Early warning signals
To calculate univariate EWS for any one time series from this community,
we would use uniEWS . We first need to select the EWS indicators
of interest to provide to the metrics argument. Autocorrelation
(“ar1”) and variance (represented by the standard deviation - “SD”)
are the most commonly used EWSs and have the largest body of research
defining their best utility (Carpenter and Brock 2006, Dakos et al.
2012b, Patterson et al. 2021). Using these metrics, we then choose the
time calculation approach (expanding), the resulting burn in period (50
data points) and the sigma threshold (two), and that we want uniEWS to return a visualisation (ggplotIt = TRUE). uniEWS only performs assessments on univariate data but
requires a two column data frame where the first column is an equally
spaced time vector and the second is the time series to be assessed. We
have chosen the third species here.