Figure Legends

Figure 1. (a) Map of study regions and presence records used in this study. Red and blue points represent sample collection locations for molecular analyses in the Western Tropical Indo-Pacific (WTIP) and the Central Tropical Indo-Pacific (CTIP). Gray crosses show presence records used for the species distribution models. (b) Network of microsatellite genetic differentiation (Cavalli-Sforza and Edwards chord distances). The topology results from pruning the network for pairwise genetic distances <0.534. The smallest chord distance (0.499) between the WTIP and CTIP lineages is shown. (c) Genetic lineage division over space based on STRUCTURE clustering (k = 2). Population abbreviations are the same as in Hernawan et al. (2017) and Jahnke et al. (2019a), and their classification to network analysis is in agreement with that of STRUCTURE (Fig. 1c).Figure 2. The realized niches for the two lineages of Thalassia hemprichii quantified via eight-dimensional hypervolumes. Circles with white rims indicate hypervolume centroids. Boundaries and shapes of hypervolumes were delineated by 10,000 points randomly sampled within each hypervolume. The photograph depicts T. hemprichii.Figure 3. Present-day continuous (a, b) and binary (c, d) habitat suitability predictions for Thalassia hemprichii by lineage-level (a, c) and species-level (b, d) Maxent models. Dashed lines represent the equator.
Figure 4 . Changes in continuous (a, b) and binary (c, d) habitat suitability forThalassia hemprichii projected by lineage-level (a, c) and species-level (b, d) Maxent models under the RCP 8.5 scenario in the 2050s. Dashed lines indicate the equator. The category “stable” represents areas predicted to be suitable under both present-day and future climatic conditions, “loss” indicates areas predicted to be suitable under present-day conditions but unsuitable in the future, and “gain” indicates areas predicted to be unsuitable under present-day conditions but suitable in the future.