Using mixed effects models in R for ecological and evolutionary insight

Using mixed effects models in R for ecological and evolutionary insight

John Morrongiello

Ecological data sets are often complex and hierarchical. For example, repeated measures can be taken from individuals or plots that are nested within sample times or sites. Whilst this complexity can be a pain (how do we deal with it?), we can use it to our advantage. Mixed effects models offer an opportunity to harness the data’s underlying complexity and explore a range of ecological and evolutionary questions not always possible with other more traditional techniques. Here, I’ll present a general overview (from a biologist’s perspective) of what a mixed model is then illustrate their different uses with a series of worked case studies. These will include: estimating long-term patterns in fish growth and thermal reaction norms from otolith data using linear mixed models (LMM), recreating recruitment time series in a data poor fishery using zero-inflated mixed models (ZINB), and exploring migration patterns from telemetry data using generalised additive mixed models (GAMM).