Figure legends
Fig 1 . Locations of reefs where field surveys were undertaken for benthic data collection. The top panel shows the location of surveyed reefs from Honduras, the bottom pannel shows the surveyed reefs of Indonesia, located in whte Wakatobi National Park.
Fig 2 . Average sea surface temperature anomaly (SSTA) in °C and standard errors (whiskers) from 2012-2019 at the Indonesia and Honduras sites surveyed in this study. Temperatures were quantified from the 52 weeks preceding the survey period which began at the 1st of June for each site, each year. Points represent the mean SSTA, while error bars are standard error.
Fig 3. MCMCglmm coefficient estimates for the response of selected major benthic components under elevated sea surface temperatures, quantified as sea surface temperature anomaly (SSTA). Coloured points correspond to the specified benthic component, representing the mean model coefficient. Horizontal bars represent 95% credible intervals, which are considered significant when they do not cross zero (grey line). The models ran separately for Indonesia and Honduras sites. The components were selected based on preliminary analysis as the most dominant components of reefs from the field surveys undertaken by Operation Wallacea volunteers from the years of 2012-2019.
Fig 4 . nMDS analysis of the benthic community composition at the Indonesia sites (A-C) and Honduras sites (D-F). A. and D. are the composition of individual benthic components from 2012-2015. B. and E. are response of individual benthic components from 2016-2019 (i.e. showing the response of the global marine heatwaves which took place in 2016/2017 (Fig. 2)). Letters represent the individual taxa which are specified in Table 1. C. and F. represent the entire ordination space of the benthic composition at each individual reef, where the red polygon bounds the sites from 2012-2015, while the blue polygon bound sites from 2016-2019. The black arrows connect the before and after pairs of each site.
Fig 5 . Change in benthic composition in response to sea surface temperature anomalies (°C). Regression curves are fitted using generalised additive models with 95% confidence intervals displayed in the grey bands. Each data point represents the shift in composition at a reef from before 2016 to after 2016, based on the Euclidean distance of sampled reefs in the nMDS analysis. Purple points show Honduras sites, while the yellow shows Indonesia sites.