Figures
Figure 1. A schematic overview of the common garden experiment. The Japanese populations were collected on September the 14th (Fuitatsuiwa; fut ) and 15th (Akkeshi; akk ) and arrived in the climate room on September the 17th. The populations from Germany were collected and brought to the climate room on September the 20th (Nordstrand; nor ) and September the 21st (Kiel; dhd ). The antibiotic treatment started September the 26th and lasted 24 hours. Algae were subsequently separated in two parts into different tanks at 15 °C and 22 °C. These tanks also received the inoculum: a combination of living algal branches from all population in a 50 mL tube with openings, to function as a source of microbes. Epibiota were sampled in the field (tf ), before the disturbance treatment (t0) and 1,2 4 and 12 weeks following the disturbance (t1 , t2 , t4 , t12 , respectively).
Figure 2. Marginal means of host performance traits recorded at the end of the experiment (t12 ) with 95% confidence intervals. The relative growth rate (RGR, A), measured as percentage wet weight increase per day between temperatures and between ranges (B). The probability of observing brittleness of the algal thallus between temperatures (C) and between ranges (D). RGR graphs include the partial residuals and for thallus brittleness the raw binary data were plotted .
Figure 3. Rarefied OTU richness (to 1132 reads, panels A-E), evenness in terms of logit transformed PIE (F-J) and within-population beta diversity measured as Bray-Curtis (K-O) and Euclidean distances (P-T) with 95% confidence limits and partial residuals. Marginal means in the field (tfield ), before holobiont disturbance (t0 ) and 7 days after holobiont disturbance at 15 °C (t1 , A, F, K, P). Post-disturbance responses over time at 15 °C (B, G, L, Q) and 22 °C (C, H, M, R), by temperature (D, I, N, S) and by range within temperature (E, J, O, T).
Figure 4. Non-metric dimensional scaling (nMDS) based on the rescaled residuals from mGLMs including the LSD as offset to correct for the effect of sequencing depth using Bray-Curtis (A-D) and Euclidean distances (E-H). . Panels A and B include samples from pre- and post-disturbance time points (tf , t0 , t1 , t2 , t4 and t12 ).. Blue and red arrows draw the post-disturbance trajectories of the 15 and 22 °C temperature groups in time. The other panels show nMDS plots using rescaled residuals from mGLMs including all terms, except the variable of interest; time (B, F), temperature (C, G) and range (D, H). The 95% confidence regions of the group centroids are shown as ellipses. Note that the smaller panels (B-D, F-H) are zoomed in and some datapoints are outside the frame limits.