Harnessing the benefits of herbarium specimen digitisation for inferring recent and ongoing plant extinctions.

Evidence for the ongoing biodiversity crisis rests on assessment of a small fraction of described species, with major knowledge gaps for most organisms, including plants. Here, we highlight how digitised herbarium specimens can be used to accelerate and improve estimates of recent and ongoing plant extinctions. We focus on species already considered extinct because they represent a special category for understanding biodiversity loss and a special scientific challenge, as their 'detection' relies on proving absence. We propose that this challenge is embodied by a neglected biodiversity shortfall, the Katuš shortfall, encompassing all facets of unquantified levels of past, present and future biodiversity loss. To address this shortfall, we review how methods for estimating the probability of species extinction can be scaled up to harness the massive amounts of digitised data being produced across the 'global metaherbarium' using artificial intelligence. Thus, we suggest that the Katuš shortfall can be diminished by shifting focus from proving absence to a probabilistic framework. This can contribute to increasing the accuracy of lists of extinct plants and reveal the true extent of the biodiversity crisis.