Fig. 4

Symbiont community structure prior to, during, and at the end of the temperature manipulations. a Lineage-specific symbiont dominance throughout the experimentation period. Relative proportions of Durusdinium symbionts at designated time points of the temperature manipulation are normalized to a range of –1 (complete dominance by Cladocopium) to + 1 (complete dominance by Durusdinium) to infer the degree of symbiont shuffling. Values are means ± standard error (n = 6). Standard error was 0 for PdC at 26°C both on Day 0 and Day 7. b Changes in relative abundance of different genera of symbionts, expressed as log₁₀[D/C] by calculating Durusdinium to Cladocopium ratio. Line and shaded area represent logarithmic regression and 95% confidence interval, respectively. c Non-metric multidimensional scaling (NMDS) ordination of symbiont communities constructed based on a Bray–Curtis dissimilarity matrix, given the relative abundance of dominant symbiont types detected during the aquarium maintenance (open dots/triangles) and at the end of temperature manipulation (solid dots/triangles). Grouping based on complete linkage at 90% similarity is indicated by circles for each treatment (dashed, pre-treatment 26°C; solid blue, control 26°C; solid red, heat 32°C). D Durusdinium, C Cladocopium