Results of model selection assessing all combinations of fixed factors using AICc values of the GLMMs.

(a) sapstain within Pinus radiata logs as a response variable in GLMMs with a Poisson lognormal error structure to account for overdispersion in the binomial stain model, (b) evidence of bark beetle (Hylastes ater and Hylurgus ligniperda) colonisation as a response in GLMMs with binomial error structure, (c) evidence of other arthropod colonisation as a response in GLMMs with binomial error structure, and (d) GLMMs with Poisson lognormal error structure testing whether the fixed covariate effects of beetle evidence and other-arthropod evidence explain the distribution of stain within and among logs. All models included site replicate and log replicate as random factors. Models within two ΔAICc units of the best model (ΔAICc = 0) were considered to have equivalent explanatory power, as indicated in bold. All models also included the random effects of site replicate and log replicate.