a) univariate early warning signals
Early warning signals developed for univariate data are the simplest form of CSD assessment and thus have received the most research effort. Table 2 describes the most common EWSs, all of which are provided inEWSmethods via the uniEWS function, and how they are calculated. Each of these are also provided in the earlywarningspackage and mathematically described in detail (Dakos et al. 2012a). The development that EWSmethods provides over that package is the diversity of approaches used to compute these EWSs beyond those available in earlywarnings , allowing users to tailor their analyses to support their use case. This primarily involves the choice of rolling versus expanding windows during calculation (Figure 2).