Meeting Abstract
All of us routinely experience performance trade-offs as we complete various tasks. For example, the faster we write, the more likely we are to make mistakes! Therefore, if we look across all 2017 SICB abstracts we could expect a negative correlation between the time spent writing an abstract and the number of mistakes it contains. However, we can also predict a positive correlation because SICB members greatly vary in their writing skills (i.e., some of us are “Darwinian demons” and are able to quickly write excellent error-free abstracts). This situation is similar to the classic “big house big car” model of life-history evolution where individuals differ in their ability to acquire resources, but nevertheless have to allocate their resources into current vs. reproduction. In this case, phenotypic correlations among life-history traits can be either positive, negative, or nil depending on whether there is respectively more, less, or equal variance in acquisition and allocation among vs. within individuals. This is one of the main reason why performance and life-history trade-offs can be hard to detect at the phenotypic level. In this talk, we show how multivariate mixed models (MMMs) can help understand performance (or any other type of) trade-offs. Indeed, MMMs allow straightforward and simultaneous examination of trait correlations at several levels of variation (e.g., within and among individuals, populations, or species, etc.). It is now relatively easy to run MMMs using softwares like R, SAS, WOMBAT, ASReml, and others, yet MMMs are underused in performance studies. We will use a few published datasets in which several individuals were repeatedly sampled for multiple performance traits to illustrate what insights can be gained by applying MMMs to detect performance trade-offs (e.g., speed vs. accuracy).
