# Also in the Article

Statistical analyses

Procedure

The effects of colony size on the probability of bleaching and the probability of whole colony mortality were estimated using binomial generalized linear models. For colonies that were not genetically identified, the effects of colony size on the probability of bleaching and the probability of mortality were compared across sites. For colonies that were genetically identified, sites were pooled for analyses because there were not enough genetically identified samples at each site to directly estimate the interactive effects of site and haplotype identity on the probability of colony mortality (see Appendix S1: Fig. S4 for analyses at Site 2). Each colony was assigned a category of 0 for alive, or 1 for dead. We modelled the probability of bleaching using two approaches. The first approach follows that used in multinomial models. For each bleaching severity category (i = 1, 2, and 3), we modelled the probability of bleaching as the log of the odds of bleaching in category i vs. not bleaching (i = 0) as: $lnPbleachingi/Pbleacing0$. In the second approach, we simply distinguished between those corals that did (i = 1) or did not (i = 0) bleach. Since colony size was only estimated on a subset of colonies for which bleaching and mortality were recorded, the effects of site on bleaching and mortality included a larger sample of colonies than the effects of colony size on bleaching and mortality. The effects of colony size were tested using χ2 log‐likelihood ratio tests of models with and without the slope coefficient. Spatial and temporal variation in the relative abundance of haplotypes was assessed using Principal Coordinates Analysis.

Note: The content above has been extracted from a research article, so it may not display correctly.

Q&A