Crowd?sourced plant occurrence data provide a reliable description of macroecological gradients

Deep learning algorithms classify plant species with high accuracy, and smartphone applications leverage this technology to enable users identify in the field. The question we address here is whether such crowd-sourced data contain substantial macroecological information. In particular, aim understand if can detect known environmental gradients shaping co-occurrences. study analysed 1 million points collected through use of mobile app Flora Incognita between 2018 2019 Germany compared them Florkart, containing occurrence by more than 5000 floristic experts over a 70-year period. direct comparison two sets reveals that particularly undersample areas low population density. However, using nonlinear dimensionality reduction were able uncover patterns both correspond well each other. Mean annual temperature, temperature seasonality wind dynamics as soil water content texture represent most important composition collections. Our analysis describes one way how automated identification could soon near real-time monitoring their changes, but also discusses biases must be carefully considered before biodiversity effectively guide conservation measures.