Confounding Environmental Colour and Distribution Shape Leads to Underestimation of Population Extinction Risk

Coloured stochastic processes show slow (red), fast (blue) or purely random (white) variation, important across biology, engineering and physics. Changing colour from white to red or blue in traditional models generates coloured stochastic series that are not normally distributed; confounding comparison with normally distributed white series. We illustrate this with a stochastic population model previously used to estimate extinction risk in coloured environments, demonstrating that previous extinction estimates were based on methodological artefacts rather than the actual effect of coloured environmental variation. We propose a method for generating normally distributed coloured series, which must be used to avoid spurious inference