Fig. 3: Adaptive genotype predictions explain spatial patterns of Acropora decline by heat exposure.

A Map displaying the number of survey sites with repeated measurements of Acropora cover across marine provinces of the Indo-Pacific (symbols)⁴⁹. Shown is also the period covered by surveys in every marine province. B The distributions of marginal coefficients of determination (marginal R²) for models explaining the change in Acropora cover using different explanatory variables: marine province of the survey site (Spatial Autocorrelation), maximal heat stress measured between surveys (Max-DHW btw. surveys), maximal historical heat stress (Max-DHW since 1985), and expected frequency of heat-adaptive genotypes––derived from the environmental genomic model. For every model, the marginal R² distribution (N = 163) is shown in a boxplot defined by a central line corresponding to the median value, a box delimited by the lower and upper quartiles, and box whiskers extending to ±1.5 times the interquartile range. Gray lines represent marginal R² estimations from distinct jackknife subsamples. C The model with the highest marginal R² from (B): the relative change in Acropora cover is represented as a function of maximal heat stress between surveys for sites with different expected frequencies of heat-adaptive genotypes (red: 0–0.42%, orange: 0.42–0.52%, yellow: 0.52–0.63; green: 0.63–1). Plot symbols correspond to the marine provinces of the survey site [as displayed in (A)]. The shaded areas show the 95% confidence intervals of the regression lines. Background map was produced using the Global Relief Model from NOAA (ETOPO 2022⁸²) and the Global Distribution Map of Coral Reefs from the UNEP World Conservation Monitoring Center⁸³.