ACD, over 20 cm above the highest unponded surveyed microatoll (Fig. 3; Supplementary Tables S2,S3).
The weighted mean HLG of all unponded living Porites microatolls at Siloso Point is 1.42 ± 0.04 m below Singapore Height Datum (SHD; uncertainty is 2 standard errors of the weighted mean of 24 microatolls). MLWS at the Siloso Point site is 1.34 m below SHD based upon the Siloso tide-gauge data from January 2021 to June 2024. Therefore, the weighted mean of the 2021 HLG of the Siloso Point reef microatolls was 8 cm below MLWS.
BIOL-L13 (the highest-elevation unponded coral surveyed on the intertidal reef flat) had a raised outer rim that was bleached but apparently alive in May 2023 (Fig. 4a). However, by August 2023, the top of the outer rim was dead and covered by sediment (Fig. 4b). Therefore, we infer that BIOL-L13 died down at some point between May and August 2023. During our surveys of July and August 2023, we also observed apparent recent diedowns on other Porites spp. microatolls (BIOL-L13a, BIOL-L38 and BIOL-L56), though we do not have close-up photographs of these corals taken prior to the partial mortality event. BIOL-L13a had a diedown of similar magnitude to that seen on BIOL-L13 (Fig. 4e) but BIOL-L56 and BIOL-L38 had patches of recently dead coral only on the tops of their outer rings (Fig. 4f-h). The living rims of BIOL-L13a, BIOL-L56 and BIOL-L38 had not grown outwards beyond the adjacent dead coral by more than a few millimetres in August 2023. Since reported growth rates of Indo-Pacific Porites corals span ≈ 0.5 to 3 cm yr, we infer that the partial mortality of BIOL-L13a, -L56, -L38 probably occurred within the preceding weeks or months and is likely coeval with the diedown on BIOL-L13.
We measured BIOL-L13's pre-diedown HLG and post-diedown HLS on 15 swath profiles extracted from the DSM (Fig. 5). The DSM was created from a LiDAR scan taken on 5th August 2023, shortly after the mid-2023 diedown. The difference between BIOL-L13's pre-diedown HLG and post-diedown HLS (on 5th August) was 1.4 ± 0.7 cm (1σ) and was largest between the north and west faces of the microatoll, where a maximum difference of 2.6 cm was measured (Fig. 5c). On 5th August 2023, we also used the total station to survey eight paired measurements of HLG and HLS at the approximate positions of the N, NE, E, SE, S, SW, W and NW swath profiles. The HLG-HLS difference from the field survey is 1.2 ± 0.3 cm (1σ).
We did not observe any diedowns between August 2023 and May 2024. The upper surfaces of BIOL-L13 and BIOL-L13a's outer rims were dead in August 2024 (Fig. 4c,d; Supplementary Fig. S1), so we infer that there was an additional, smaller diedown on BIOL-L13 and -L13a between May and August 2024.
While BIOL-L13 was bleached in the month(s) before the diedowns (April 2023 and May 2024), bleaching was also present on many other Porites spp. corals that were exposed at low tide, including those that survived without partial mortality (Fig. 4i,j, and Supplementary Fig. S1). Therefore, bleaching does not necessarily presage diedowns on the Biola reef microatolls.
We combine water levels and coral elevations to determine the threshold exposure duration for diedowns. From the Pulau Biola reef, we use BIOL-L14, BIOL-L3 and BIOL-L13 as these corals have been continuously monitored since 2022 and therefore have the best constrained growth history of any unponded microatolls at this site.
BIOL-L14 has several concentric rings on its dead upper surface, which suggests this coral's morphology has developed in response to multiple short-lived diedown events (Fig. 4i). However, despite being fully exposed during several field visits, this coral did not have a diedown between June 2022 (when first photographed) and August 2023 (when the HLG was surveyed, and the coral was scanned using LiDAR). In June 2022, the microatoll's living outer ring was approximately 1 cm thick, so, given a possible growth rate between ≈ 0.5 and 3 cm yr, BIOL-L14's most recent diedown may have occurred during 2021.
While HLS can be precisely measured from radial slabs of coral using annual density or luminescence banding, we did not have permission to sample live coral for this study. Therefore, we estimate the post-diedown HLS using non-destructive LiDAR scans. To estimate the HLS after this most-recent diedown, we measured ring thickness from 14 swath profiles extracted from the LiDAR-derived DSM at angles perpendicular to the microatoll perimeter (Supplementary Fig. S9). We assumed ring boundaries beneath the microatoll surface are parallel to the outer living rim and that vertical and lateral growth rates are equal. Mean ring thickness across all profiles is 2.5 ± 0.64 cm, and HLG in August 2023 was 41.0 ± 1.53 cm above ACD. Therefore, we estimate the HLS after BIOL-L14's most recent diedown to be 38.5 ± 1.66 cm above ACD.
We assume water levels recorded at the nearby Raffles Lighthouse tide gauge (Fig. 2c) are representative of the tides and sea-level anomalies on the Biola reef. Under this assumption, BIOL-L14 would have been fully emersed for at least an hour on 19 days in 2021 when the recorded water level was lower than the base of the coral for a continuous 1-hour period (Fig. 6c). While we did not monitor the Biola reef microatolls in 2021 (the most recent year of our tide-gauge data), full subaerial exposure of intertidal corals was recorded on Singapore's other reefs between April and August 2021, so emersion of the Biola reef microatolls during these spring low tides seems reasonable based upon observations elsewhere in the region. The tide-gauge data also suggests BIOL-L14 would have been fully exposed for at least two continuous hours during four days between May and July when water levels were lower than the reef substrate surrounding the coral (Fig. 6d). Since at least part of BIOL-L14 must have remained alive throughout 2021, we therefore infer that BIOL-L14 can survive more than 2 h out of water.
BIOL-L14 would not have been fully subaerially exposed for three or more consecutive hours per day during 2021, but this coral would have been partially exposed for a three-hour duration on several occasions. The estimated HLS of BIOL-L14's most recent diedown coincides with the lowest elevation exposed for 3 h per day between 28th and 30th May 2021 (Fig. 6e). If the most recent diedown happened in 2021, at least 3 h of continuous subaerial exposure above the estimated HLS (≈ 39 cm ACD) may have been sufficient to kill the part of the coral's outer rim that was exposed above the water for this time. Coral polyps below ≈ 39 cm ACD must have survived throughout 2022, and coral polyps in this lower region of BIOL-L14's outer perimeter would have been exposed for less than 3 h per day in 2022.
BIOL-L3, another microatoll first observed in 2022 and with no diedowns from April 2022 to August 2023, would also have been partially exposed for 3 h down to the elevation, within error, of its most-recent HLS in June 2021 (Supplementary Text 3; Supplementary Figs. S10, S14).
BIOL-L13's most recent pre-2023 diedown (based upon ring thickness estimated from the DSM; Supplementary Figs. S11, S12) would have been approximately 10 cm higher than that of BIOL-L14 and BIOL-L3 (Fig. 6; Supplementary Fig. S14). As such, BIOL-L13 would have been both fully and partially exposed more frequently and for a longer maximum daily exposure duration than BIOL-L14 or BIOL-L3. In 2021, BIOL-L13 would have been partially exposed for approximately 3.5 h down to the elevation of the pre-2023 diedown HLS according to the Raffles Lighthouse tide-gauge data (Fig. 6f).
Microatolls on the Siloso Point reef are inferred to have died down in early-mid 2020, but no diedowns were observed between July 2020 and the start of 2021. Bleaching was observed on the Porites microatolls at Siloso Point in May 2021, and we noticed partial mortality of some Siloso Point microatolls in mid-2021 (Supplementary Text 4; Supplementary Fig. S15).
Of the corals surveyed in 2020 and 2021, SILO-L1 and SILO-L2 had the highest HLG prior to the 2021 diedown. The top of SILO-L1's outer ring died between 29th May and 13th June 2021 (Supplementary Figure S15). A similar diedown of centimetre-scale magnitude was observed on SILO-L2 when it was photographed on 27th June 2021. SILO-L4, a microatoll with lower-elevation pre-diedown HLG, also had a centimetre-magnitude diedown between 29th May and 28th June 2021. Assuming coeval partial mortality on SILO-L1, -L2 and -L4, these field observations therefore constrain the diedown timing to the period between 29th May and 13th June. However, SILO-L5, which was first photographed on 27th June 2021, showed no indication of a diedown within the preceding months (Supplementary Figure S15). None of SILO-L1, -L2, -L4 or -L5 had signs of new partial mortality between June and October 2021 (Supplementary Figure S16).
We surmise that the upper parts of SILO-L1, -L2 and -L4 were partially exposed for a sufficient duration to cause mortality of the upper parts of the coral colonies, but the small magnitude of the 2021 diedown on SILO-L1, -L2 and -L4 suggests the threshold exposure duration was exceeded only at the tips of their living rims (within approximately a centimetre of their HLG in May 2021). The pre-2021 diedown HLG of SILO-L1 and SILO-L2 was surveyed on 1 st May 2021 and 24th July 2020, respectively (Fig. 7). The pre-diedown HLG of SILO-L4 was surveyed on 1 st April 2021 (Fig. 7). We extrapolated the surveyed HLG values until the end of May 2021 by assuming a vertical growth rate between 0.5 and 3 cm yr (Supplementary Table S4). From the extrapolated HLG and the Siloso tide-gauge record, we infer that SILO-L1 and SILO-L2 were exposed for two continuous hours per day on six occasions in March and April 2021 (Fig. 7a), though no diedowns were observed at these times. The upper tips of SILO-L1's and SILO-L2's outer living rims were first exposed for approximately three consecutive hours during the spring low tides of 28th to 30th May 2021 (Fig. 7b). Similarly, SILO-L4 was first exposed for more than 2 h on 28th May (Fig. 7a). Therefore, the vertical growth of these corals appears to have been limited by the relatively long exposure duration at the end of May 2021. Since SILO-L1 and SILO-L2 had the highest pre-diedown HLG of any surveyed corals on the Siloso Point reef, we suggest 3 h of exposure per day may be the upper limit of Porites microatoll survival at this location.
Prior to the 2021 diedown, the highest HLG of SILO-L5 was similar to that on SILO-L4, both of which were lower than the highest HLG on SILO-L1 and SILO-L2. No part of SILO-L4 or SILO-L5's living outer ring was subaerially exposed for three or more hours in 2021 (Fig. 7b), but at least part of SILO-L5's outer ring was exposed for two or more hours on ten occasions between May and July. We infer that SILO-L5 was not exposed for long enough for a diedown to occur, even though the maximum HLG of SILO-L5 and SILO-L4 was similar at the end of May 2021.