Conclusion
The ability to use accessible and easy to interpret tools are key for ecological monitoring. In this note we present EWSmethods , an R package consolidating the simplest methods of early warning signal assessments into a coherent suite of metrics and visualisations. Each function is consistent in its parameterisations, terminology, and output to allow any user to interpret the assessment confidently, regardless of the data dimensionality or EWS approach.
It would however be remiss to overlook the pivotal earlywarningspackage and work of Dakos et al. (2012). EWSmethodsinnovates on earlywarnings by providing alternative calculations (rolling vs expanding windows) and data types (univariate vs multivariate), but does not provide the additional modelling techniquesearlywarnings supports (diffusion-drift-jump models, BDS tests etc). We direct readers to that package on github (as it is no longer maintained on CRAN at the time of writing - https://github.com/earlywarningtoolbox/earlywarnings-R) for the typical rolling window EWS approach due to the additional modelling capabilities it provides. EWSmethods better supports multivariate analyses and standardises across univariate EWSs, multivariate EWSs and machine learning models to allow comparability. It also provides access to purpose-built machine learning models not otherwise available to R users. Consequently, users are able to explore an ensemble of generic forecasting methods to identify oncoming transitions and tipping points in their system.
Generic approaches also facilitate wider research interest into the universal challenge of identifying oncoming tipping points. Resilience-based approaches are critical for the management of globally imperilled systems (Folke et al. 2010, Oliver et al. 2015, Capdevila et al. 2022) but are applicable in other disciplines. Remotely sensed data could allow global level tipping point assessments for example (Forzieri et al. 2022), individual mortality risk may be detectable (Cailleret et al. 2019) or positive thresholds can be encouraged (Lenton et al. 2022). The low barrier to entry that EWSmethods provides for R users can aid the development of these developing research avenues.
To cite EWSmethods or acknowledge its use, cite this Software note as follows, substituting the version of the application that you used for ‘version 1.0’: